If MSI testing becomes universal, says Dr. Hamilton, “then it needs to be clear that the results of microsatellite instability status are looked at and correctly interpreted by clinicians.” Even a simple positive or negative result can be problematic, “believe it or not.”
April 2017—Molecular testing for colorectal cancer is not for the faint of heart.
While that’s not news to Stan Hamilton, MD—he’s head, Division of Pathology and Laboratory Medicine, and the Frederick F. Becker distinguished chair in cancer research, University of Texas MD Anderson Cancer Center—he was reminded of this fact recently when a friend looked at the multipage molecular pathology report on his own tumor. “He called and basically said, ‘What are you guys doing?’” recalls Dr. Hamilton, noting that his friend, an engineer, is well versed in reading technical reports. “And he was completely befuddled by what he saw.”
Dr. Stan Hamilton (right) at the University of Texas MD Anderson Cancer Center with George Chang, MD, MS, a professor in the Department of Surgical Oncology and chief of the section of colon and rectal surgery. The guideline on molecular biomarkers for the evaluation of colorectal cancer “has brought the quality control aspect [of testing] front and center,” says Dr. Hamilton.
Dr. Hamilton adds, “He asked, ‘Why can’t you make this easier on us patients?’”
And that’s just the report. Left unsaid, but equally valid, is another question: Why can’t molecular testing be easier for clinicians and pathologists, too?
With colorectal cancer, ease is a distant, likely nonexistent, goal. Molecular testing options are complicated to navigate. Even when a molecular portrait of the tumor has emerged, “We don’t have a lot of active agents to choose from,” says Carmen Allegra, MD, chief of oncology and hematology, University of Florida, Gainesville.
Yet the literature continues to grow, and with it come new approaches to using older tests, as well as added data about promising biomarkers. Trying to make sense of it all is a fresh guideline on molecular biomarkers for evaluating CRC, representing the best and the brightest from the CAP, American Society for Clinical Pathology, Association for Molecular Pathology, and American Society of Clinical Oncology (Sepulveda AR, et al. Arch Pathol Lab Med. Epub ahead of print Feb. 6, 2017. doi: 10.5858/arpa.2016-0554-CP).
Unlike previous guidelines from ASCO and the National Comprehensive Cancer Network, says lead author and AMP co-chair Antonia Sepulveda, MD, PhD, that input from four key societies meant unusually heavy emphasis on laboratory testing as well as the more typical guideline coverage of required testing for targeted and conventional therapies. “This has never really been done with such a global scope,” says Dr. Sepulveda, professor and vice chair for translational research, and director, Division of Gastrointestinal Pathology, Department of Pathology and Cell Biology, Columbia University, New York City.
“If you look at the recommendations,” adds Dr. Hamilton, who was the CAP co-chair, “the vast majority of them deal with how to get the testing done. This is not an inconsequential issue. But this is not simply a lab testing guideline. It includes the clinical utility of the tests, and with expert opinion from the medical oncologists who are ordering the tests.”
Another strength, says Dr. Sepulveda, is that the guideline is based on a systematic literature review (with extensive tables to prove it) and levels of evidence, using National Academy of Medicine (formerly Institute of Medicine) standards for developing clinical practice guidelines. This should chase most bias from the recommendations, says Dr. Hamilton. Moreover, even with its 21 statements—“It’s a large and comprehensive guideline,” Dr. Sepulveda says—updates are likely. By following the academy’s standards, the new guideline should remain relevant for some time, she says.
Reflecting the complicated nature of CRC testing, “There was an enormous amount of literature—thousands of papers that we combed through,” says Dr. Allegra, the ASCO co-chair. The major tests for CRC are not new, but their application is evolving—a fact reflected in the guidelines. And in the meantime, researchers continue to find new puzzles in need of solving.
As with the city of Pittsburgh, there are three main rivers to follow in CRC testing: DNA mismatch repair, or MMR, status; BRAF mutation; and RAS mutation.
Testing for MMR, either by immunohistochemistry for the four MMR proteins (MLH1, MSH2, MSH6, and PMS2) or by microsatellite instability DNA-based testing, has evolved from its use in a small group of patients with an inherited disease to a wide variety of clinical decisions, including whether to give postoperative adjuvant therapy and how to treat patients with advanced disease. Oncologists might order that same test for different reasons, says Dr. Hamilton. “That, in fact, is part of the reason we’ve now recommended universal testing in the guideline.”
The first use to emerge was as a marker for Lynch syndrome. MMR is also used as a prognostic marker.
Most recently, researchers have recognized the value of the MSI-high (i.e. high level of microsatellite instability) in patients with advanced disease (about five to six percent of CRC patients) in predicting response to immunotherapy with immune checkpoint inhibiting drugs, specifically pembrolizumab (Le DT, et al. N Engl J Med. 2015;372:2509–2520).
In patients with colorectal cancer, about 20 percent have defects or mutations in one of the DNA repair genes. In about a quarter of those patients, the mutation is based in their germline, the underlying mechanism of Lynch syndrome. As Dr. Hamilton notes, knowing MMR status is crucial not only for managing these patients but also to encourage earlier screening of at-risk family members. For the other 75 percent of patients with mismatch repair deficiency, the mutation is sporadic.
In both groups, MMR mutations carry prognostic information. Those with a deficiency typically, and counterintuitively, have a better outcome, regardless of the stage of their disease. In the case of a stage II patient with an MMR outcome, for example, “They have an extremely good outcome, to the point where we generally don’t consider those patients for any kind of adjuvant therapy after their primary surgery,” says Dr. Allegra.
With other tests, figuring out the next clinical steps often resembles the search for winning lotto numbers.
BRAF mutation (occurring in about eight percent of patients with advanced colorectal cancer) engendered lively discussion when the guideline creators looked at its role as an adverse predictive marker.
“In the report,” Dr. Allegra says, “we said that a BRAF mutation means you don’t do as well with an EGFR inhibitor. But the benefit isn’t zero—that’s what the data showed.” While many clinicians balked at the idea of using EGFR inhibitors in people with BRAF mutations—“They thought it was crazy,” Dr. Allegra says—a hard look at the data suggests some benefit. “So it’s hard to say you shouldn’t use it at all.”
As Dr. Hamilton explains, it’s well known that this gene mutation behaves differently in different subsets of patients. For patients with high levels of microsatellite instability and BRAF mutation, “the BRAF mutation doesn’t seem to matter as much. The outcomes are somewhat worse, but not substantially so.”
In contrast, patients with microsatellite stable or MSI-low tumors and BRAF mutation have a far worse outcome and generally present with more advanced disease, Dr. Hamilton observes. They also seem to be more resistant to chemotherapy. During the guideline discussions, “There was some concern about whether this was related to the fact that the BRAF mutation just conferred a worse prognosis, or whether it really was a predictive marker and could be used to make decisions about therapy.”
Another possible use for BRAF popped up in an abstract presented at ASCO’s 2017 Gastrointestinal Cancers Symposium. Research by Dr. Hamilton’s colleague at MD Anderson, Scott Kopetz, MD, PhD, suggested simultaneous EGFR and BRAF inhibition prolonged progression-free survival in patients with advanced disease. Patients were given a combination of cetuximab and irinotecan, with and without vemurafenib, a BRAF inhibitor. Patients who received the latter drug as part of their regimen “did much, much better than those with the classical chemotherapy,” says Dr. Allegra (median PFS of 4.4 versus 2.2 months; disease control rate of 67 percent versus 22 percent). He considers this a major advance and says, “It was probably the most important paper presented at GI ASCO in January.”
Dr. Hamilton, who has collaborated on a number of papers with Dr. Kopetz, adds, “Scott is an absolutely superb researcher. I think the world of him. And I think he’s right.” But, he adds, it is an abstract.
While that abstract was too newly hatched to influence the guideline, the authors had their hands full trying to assess other possible breakthroughs.
“It took us two years to get this done, because every time we turned around and thought we were getting toward the end, something new, like extended RAS, popped up,” says Dr. Hamilton.
For patients with advanced disease, anti-EGFR drugs such as cetuximab and panitumumab are part of the oncologist’s armamentarium. Less well recognized is that patients who have a KRAS mutation generally will not respond well to those drugs. “There have even been a few studies that have suggested they actually get worse,” says Dr. Hamilton.
In looking at targeting the EGFR pathway, the guideline authors looked for literature for or against BRAF testing. “So far the evidence is insufficient to make a recommendation,” says Dr. Sepulveda. Ditto for markers such as PIK3CA and PTEN. “So while we can test these genes for other reasons, they are not useful at this time for making decisions about anti-EGFR therapy,” says Dr. Sepulveda.
The KRAS discussions thus were relatively tranquil. But while the guideline was being developed, the NRAS story began to be told.
An NRAS mutation should sound an alarm. For these patients, anti-EGFR agents don’t work. “They have zero activity,” says Dr. Allegra. “And they’re toxic. Sparing patients from that therapy is important.”
“This has evolved quickly in the last 18 months to two years,” he continues. Those who had been looking at RAS had mostly confined their search to a couple of hot spots. “But what’s become apparent is that if you have a mutation, probably regardless of where that mutation occurs, it carries with it a negative predictive value.”
About half of patients with colorectal cancer have a KRAS mutation. Another five to eight percent has an NRAS mutation.
“A large meta-analysis, summarizing a number of primary trials, provided additional data that led to recommendation No. 1 in our guideline, for expanded [also called extended] RAS testing,” Dr. Sepulveda says (Sorich MJ, et al. Ann Oncol.2015;26:13–21). Patients being considered for such treatment must receive RAS mutational testing, including analysis of KRAS and NRAS codons 12 and 13 of exon 2; 59 and 61 of exon 3; and 117 and 146 of exon 4.
The authors also had to pause in their discussions about mismatch repair testing.
IHC is relatively easy to do and relatively inexpensive. Turnaround times are rapid, and it can be done with small amounts of tissue. “That’s obviously a terrific screening test,” Dr. Hamilton says.
But it’s now recognized that there are situations where IHC doesn’t work.
One occurs in patients with Lynch syndrome who have a gene mutation that abrogates the function of the gene but doesn’t cause loss of immunoreactivity. In other words, the expression exists in the tumor, but the protein is nonfunctional. “So the patient still has high levels of microsatellite instability and is still generating the phenotype of an MSI-high cancer, but the immunohistochemistry is often not able to pick that up,” says Dr. Hamilton. “There are sometimes subtle differences in the tumors, in the pattern of immunohistochemistry, that can give you a clue that something’s up,” including patchy expression or a peculiarity in the nucleus, where the staining may be less than in a typical case. “It just doesn’t look quite right.”
A more recently identified problem is that in some tumors affected by MSH6 mutations, the IHC will yield an abnormal result—no protein expression—but those cases do not show high levels of MSI. “It’s hard to understand why that occurs, but we see it,” says Dr. Hamilton.
“The other thing now recognized is the biallelic inactivation by somatic mutation,” Dr. Hamilton continues, “where there are point mutations in both copies. One of the mismatch repair genes inactivates them, but again, doesn’t affect the protein.” In this subset of cases, it’s not the methylation mechanism that extinguishes expression; rather, the protein is nonfunctional—the immunoassay sequence is changed by the mutations.
IHC and molecular MSI testing have their pros and cons, clearly. The best choice should reflect patient population and the needs of the ordering physician. “If you’re looking for Lynch syndrome on the basis of a clear-cut family history, it’s probably quite reasonable to start with a molecular test,” says Dr. Hamilton. If that comes back abnormal, “move to one of the family cancer germline mutation panels, and don’t worry about [IHC] in the tumor itself.”
If a patient is older, and methylation of MLH1 is likely, “starting off with the molecular sequencing spends a lot of money and takes up time,” but it may be worth it, Dr. Hamilton says. “The bottom line is, if you want to make certain of what you’re dealing with, you’ve got to do a fair amount of work.”
IHC will tell physicians which genes are involved. If BRAF is mutated, then Lynch syndrome is highly unlikely. “But that’s not perfect,” Dr. Hamilton says. It’s also important to test MLH1 to determine if methylation is present. “But that’s not perfect, either, because methylation does occur in patients who have Lynch syndrome as the underlying cause.” And don’t forget the aforementioned biallelic somatic gene mutations that can lead to abnormal protein—and the need for gene sequencing. “Finally, we now recognize a hypermutable—some refer to it as ultramutated—group with high mutational burden due to abnormalities in the polymerase epsilon and delta genes. Those do not have MSI at all.”
“Fortunately,” says Dr. Hamilton, “we’re beginning to understand much better what we’re dealing with. But unfortunately, it turns out to be very complicated to do the whole workup to get the answer in these individual patients when you get these unusual results.”
Dr. Hamilton pauses, then plunges ahead with a rueful laugh. “I’m still not done. To make things even worse,” he says, studies of germline testing have shown that looking at the phenotype in families has a significant error rate in identifying which genes are abnormal. “Cases that look like Lynch syndrome turn out to be something like PTEN deficiency.”
The guideline can help with even these most complicated cases. Genes are fickle—a fact not always reflected in other guidelines, says Dr. Hamilton. “Frankly, most of the other guidelines looking at results of assays, particularly panel sequencing, are not looking at how you got those results,” Dr. Hamilton says. “This guideline has brought the quality control aspect front and center. I’m not casting aspersions. It’s simply a matter of how they think about that.”
In colorectal cancer, there are few straight lines through the testing process. Even KRAS testing—ostensibly one of the least complicated steps—has its hiccups. Given the exceptions that may turn out to be unexceptional, and the growth of targeted therapies, stepping precisely through CRC testing, with close attention to the quality control directives in the guidelines, becomes even more crucial, says Dr. Hamilton.
Dr. Sepulveda drives home that point when she highlights the predominance of statements devoted to lab-specific issues, such as types of tissue and fixative to use and the benchmarks for choosing them, turnaround times, analytical sensitivity, reporting clarity, quality improvement measures, and the like. “There’s a lot there,” she says. “It’s all important.”
“We tried to cover the waterfront,” Dr. Hamilton adds.
The guideline could add another level of clarification to the MMR discussion as well. Calling for its universal testing should nudge third-party payers. “That’s one of the goals of a national guideline,” says Dr. Allegra. “If you say something ought to be tested, it’s harder for third parties to say no.”
Despite the high stakes, molecular testing results can sometimes fall through the cracks, even when delivered promptly. How can pathologists make sure the results are seen by the right pair of eyes, understood, then acted on?
In the current National Cancer Institute-MATCH trial, for example (a precision medicine trial in which specimens are evaluated for a series of markers to qualify patients for one of 24—soon to be 30—arms), two of the MSI genes are used for determining access to one of the arms. Pathologists are clear on the drill: If a result is positive, that means the gene retains expression; a negative result means loss of expression. And a loss of the protein (i.e. a negative result) is a positive for receiving therapy.
“You would have thought that would have been fairly clear to everyone, but almost on a weekly basis we got a contact from a site: ‘Does that positive [result] mean my patient’s eligible for immunotherapy?’” says Dr. Hamilton.
Counterintuitive reasoning is one problem. Electronic medical records, oddly, are another.
Dr. Hamilton explains: Pathology reports have actually become quite clear. “The CAP has been working on that for years, and the biomarker guidelines are out there [he served on the biomarker reporting committee], and most pathology departments have taken them to heart,” he says. The guidelines, when followed, ensure results are clearly presented—typically in tabular form and thus highlighted in the text.
So far, so good. But when the results become part of the EMR, data from the laboratory information system get converted. Depending on the EMR being used, “You lose that nice visibility when the clinician looks at it,” Dr. Hamilton says.
Dr. Hamilton didn’t recognize the problem until he became involved in the NCI-MATCH trial and saw what happened with certain EMRs. “It can be really tough to find the results in these reports.” The problem extends to reports from the major reference labs. “Their reports are very well laid out,” Dr. Hamilton says. “It’s very easy to get the information at a glance.” When the information hits the EMR, however, all bets are off. The difference, in some cases, “is night and day.”
It’s more than frustrating. “Frankly, it’s a patient safety and quality of care issue,” says Dr. Hamilton. Targeted therapies and immunotherapeutic agents associated with these assays yield good results in a significant proportion of patients, he says. “It’s a real issue that needs to be addressed,” especially given the pressures on already-busy clinicians to see more patients.
Dr. Allegra agrees about the busy part. Even if the results are clear in the EMR (which seems to be the case at his institution), he voices frustration at how test answers often come to him piecemeal. “As docs get busier and busier, it may not get acted on.”
The guideline’s authors—experts all—breezily admit to not following them to the letter in their own practices. It’s to be expected, given their patient populations, which often include those with advanced disease who’ve run out of options elsewhere. As Dr. Allegra puts it, “You go to the university, and for those patients, if they’re lucky enough to have a certain kind of genetic defect, the magic does happen.”
Dr. Hamilton is involved in a large phase one program that is testing tumors in an attempt to develop biomarkers in tandem with identifying toxicity. “We’re requesting testing on a very high percentage of patients with advanced disease with a panel of genes related to that approach. We’re looking at pathways that may be important, not only as targets but also as modifiers of the response and resistance to therapy.”
Dr. Kopetz’s work fits in with those efforts to identify resistance to mutation. Work like this will, it’s hoped, help explain why BRAF inhibitors that work spectacularly in melanomas with a BRAF mutation don’t translate to CRC. “Colon didn’t respond at all,” says Dr. Hamilton. “There were resistant pathways that developed in EGFR and C-meth. When the drug blocked the BRAF pathway, these other pathways were activated, and the tumors kept growing.”
While this deeper understanding is starting to become part of general usage, “We’re not there yet,” says Dr. Hamilton. “In particular, the understanding of which pathways and how many of them can be inhibited without introducing horrible toxicities needs to be answered.” The other problem is that combinations of co-mutations are variable, he says. “So we’re trying to figure out in advance what the most common co-mutations are that you might want to try to inhibit.”
At Columbia, Dr. Sepulveda says, every case of primary CRC undergoes IHC testing for the four MMR proteins, as well as MSI testing. “Some of the immunotherapy clinical trials ask for MSI status of the tumor, so we want that information up front.” They also use a commercial next-generation sequencing panel, do extended RAS testing, and routinely test for BRAF and PIK3CA. Though the guideline does not recommend PIK3CA, compelling retrospective data show that it might be a useful biomarker for patients postsurgically, as well as to qualify patients for a clinical trial. “Because we are an academic center, we decided to go ahead and include this on our reports,” she says.
Dr. Hamilton blithely says there was almost no controversy during the guideline discussions—until he’s asked about turnaround times. The memory of that, like childbirth, was something he managed to forget. “I guess that was a Freudian thing on my part,” he says.
Oncologists, naturally, want test results delivered as if borne by the wing-footed Mercury.
Dr. Hamilton is sympathetic. “Think of it from the patient perspective,” he says. Those with advanced disease understand the gravity of their situation. “Sitting and waiting for lab results to come back is awful. We have a duty to patients to do this as rapidly as possible.”
But real limitations weigh on laboratories as well, such as having sufficient staff to pull a case and review it, in the case of a resection, or sending out a biopsy specimen when that’s the only tissue available from that patient. “These are practical, day-to-day issues that confront us all,” says Dr. Hamilton. “If we had unlimited resources and a guarantee that specimens we’re going to send out are going to be sent back, that obviously would make things much easier.”
Hence, the intense back and forth among physicians. “Some of the pathologists took issue with how fast we were recommending tests and reports got returned,” says Dr. Allegra. Initial suggestions were deemed impractical, he says, given that pathologists can’t always control how quickly they receive tissue for testing.
“There were some areas where there was consternation and handwringing,” says Dr. Allegra. In the end, the guideline says things like, “Laboratories must provide clinically appropriate turnaround times,” “molecular biomarker results should be made available as promptly as feasible,” and “It is suggested that a benchmark of 90 percent of reports be available within 10 working days of tissue receipt in the molecular diagnostics laboratory.”
Says Dr. Allegra: “There was some hesitancy to make things very strict—everyone wanted a little flexibility.”
Apart from some spirited TAT discussions, there was good concordance between the pathologists and oncologists. “That’s a tribute to Antonia Sepulveda as she led the group,” Dr. Hamilton says. What few differences did emerge often had to do with practice styles, he says. “It’s not unexpected. There are some of us who are early adopters, and there are other people who look for a much more advanced level of evidence before they’ll start to take a new approach.”
Karen Titus is CAP TODAY contributing editor and co-managing editor.