June 2008
Feature Story

Karen Titus

Oncologists share a clear goal—pick the best treatment for each patient. In the era of targeted therapy, that’s easier for some cancers than others. With chronic mye­loid leukemia, they can turn to Glee­vec. With breast cancer, antiestrogens and HER2/neu inhibitors are options for patients expressing these receptors.

“The question is, when are we going to get individualized therapies for lung cancer?” asks Paul A. Bunn Jr., MD, James Dudley chair in cancer research and professor and director, University of Colorado Cancer Center, Aurora.

The “when” in his question implies certainty. That stands to reason, given the number of promising non­che­mo­therapy agents available or being considered for first-, second-, or third-line treatment as well as in adjuvant settings. Many are related to epidermal growth factor receptor, including the EGFR ­tyro­sine kinase inhibitors erlotinib (Tarceva) and gefitinib (Iressa) and the monoclonal antibody cetuximab (Erbitux), which binds to EGFR. Another monoclonal antibody, bevacizumab (Avastin), inhibits the action of vascular endothelial growth factor, or VEGF, and is an antiangiogenic agent.

But if the trains are on the tracks, there’s no telling when they’ll pull into the station. Sorting through who should get what requires layers of tests: histologic subtyping, EGFR mutations, EGFR amplification, K-ras mutations, and perhaps even protein expression. When pathologists are faced with lung cancer patients being considered for a first-line or adjuvant treatment, Dr. Bunn says, “They ask us, ‘What do you want us to do?’ ”

“It’s becoming very complicated,” says Sanja Dacic, MD, PhD, assistant professor of pathology, University of Pittsburgh Medical Center. Solvable, yes, but probably not soon. Imagine the “Sabre Dance” sans xylophone and trombones—the piece will reach its destination, but with far less propulsion.

In fact, only recently have pathologists begun making one big change needed in lung cancer diagnostics—distinguishing between different subsets of non-small cell lung cancer, which are the objects of targeted therapy.

That’s the first step. “Physicians want to know precisely what kind of tumor it is,” says Dr. Dacic.

In the past, lung cancer histology involved little more than distinguishing between small and non-small cell tumors; only the former was considered to be sensitive to chemotherapy. “Fortunately, that has changed,” says Fred R. Hirsch, MD, PhD, professor of medicine and pathology, University of Colorado Denver School of Medicine. “But that makes a huge responsibility for the pathologist to make an exact classification.”

“Up until two or three years ago, we were allowed to make the diagnosis of just non-small cell lung carcinomas,” says Dr. Dacic. Indeed, she says, “Some people think non-small cell carcinoma is still an acceptable diagnosis. It’s not.”

Pathologists should now go beyond H&E in cases where a histologic diagnosis is not clear, Dr. Dacic says. “This entire field of targeted therapies is going to change our practice, and we just have to adjust. It should be like a reflex: Once you get a lung cancer and you cannot make a diagnosis on H&E, just order a histochemical stain that is cheap, available, and easy to interpret.” A mucicarmine is one good option, she says.

The need to make NSCLC distinctions comes chiefly from two sources, says Marc Ladanyi, MD, attending pathologist and chief, molecular diagnostics service, Department of Pathology, Memorial Sloan-Kettering Cancer Center. One is that EGFR mutations, which identify patients who are most likely to respond to the EGFR inhibitors, are found almost exclusively in adenocarcinomas as well as occasionally in adenosquamous carcinomas. “And the feeling is that the few squamous carcinomas that have been reported to have EGFR mutations may have been inadequately sampled adenosquamous carcinomas,” says Dr. Ladanyi, adding, “That’s kind of hard to prove.”

The second source relates to Avastin, which offers significant survival benefit for patients with NSCLC when combined with the chemotherapy drugs paclitaxel (Taxol) and carboplatin; however, patients with squamous cell morphology who receive Avastin face a significant risk of fatal pulmonary hemorrhage, or FPH. In fact, for any antiangiogenic agent, “Histology is a major issue,” says Dr. Bunn.

It’s not clear yet whether the culprit is the tumor’s histology or its central location next to blood vessels. “What is the egg and what is the chicken?” Dr. Hirsch asks.

One of the authors of the study that raised this concern, done by the Eastern Cooperative Oncology Group, or ECOG (Sandler A, et al. N Engl J Med. 2006;.355:.2542–.2550), notes that he and his colleagues have another paper in press in the Journal of Clinical Oncology looking at the factors associated with FPH. That analysis, says David H. Johnson, MD, indicates the only factor that seems to correlate with FPH is tumor cavitation present at diagnosis, and not location per se.

Dr. Johnson, deputy director and Cornelius Abernathy Craig professor of medical and surgical oncology, Vanderbilt-Ingram Cancer Center, Nashville, Tenn., says he and his colleagues continue to analyze the data from the ECOG trial that led to Avastin’s approval. Though histologic evidence of predominantly squamous-cell cancer—that is, 50 percent or more of the tumor is squamous—was one exclusion criterion, Dr. Johnson notes that the study did not include a central pathology review. In addition, approximately 20 percent of the 878 patients enrolled were classified as not otherwise specified, or NOS, by the local pathologist. If sufficient biopsy material had been available for review, some of these patients might have been found to have predominantly squamous-cell tumors. Dr. Johnson is looking at outcome and toxicity in patients with a NOS diagnosis versus those classified as unambiguous adenocarcinoma, and says he hopes to have more information this summer.

That may be only the tip of the iceberg. Says Dr. Bunn, “This goes beyond Avastin to the small-molecule inhibitors as well.” The ESCAPE trial, which looked at carboplatin and paclitaxel with or without the multitargeted kinase inhibitor sorafenib (Nexavar), was stopped earlier this year in part because survival was significantly worse in patients with squamous-cell carcinoma, he says. (The addition of sorafenib was not more effective than chemotherapy alone.)

The various subtypes of NSCLC tumors exhibit some overlap. “In fact, if you look hard enough in squamous carcinomas, you will find that many of them have at least a little bit of mucin production. So the World Health Organization uses an arbitrary cutoff—if less than 10 percent of the cells are making mucin, we still call it squamous,” explains Wilbur A. Franklin, MD, professor of pathology, University of Colorado Denver Health Sciences Center.

Another murky entity is bronchioloalveolar carcinoma, or BAC, which seems to be a specific entity with its own prognosis and response, especially to EGFR inhibi­tors. At a recent meeting held by the International Association for the Study of Lung Cancer, participants discussed revising the definition of adenocarcinoma, particularly BAC. “It is a work in pro­gress,” sighs Dr. Hirsch.

Adding to the difficulties, small sample size can make it difficult to render a precise diagnosis, particularly in patients with advanced cancers. Dr. Johnson alluded to this when he mentioned the lack of biopsy material for review. “We often get scant material for review in patients with stage IV NSCLC—typically FNA cytology,” he says.

The problem often spills over into the next layer of testing. Because so many patients either relapse and require second- or third-line treatment, or progress on first-line treatment and are considered for an inhibitor therapy, it’s important for the molecular diagnostics team to have adequate tissue to work with for additional testing. “If you have 10 slides and they’re used up by immunos in surg path,” Dr. Ladanyi says, then the oncologist will need to request additional slides down the road. It’s all about communication—“the oldest advice in the book,” Dr. Ladanyi says. “Surg path needs to communicate with the oncologists about what they need. They may not be looking for a very complete immuno panel on the tumor; they may just want confirmation that, yes, it’s a lung adenocarcinoma.”

All that is merely the first step in lung cancer testing. “Not only do we need to know what the histologic subtypes of lung cancer are—which is actually new in itself—but increasingly we’re being asked to provide some molecular information,” Dr. Franklin says.

The Molecular Assays in NSCLC Working Group recently published its recommendations on EGFR molecular testing in the Journal of Clinical Oncology: Eberhard DA, et al. 2008.;.26:.983–994. If nothing else, the discussion demonstrates the complexity of the molecular story in lung cancer. “There’s a lot of cross-talk between the different pathways,” Dr. Dacic says.

Dr. Ladanyi’s group at Memorial Sloan-Kettering has moved aggressively into using molecular testing for managing patients with lung cancer. One of the three groups that independently identified the EGFR mutation in lung cancer was from his institution, he says, “and our oncologists quickly realized that this was going to be an extremely useful test.” They implemented it early, in 2004, he says. All lung cancer patients who are resected at Memorial Sloan-Kettering are tested prospectively; if an adenocarcinoma is identified, it’s tested reflexively for EGFR and K-ras mutations. EGFR mutations are, of course, positive predictors of response to EGFR inhibitors; a K-ras mutation is a negative predictor of response to EGFR inhibitors, though its role as a sole selection marker looks iffy right now.

Tumor samples need to contain at least 50 percent tumor cells; when they don’t, Dr. Ladanyi and his colleagues will trim away nonneoplastic areas from the paraffin sections to ensure they’re studying areas that are as tumor-rich as possible. Dr. Ladanyi cautions labs to be aware of this nuance if they outsource DNA-based mutation testing. “If you’re not paying attention to this process, you may send a section of lung with a few tumor nests in lymphatics—when the results come back negative, it’s basically meaningless.”

EGFR mutation testing is straightforward, in Dr. Ladanyi’s opinion. Right now Memorial Sloan-Kettering tests for the two most common EGFR mutations, the exon 19 in-frame deletions (most frequently small deletions of 15 or 18 base pairs) and the exon 21 point mutation, L858R, which together account for about 90 percent of cases. “We use mutation-specific PCR assays that are more sensitive than direct sequencing. We can comfortably pick up about 10 percent tumor cells, and can probably go down to five percent,” he says.

They do the K-ras mutations using direct sequencing, which is not as sensitive as the EGFR mutation testing—it detects about 25 percent tumor cells, Dr. Ladanyi estimates. “That’s the lowest I would be comfortable with,” he says, “so that’s why we’re aiming for at least 50 percent tumor cells in the sample, just to be extra sure we’re well above those sensitivity limits.”

EGFR amplification is another of NSCLC’s many enigmas. “In our hands, EGFR amplification in the absence of EGFR mutation doesn’t seem to predict response at all,” says Dr. Ladanyi.

Yet there’s been no shortage of discussion in the thoracic oncology community about EGFR mutations versus amplification, Dr. Ladanyi says. He suspects the mutations are the fundamental biological lesion, noting that this has been worked out functionally in studies involving mice, with the mutation shown to trigger lung adenocarcinomas. But early studies have also demonstrated the mutant allele often then undergoes amplification, supercharging the signal deregulation in the cells.

Some early patient studies may not have distinguished between squamous-cell and adenocarcinoma tumors. Yet another reason to view early studies with caution is that many lacked statistical power, Dr. Ladanyi says. The studies looked retrospectively at materials from the early clinical trials of EGFR inhibitors, with researchers able to obtain samples from a small subset of patients; among those, only a small subset would have had mutations.

Since then, Dr. Ladanyi says, many more studies, especially from Asia, with much larger numbers of EGFR mutant patients, have definitively shown that treating these patients with inhibitors triggers strong response and extends survival substantially. Gefitinib, which was not approved by the FDA, is a useful drug in Japan and elsewhere in the Far East.

Unlike with EGFR-based therapies, there currently are no good molecular markers for selecting tumors for treatment with Avas­tin on the basis of molecular techniques. Though some candidates are being look­ed at, Dr. Frank­lin says, “The fact of the matter is it’s mostly the benign, stromal tissues that are targets here. So it may be difficult to select a marker in this setting.” Dr. Dacic suggests one possibility may be an assay that distinguishes between the bound and free forms of VEGF.

EGFR is in the same family of proteins as HER2, so many of the fixation and staining requirements and restrictions that accompany HER2 can be applied to EGFR, says Dr. Franklin. While accepted protocols are standard in most laboratories, he says, different fixation methods and practices for embedding aren’t unheard of, so it may be worthwhile for pathologists to make note of the requirements for the demonstration of the antigen in lung tumors. “Almost like a refresher course,” he says.

As with HER2, antibody choice is up for discussion. Dr. Franklin reports having used both the Dako and Zymed antibodies in his lab, with marginal differences between the two. Whether these differences are significant overall is not entirely clear. His lab did a study using the criteria that were used for the BR.21 trial (Shepherd FA, et al. N Engl J Med. 2005;.353:.123–132), a Canadian study that selected a 10 percent cutoff using the Dako antibody. “We’ve used a similar kind of scoring system, except we factored in the intensity of staining,” Dr. Franklin says.

The EGFR signaling pathways are complicated, Dr. Franklin says, and molecular abnormalities in different cancer cells may affect different parts of the pathway. “You also have to take into account not only the object of staining and other testing, but how the drug that targets the pathway works and how test results should be interpreted in the context of the way the drug works. So for the small-molecule inhibitors, one procedure might provide you with specific cutpoints, whereas for antibody-based anti-EGFR therapy, you may be able to use different cutpoints or different tests.” That’s one of the reasons this area is in flux now. “I think it’s still important to know as much about the whole pathway as possible,” he says.

Dr. Bunn sees another resemblance between breast cancer testing and NSCLC testing: At least for now, it seems reasonable to order several tests.

Using IHC, he says, 80 to 90 percent of patients will exhibit positivity. With EGFR gene copy number, 40 to 60 percent of patients will be positive. For EGFR mutation testing, eight percent of patients in Western populations will be positive. For K-ras mutations, about 10 to 12 percent of patients overall will be positive (20 to 25 percent of adenocarcinoma cases).

Though Dr. Bunn is an advocate of FISH testing, he says “It’s most reasonable to get them all.”

“For example,” he continues, “I’ve got a faculty member whom I’m treating, a never-smoking Chinese-American male who had gene amplification for EGFR, did not have an EGFR mutation, and was K-ras mutated.” Treatment led to a complete response, Dr. Bunn reports. “But if I only got a K-ras mutation, then I wouldn’t have treated him.”

His view is also influenced by the FLEX study (a phase three trial comparing vinorelbine/cisplatin to the same chemotherapy plus cetuximab), details of which were to be discussed in an ASCO plenary session early this month. The primary endpoint was survival. To qualify for the trial, patients had to be EGFR positive by IHC. There’s good data to suggest that if one used FISH, the hazard ratio would be considerably better in FISH-positive patients compared with IHC-positive patients. “If everyone just gets an immunohistochemistry, I think the clinician’s not going to get as much information as he might get from both IHC and FISH,” Dr. Bunn says.

The FLEX trial’s primary endpoint of superiority in survival with cetuximab was met, which makes it likely the FDA will approve cetuximab for lung cancer, Dr. Bunn says, probably with a label derived from the study—IHC-positive disease. This would make it the first drug for lung cancer approved in conjunction with a required pathology test. “While this would herald the start of personalized medicine for lung cancer, it might utilize a suboptimal test,” he laughs.

Even with multiple tests, the information may not be enough to guide physicians, given the permutations that can make all but the hardiest oncologists go glassy-eyed.

If Tarceva or Iressa are used as first-line drugs, for example, would it be better to alternate them with chemotherapy? “It’s not good to be giving them at the same time,” Dr. Bunn says. “The question is, can you alternate them and get added benefit of both? That’s something that’s under study but isn’t known. But if you’re going to use cetuximab in somebody who’s got high copy number, you would do it with chemotherapy at the same time.”

Another question: “Let’s say you’re Avastin-eligible, and you’re EGFR positive. Should you get both Avastin and cetuximab?” Dr. Bunn asks. It appears to be safe, based on a Southwest Oncology Group trial; the group now intends to look at it further in a randomized phase three trial. “If it was me, and I was gene amplified and had adenocarcinoma, I would certainly consider them both,” Dr. Bunn says.

Dr. Bunn points to another study, a 1,700-patient trial that compared whether the chemotherapy drugs pemetrexed (Alimta) and cisplatin were as good as another chemotherapy combination, gemcitabine (Gem­zar) and cisplatin. Against non-squamous tumors, the pemetrexed/ cisplatin combo had the upper hand; in tumors with squamous histology, the gemcitabine/cisplatin pairing was more effective. Another subset of otherwise nonspecific tumors did better with Iressa. Subsequently, in Europe, Alimta was approved for both first- and second-line treatment of non-squamous-cell carcinoma. Though it’s not yet clear how or if the FDA will respond, it’s one more reason, as Dr. Bunn sees it, to make histologic distinctions in NSCLC.

And while Avastin is approved now only for non-squamous histology, there’s some discussion it might be useful in patients with squamous-cell carcinoma who were surgically treated who now present with minimal disease or distant metastases.

Don’t expect matters to become easier anytime soon. Early trials were focused on the small-molecule inhibitors, but the next set of data will be teasing out the role of therapeutics based on antibodies. Here there may be a greater role for immunohistochemistry, Dr. Franklin says. “Perhaps that will also bring some life back to FISH, which has been dismissed in many quarters.”

“When you look at just one or two markers in a clinical trial,” he adds, “you’re not going to get to the bottom of the mechanism for individual patients.” That precise context is where the tests will be most valuable, giving lift to the concept of targeted therapies.

Says Dr. Bunn, “We’re just hoping that pathologists are into this as much as oncologists are, because we can’t have personalized medicine without the pathologists doing the tests.”