Guidelines reset horizons of molecular testing in NSCLC

Anne Paxton

February 2017—It doesn’t happen often. But from time to time, says Gene Finley, MD, director of medical oncology at Allegheny Health Network in western Pennsylvania, a patient who is at death’s door will make such a dramatic recovery with therapy that clinicians refer to it as a “Lazarus effect.”

Speaking at a November 2016 webinar produced by CAP TODAY and sponsored by Pfizer Oncology with the collaborative assistance of the Association of Community Cancer Centers, Dr. Finley traced his enthusiasm and advocacy for targeted therapy to one of his adenocarcinoma patients, about eight years ago, who showed such a response. “He presented with severe respiratory failure. He was on high levels of oxygen and in desperate straits. On the x-ray he had a diffuse alveolar and interstitial pattern throughout both lungs. He couldn’t breathe.”

After being treated with chemotherapy, the patient had an acute myocardial infarction, and once he recovered from that, molecular testing showed he had a driver mutation of EGFR with a deletion in exon 19; his subtype of cancer, making up about 15 percent of all adenocarcinomas, was targetable by erlotinib. “At that point, we had the approval of Tarceva. I put him on the drug and within a week he had this Lazarus effect,” Dr. Finley said.

Since those early days of targeted therapy, “As time has gone on, we’ve been able not only to subtype non-small cell lung cancers by histology into squamous and adenocarcinomas, but also to drill down into the molecular underpinnings of the cancer.” And given what is now known, he said, “We really have to redouble our efforts to test all of our patients for these important mutations.”

Dr. Finley was one of four presenters in the web­inar titled “CAP/AMP/IASLC Guidelines for NSCLC: Current Experience and Future Update.” A pathologist, medical oncologist, radiation oncologist, and surgeon joined as panelists in the webinar to describe how new guidelines and technology are helping molecular testing recharge hope for later-stage NSCLC patients.

Three groups—the International Association for the Study of Lung Cancer, the CAP, and the Association for Molecular Pathology—teamed to produce in 2013 the first set of guidelines for molecular testing for lung cancer predictive biomarkers. Those guidelines, still current, were based on the two FDA-approved and clinical-trial-proven tyrosine kinase inhibitor targets, EGFR and ALK. The American Society of Clinical Oncology endorsed the guidelines soon after they were issued.

But “we knew that would not be the end of the story,” said webinar co-presenter Philip T. Cagle, MD, a professor of pathology and genomic medicine at Houston Methodist Hospital and Weill Cornell Medical College. The National Academy of Medicine (formerly the Institute of Medicine) is rather strict, he noted, in requiring that guidelines be updated regularly. Thus, the first guidelines had barely been issued when the three organizations began work on revisions. “This field is moving very rapidly,” said Dr. Cagle, editor-in-chief of Archives of Pathology & Laboratory Medicine.

In the webinar, Dr. Cagle previewed the significant changes clinicians and pathologists can expect in the next edition of the guidelines, scheduled for release later this year. When an initial draft of the revised guidelines was completed in 2016 by the co-chairs and expert panelists—an international team of pulmonary and molecular pathologists and oncologists—they asked for public comment and those comments have now been reviewed.

Two major recommendations in the guidelines will remain largely unchanged: Physicians must use EGFR molecular testing to select lung adenocarcinoma patients for EGFR-targeted therapy irrespective of clinical characteristics or when adenocarcinoma cannot be excluded. The same requirement holds for ALK. A recommendation that will be added in the revised guidelines will state that physicians may use EGFR and ALK testing in tumors with histologies other than adenocarcinoma when clinical features suggest a higher probability of an oncogenic driver. “For example, if the patient whose lung cancer is diagnosed as squamous cell is a never-smoker, who is young, it’s certainly up to the individual treating physicians and their pathologists about doing the testing then,” Dr. Cagle said.

Whenever possible, the new guidelines will state, next-generation sequencing should be used for molecular testing. That recommendation arose because of the widening commercial availability of NGS since the first guidelines were developed. Dr. Cagle was quick to add that the guideline authors believe there is nothing wrong with doing single-gene tests. “However, turnaround time is an issue and the ability to test for other potential targets makes us recommend that multiplexed genetic sequencing be used. This is in part, of course, because tissue is often limited. We realize that NGS is not something that will always be possible in every setting.”

Another important change is to the current preference for cell blocks over cytology smears for specimen testing. “Now, we’re going to say that cytology smears or cell blocks are both equally suitable for lung cancer biomarker molecular testing.”

The National Academy of Medicine, Dr. Cagle noted, sets a high bar for recommendations. “When we recommend something, that means there’s a great deal of evidence” in support. With this standard as a basis, the revised guidelines address five new “key questions” for pathologists and clinicians:

1. What other genes, previously not addressed, should be tested in lung adenocarcinoma?
2. Is immunohistochemistry reliable for screening for ALK translocations?
3. In patients who are undergoing treatment with targeted tyrosine kinase inhibitors, what are the types and rates of secondary resistance?
4. Are there biomarkers that are predictive of clinical outcome in squamous and small cell carcinomas?
5. What are the clinical performance characteristics of circulating DNA/CTC in plasma when used for diagnosis of primary lung adenocarcinoma or relapse?

As to question No. 1—genes, previously not addressed, that should be tested in lung adenocarcinoma—the candidates the guideline authors chose had strong research literature on their side. They are ROS1 (one to two percent rearrangement), RET (one to two percent rearrangement), BRAF (four percent mutation, half are non-V600E), MET (three percent, exon 14 deletion and amplification), ERBB2/HER2 (two percent mutation), and KRAS (30 percent mutation).

Even as they were drafting the first guidelines, Dr. Cagle said, the authors knew that ROS1, based on case reports and small series, would soon join the list of genes that should be tested. “Oncologists were often asking for ROS1 testing and were treating with crizotinib on that basis off-label,” and the FDA and clinical trials have reinforced ROS1’s place. “So we will recommend that physicians use ROS1 molecular or cytogenetic testing on all lung adenocarcinomas in selecting patients for ROS1 targeted therapy.”

For ROS1, there is no companion diagnostic so a test will not be specified. The guideline will say ROS1 immunohistochemistry can be used as a screening test in lung adenocarcinoma; however, it needs to be confirmed by a molecular or cytogenetic method when found to be positive.

For the other markers (RET, BRAF, HER2/ERBB2, MET, and KRAS), the guideline authors came to a different conclusion based on the literature. For KRAS itself, Dr. Cagle said, there is no particular drug, so this was investigated but determined it would not be a target at this time.

The others have been found to be potentially targetable, but there is no FDA approval and clinical trials are limited at this time, Dr. Cagle said.

“We say that adenocarcinomas can be tested for these and maybe test other cell types, again when certain clinical criteria suggest there could be an oncogenic driver. But these markers are not indicated at this time as routine standalone tests outside the context of a clinical trial.” The markers could, however, be performed as part of a large panel employing next-generation sequencing, either in an initial workup or when routine EGFR, ALK, and ROS1 testing is negative.

Asked whether testing for ROS1 will become a new standard of care, Dr. Cagle confirmed it would, just like EGFR and ALK. “And it will have the same basic stipulations surrounding it.” In terms of which stages to test, clinical trials and FDA approvals have looked at using ROS1 in advanced-stage lung cancers, but he says early-stage patients are being investigated to see if they might benefit from testing too. “We know that with the patients who are in early stage who are presumably cured by surgery, at least half of them are going to die of their lung cancer at a later date.” Clinical trials are now underway on testing and targeted therapy for EGFR and ALK in those early-stage resected lung cancers.

In the meantime, he said, “There’s no reason that any particular hospital cannot go ahead and do the testing anyway.” His own hospital, Houston Methodist, is now doing reflex testing for all stages.

“When the pathologist makes a diagnosis for nonsquamous, non-small cell lung cancer, we’re going to go ahead and send it for testing” on a reflex basis. “We’re not even asking what stage it is, and the hospital will absorb the costs on that.” The reflex panel consists of EGFR, ALK, and ROS1, as well as RET, BRAF, and HER2, looking for mutations, primarily the exon 20 insertions. In lung cancer, he noted, HER2 is used quite differently from breast cancer. “For MET, we’ll be looking at the exon 14 skipping mutations.” A new in-house NGS translocation assay will replace the use of FISH, he said.

For ALK, “there can be some discordance between FISH and ALK immunohistochemistry, but it generally shows very excellent concordance. The FDA has approved the Ventana D5F3 ALK antibody and assay for the drug crizotinib. So if one has appropriately approved antibodies that are validated in-house, we think this is equivalent to FISH.”

For lung cancer, the ALK1 antibody that was used for anaplastic lymphomas is not a good antibody, he said. “We’ve known that for a long, long time. But for the newer antibodies, you’re going to need access to some positive cases in order to validate them.” He believes some of the pharmaceutical companies would be willing to help if a laboratory wishes to set up such a validation.

ROS1 is also a good antibody, the literature indicates. “But it’s not to a point yet where we would say if you have a positive ROS1 that you do not have to do the FISH or some other test. So if you screen with ROS1 and it’s positive, you will want to go ahead and confirm with FISH or PCR, or some other method.”

When laboratories perform individual biomarker tests sequentially, some want to do the KRAS test first, Dr. Cagle noted. “If it is KRAS positive, that will exclude other biomarkers including EGFR, ALK, and ROS1 most of the time, though not 100 percent of the time.”

Repeated debate has taken place on how such sequential testing may use up what limited tissue is available before testing shows the potentially targeted biomarker. “Then the other thing, of course, is the amount of time it takes to do testing sequentially,” Dr. Cagle added. He maintains that it is better—in terms of time spent and preservation of limited tissue—to try to do the multiplex testing.

Regarding the second key question in the revised molecular testing guidelines—whether IHC is reliable for screening for ALK translocations—the guideline authors answered in the affirmative. The literature and Food and Drug Administration approvals back up the finding that physicians can use immunohistochemistry as an equivalent alternative to FISH for ALK testing. “So we’ll be making that recommendation,” Dr. Cagle promised.

On the subject of patients’ secondary resistance to the tyrosine kinase inhibitors for EGFR and ALK (question No. 3), Dr. Cagle said there are several considerations. “There are drugs, for example, that can be used in patients who have developed resistance due to EGFR T790M mutations.” The guideline authors decided that the testing for the EGFR T790M mutation is worthwhile where the patient has developed resistance to EGFR TKI, because resistance is a frequent occurrence. “For the others, the main thing is just whether there are appropriate performance characteristics for the detection of secondary mutations, including for ALK.”

The authors decided against recommending any biomarkers as predictive of clinical outcomes for therapy for squamous and small cell carcinoma (question No. 4). Various mutations have been identified in squamous cell carcinoma, Dr. Cagle said. “But based on the literature review, none have reached the point where testing specifically for those mutations can be recommended, although this may change in the future.”

Also still awaiting further evidence are the clinical performance characteristics of the circulating DNA and tumor cells in plasma when used for diagnosing primary lung adenocarcinoma or relapse (question No. 5). While the evidence is not yet strong enough for the authors to make a recommendation to use circulating tumor cells, “this is a very exciting area,” Dr. Cagle said.

The comment the authors agreed on is that physicians can use cell-free DNA from the plasma if tissue is limited or insufficient for molecular testing. “In particular, cell-free DNA for EGFR T790M testing may be used when there’s been progression on an EGFR TKI,” the guidelines will say. If the plasma test is positive, then treatment with one of the new drugs might be warranted; if negative, then another biopsy of tissue should be considered to determine whether there is a false-negative.

When Dr. Finley began practicing oncology, roughly a generation ago, median survival of lung cancer patients was six months; it is now one year. Moreover, the targeted therapies that molecular testing has made possible have reduced the toxicity and associated discomfort of treatment. Applauding these trends, he said, “We hope chemotherapy will be relegated to the ash heap of history as we better define the molecular genetics of lung cancer and are better able to utilize the immune system to fight the disease.”

The subset of EGFR-driven NSCLC patients has certain clinical characteristics, Dr. Finley noted. “They are often Asian, because the prevalence of mutations in Asians is high. They’re often women, and they are often never-smokers, and they have this nonsquamous or adenocarcinoma histology.”

Some large studies of those with EGFR driver mutations (LUX-Lung 3 and LUX-Lung 6 trials) have shown no difference in overall survival rates between those who were initially treated with targeted therapy and those whose initial treatment was chemotherapy. But many of the patients in those studies were switched at time of initial progression to targeted therapy after starting on chemotherapy, Dr. Finley said, which is the likely explanation for the same overall survival.

“Subsequent subset analysis demonstrated that if you have the most sensitive mutation of the EGFR driver mutations—that is, the deletion in exon 19—you in fact had an overall survival benefit if you started with the appropriate targeted agent—in this case, afatinib [Gilotrif], or you could imagine using Tarceva [erlotinib] or Iressa [gefitinib] as well.”

The subset with this targeted therapy has unusually high median overall survival of about 2.5 to three years, he said. “This to me is a very striking result, and it continues to drive my thinking around molecular testing and the importance of it.”

Unfortunately, when he and others searched the registry data at two of Allegheny Health Network’s hospitals, of about 200 patients with nonsquamous NSCLC, “we found we were only testing about half of the patients for these important drivers and that the frequency of driver mutation was lower than what was seen in the published literature”—eight percent versus about 15 percent for EGFR and two percent versus about five percent for ALK. One finding that fell out of the data was that the hospitals were testing more patients with large biopsies than patients who had small biopsies. And this was true for both EGFR and ALK FISH testing.

He set out to determine via surveys and more why AHN’s 50 percent testing rate was lower than the 70 to 80 percent in the literature. One of the barriers to molecular testing was the Medicare 14-day rule. Another was the physicians’ unwillingness to subject the patient, who often had multiple comorbidities, to repeat testing. “So we had a number of meetings to try to increase awareness and stakeholder education, and we came up with some pathways to encourage large tissue biopsies, to re-biopsy if necessary, and if unable to re-biopsy, to use blood-based testing. And we’ve been incorporating this into our general practice both at the community and the academic level.”

Demographics were part of the explanation for the lower prevalence of the driver mutations, Dr. Finley noted. Allegheny Health Network extends from tertiary or quaternary care sites in large Pittsburgh hospitals all the way to community sites in rural Punxsutawney, but western Pennsylvania as a whole has few Asians and a large population of Caucasians, the elderly, and smokers, all of whom have low mutation rates. “So the idea of using blood-based testing is very appealing because it would then allow us to interrogate more tumors without the patient having to undergo an invasive biopsy.” Tumor heterogeneity as well is occasionally seen and is another reason to opt for a blood-based assay.

“The advantage of a blood-based assay is that as all cells within the tumor mass itself undergo apoptotic or necrotic cell death, they release their contents into the blood, and this can be detected as cell-free plasma DNA through a variety of techniques.” This liquid biopsy, Dr. Finley said, is likely to be used more and more in the future, to look for driver mutations and to use tumor cell capture techniques that might allow interrogation of cells for PD-L1 expression. “Exosomes, which contain DNA, RNA, and protein, and perhaps will have some interesting signatures with respect to either proteins or microRNAs, may be useful in stratifying our patients.”

With the goal of testing all adenocarcinomas or nonsquamous cancers for the key genes, Allegheny Health Network is now requiring that all stage IIIB and stage IV be tested and is looking at reflexive testing for all lung cancers. “We are moving toward reflexive testing of all nonsquamous cancers. We prefer large tissue biopsies. We like to avoid the delay associated with the Medicare 14-day rule and we’ve sort of eliminated that as a barrier. We do re-biopsy, and we have used blood-based testing very successfully.”

Whether they start with chemotherapy and move on to targeted therapy or start with targeted therapy alone, “patients with driver mutations have a better prognosis in general.” They’re a “lucky subgroup of patients,” he said, and “it behooves us to identify them.”

[hr]

‘A marriage of virtual and real bronchoscopy’

[hr]

As a community hospital, Southside Regional Medical Center in Petersburg, Va., has some differences in resources, said David R. Penberthy, MD, MBA, medical director of radiation oncology at Southside Regional, which serves an area that has high cancer incidence. But the cancer care program where he works has received a “gold” level Outstanding Achievement Award for the last two survey cycles from the American College of Surgeons’ Commission on Cancer. “When ACCC [Association of Community Cancer Centers] approached us about looking at our rates of molecular testing, we wanted to compare ourselves to the standard benchmarks out there and see how we’re doing.”

A process improvement plan was eventually the result. In looking at cases between Nov. 2, 2012 and Oct. 31, 2013, “we found that we performed molecular testing on only four of six stage IIIB and seven of 17 stage IV lung adenocarcinomas.” But collecting the data was difficult.

“We thought we could answer the question more efficiently and it was suggested that we add molecular test results as a data field in the lung cancer registry, then track the molecular testing in a pathology database.” Southside Regional’s pathologists did exactly that, and the subsequent analysis was more easily obtained.

In the second assessment, in 2015, there were 138 cases in the lung cancer registry; of those, 47 were adenocarcinomas and three had mixed histologies (adenosquamous). Molecular testing was performed on 80 percent of stage I, 60 percent of stage II, 58 percent of stage III, and 61 percent of stage IV. The higher percentage in stage I is owing to the stage often having been undetermined at the time of biopsy, Dr. Penberthy pointed out. “So we’ve gotten into the habit of testing any sample that is nonsquamous.”

As frequently happens, collecting the data helped trigger improvement. “We had an average of about 6.6 days turnaround time for the first year, so it was taking almost a week to get results back. But the most recent six-month period showed we’re starting to improve, and we got a little smarter as we went along with our process, partly due to some improvements such as reflexive testing.” The goal, he said, is a turnaround time of five working days. “Typically, the oncologist will order the molecular testing and then schedule follow-up about two weeks later to discuss the results and formulate the treatment plan.”

The hospital found there were many reasons molecular testing was not performed. “Sometimes a test was ordered and the sample was inadequate, and sometimes patients have significant comorbid conditions and did not want to re-biopsy. We also had some physicians not ordering the test because the patient was end-stage or referred to a tertiary care facility.” Some patients refused treatment and one refused to be evaluated by medical oncology.

Other challenges are third-party payment issues and the limited time that’s available before the management decision must be made. “And we have run into the Medicare 14-day rule on more than one occasion,” Dr. Penberthy said. If there’s a potential delay in obtaining results, that can be a problem for the patient who wants treatment to begin quickly.

“Our molecular testing process is moving toward reflex testing,” he said, “but it’s not happening all the time.”

So what can be done to standardize testing and reduce the delays? Document the reasons molecular testing isn’t done so it can be tracked (“we need a better documentation process,” he said), analyze testing rates, set a protocol for mode of biopsy (“right now that’s still operator choice,” he said), and move to reflex testing to expedite clinical decision-making. But as for so many other things, “I would say it all boils down to people and collaboration,” Dr. Penberthy said. For them and others, that means tumor boards and telephone calls.
[hr]

Anne Paxton is a writer and attorney in Seattle.