November 2011
Feature Story

William Check, PhD

A new duo has entered our lexicon—Xalkori and ALK. Admittedly, it doesn’t roll off the tongue like Laurel and Hardy, gin and tonic, Batman and Robin, or ham and eggs. But, to a small subset of patients with non-small cell lung cancer (NSCLC), this combo could be a lifesaver.

On Aug. 26, the Food and Drug Administration announced it had approved the tyrosine kinase inhibitor crizotinib (Xalkori, Pfizer) to treat patients with late-stage (locally advanced or metastatic) NSCLC who express a rearrangement of the anaplastic lymphoma kinase (ALK) gene. On the same day, the FDA approved a companion diagnostic test to determine whether a patient carries an ALK gene rearrangement—the Vysis ALK Break Apart FISH Probe Kit. Xalkori and ALK can be compared with another power couple, sperm and egg: Together, they create a new entity that is greater than either component by itself.

Crizotinib was reviewed under the FDA’s priority review program and approved under the accelerated approval program, which allows the agency to approve a drug to treat a serious disease “based on clinical data showing that the drug has an effect on an endpoint that is reasonably likely to predict a clinical benefit to patients.”

In an unusual move, the FDA approved crizotinib on the basis of a single prospective, phase two, nonrandomized clinical trial, Robert Sweetman, MD, senior director of medical affairs in Pfizer’s oncology division, told CAP TODAY. Moreover, the trial measured not survival but response rate. Using conventional criteria (30 percent or greater reduction in tumor size), crizotinib had a response rate of 61 percent, Dr. Sweetman says, adding: “This is very good for stage three or four disease that had already failed two or three regimens.” Response rates in such patients are usually in the 10 percent to 20 percent range.

Achieving such a remarkable response rate resulted from using the Vysis ALK assay to select patients with tumors that were highly likely to be sensitive to crizotinib. Commenting on the value of the companion diagnostic, Dr. Sweetman says, “We don’t want to expose people to side effects if there is not much potential for benefit.” Use of the companion diagnostic to qualify patients for crizotinib therapy also makes it possible to avoid unnecessary cost, which in the case of crizotinib and other new tyrosine kinase inhibitors is considerable.

The FDA’s approval of crizotinib on the basis of a surrogate endpoint and its simultaneous approval of the ALK assay has been taken to reflect both the value of the new targeted therapy and the agency’s eagerness to promote the use of companion diagnostics. “Companion diagnostics are not a novel concept,” notes Kathryn Becker, PhD, global marketing director for oncology at Abbott Molecular, the manufacturer of the Vysis ALK FISH test. She points to a recent FDA guidance document on this topic, as well as the FDA’s recent approval of Roche’s drug for melanoma, vemurafenib (Zelboraf), together with approval of the Roche companion diagnostic to detect the BRAF V600E mutation that predicts response to vemurafenib. “More pharma companies are partnering with diagnostic companies early in development, then taking the drug and the assay to FDA together,” Dr. Becker says.

“Overall, it’s a fantastic story that will make a big difference to many lung cancer patients,” Marc Ladanyi, MD, of Memorial Sloan-Kettering Cancer Center, says. “It moved very quickly from discovery [of ALK rearrangements in NSCLC] in 2007 to FDA approval in 2011,” says Dr. Ladanyi, who is attending pathologist in the Molecular Diagnostics Service and a senior investigator in the Human Oncology and Pathogenesis Program and the Department of Pathology.

However, Dr. Ladanyi expresses concern about laboratories having to use the FDA-approved assay to qualify patients for crizotinib. “I don’t know yet how the cost of the FDA-approved FISH test compares to the cost of the previous Abbott ASR [analyte-specific reagent] for ALK FISH,” he says. “FDA’s approval mechanism for companion diagnostics could drive up health care costs if the approved kit is much more expensive.” Before the Vysis kit was FDA-approved, the Molecular Diagnostics Service at Memorial Sloan-Kettering had already been offering ALK FISH testing for lung cancer and anaplastic large cell lymphoma using the Abbott ASR for ALK FISH for several years.

John Iafrate, MD, PhD, director of molecular diagnostics at Massachusetts General Hospital and associate professor of pathology, Harvard Medical School, agrees that the new drug-diagnostic combination “is going to be a very big deal.

“With targeted therapies, the diagnostic is almost as important as the therapy itself,” he says.

Dr. Iafrate, too, expresses concern about the impact of the FDA’s approval process for companion diagnostics. He speaks from a position of expertise, having been a leader of the research group that showed in 2008 that ALK rearrangements make tumors susceptible to crizotinib. That work was done using a FISH assay developed in his laboratory, which was also the central coordinating laboratory for the phase one trial of crizotinib. “One complicating issue for pathologists and people who run labs is what will happen to homebrew or lab-developed tests,” Dr. Iafrate says. “I don’t have a great answer to that.” But he has a hunch: “I think we will see in the long term that the most efficient course for society is to have more competition rather than less.” Dr. Iafrate acknowledges it is the FDA’s responsibility to regulate tests to produce the best product for patients. “However,” he says, “there has to be balance between competition in the marketplace and safety for patients.” He notes the particular irony that, after his laboratory developed a FISH test for ALK that was used in the phase one trial, “For FDA to mandate that we now have to use a commercial test that is more costly is kind of frustrating.”

Marileila Varella-Garcia, PhD, professor of pathology at the University of Colorado Anschutz Medical Campus, is another laboratorian who developed her own ALK FISH assay. Starting in 2007, Dr. Varella-Garcia used her laboratory-developed test to investigate ALK mutations in NSCLC. Subsequently, her lab was one of two selected to perform molecular analysis for the phase two crizotinib trial using the Vysis assay. “After FDA approved the [Vysis] IVD, I thought it was very possible that insurance companies would deny reimbursement if we didn’t use the approved test,” says Dr. Varella-Garcia, who is also director of the cytogenetics core in the University of Colorado Cancer Center. “That puts laboratories in a difficult position. So in September we changed to the FDA-approved assay for clinical testing.” But the cost of the assay has increased substantially and she still uses her LDT to answer research questions. “It’s much cheaper and more specific for the individual fusions than the Vysis assay,” she says.

Before discovery of the biological significance of ALK gene rearrangements, crizotinib was just another potential oncology agent going through the usual testing process in unselected patients who had failed other lines of therapy. “Initially, we were not even focusing on lung cancer,” Pfizer’s Dr. Sweetman says. Then, in 2007, data were published showing that between three percent and five percent of NSCLC carried an ALK gene rearrangement (Soda M, et al. Nature. 2007;448:561–566). Dr. Iafrate and colleagues quickly showed that human cell lines harboring ALK rearrangements were susceptible to tyrosine kinase inhibitors like crizotinib that selectively target the anaplastic lymphoma kinase (McDermott U, et al. Cancer Res. 2008;68:3389–3395). Pfizer got the go-ahead from the FDA to evaluate crizotinib in lung cancer patients prospectively selected with a molecular assay, Dr. Sweetman says. “FDA is very interested to make sure, as personalized medicine evolves using tests to select patients, that we have an approved companion diagnostic at the same time as therapy is approved.” Pfizer partnered with Abbott to devise this companion diagnostic for two reasons. First, Abbott made the fluorescent probes used in the LDT that Dr. Iafrate and his group devised. Second, “Abbott has a global capability for product registration,” Dr. Sweetman says.

A phase one trial was conducted in three countries with Dr. Iafrate’s LDT as the confirming assay. In the phase two trial, all 136 patients were selected using the Abbott test kit. The objective response rates and median duration of response for these studies were 61 percent and 48 weeks and 50 percent and 42 weeks, respectively. Overall survival has not yet been reported. In a retrospective analysis, survival of ALK-positive, crizotinib-treated patients in the phase one trial was similar to that of NSCLC patients with an EGFR mutation treated with an appropriate tyrosine kinase inhibitor. One-year survival was 71 percent and 74 percent, respectively (Shaw AT, et al. Lancet Oncol. 2011;12:1004–1012).

Resistance developed in some patients during crizotinib treatment, which is usual with single-agent therapy. De novo mutated ALK genes could explain some cases in which an ALK-positive patient does not have a response to crizotinib (Sasaki T, et al. Cancer Res. 2011;71:6051–6060). In some patients resistance was due to a mutation in the ALK binding pocket where crizotinib fits; other tumors activated an escape pathway. Dr. Varella-Garcia has reported finding EGFR mutations in ALK-positive patients who do not respond to crizotinib. Molecular analysis suggests that second-generation ALK tyrosine kinase inhibitors could be effective against the first type of resistance (Katayama R, et al. Proc Natl Acad Sci USA. 2011;108:7535–7540). For the second type of resistance, an effective approach might be to combine crizotinib with a second TKI, such as erlotinib in the case of an EGFR activating mutation.

Ongoing trials—part of Pfizer’s post-approval commitment to the FDA—include a randomized controlled trial in which single-agent crizotinib is being evaluated versus platinum-based double therapy as first-line treatment in ALK-positive patients. A hint that crizotinib has greater efficacy in untreated patients comes from the phase one trial: Among 15 previously untreated patients, the response rate was 80 percent.

The economic value of selecting patients for crizotinib treatment is underscored by its cost: $9,600 per month. Crizotinib is dosed as long as patients show benefit. “Patients can tolerate this agent for a long time,” Dr. Sweetman says. Progression-free survival in the phase two trial was about 10 months, for a total per-patient cost approaching $100,000. Pfizer has developed co-pay and payment assistance programs for privately insured and uninsured patients.

Compared with the cost of the drug, the cost of the diagnostic is a minor factor. Abbott’s Dr. Becker puts the cost of the Vysis FISH assay at less than $250 per test per patient. Testing 100 patients will cost $25,000. If four ALK-positive patients are found and treated, with a median duration of 10 months, the cost of treatment will approximate $400,000.

As its name implies, ALK TK was first found as part of a fusion protein in anaplastic large cell lymphoma (Morris SW, et al. Science. 1994;263:1281–1284). Before introducing ALK testing for NSCLC, most cancer centers offered the test for lymphomas. “There wasn’t that much volume,” Dr. Iafrate says, “since ALCL is a rare disease. But we did learn from ALCL that ALK immunohistochemistry can be a good marker.”

Susan Jewell, PhD, scientific affairs manager for the ALK test at Abbott Molecular, explains that a fusion between the ALK gene and certain other genes produces constitutive ALK TK activation, which drives the transformation process. In NSCLC, the most common of these companion fusion genes is the one for echinoderm microtubule-associated protein-like 4 (EML4), so that the qualifying mutation is sometimes simply referred to as EML4-ALK. However, Dr. Jewell says, fusion with genes other than EML4 can constitutively activate ALK. The Vysis ALK assay is configured to detect any ALK rearrangement, leading to high sensitivity. Specificity is difficult to define, since the Vysis assay itself has been used as the gold standard.

In its kit form, the assay is more than just a probe, Dr. Jewell says. “We provide positive and negative control slides as well as processing reagents specialized for the probe. We also have a range for what positive and negative should look like.” (With the introduction of the new FDA-approved test, distribution of the Vysis LSI ALK Dual Color Break Apart Rearrangement Probe will be discontinued on Nov. 30.)

While ALK-positive NSCLC patients are more often young and non- or light smokers (Shaw AT, et al. J Clin Oncol. 2009;27:4247–4253), screening for ALK cannot be limited to people with these characteristics. Still, Dr. Iafrate says, “When there is a rare genotype, it is nice to know the clinical profile. When a young never-smoker has lung cancer, the clinician needs to think ALK.”

In Dr. Ladanyi’s testing algorithm at Memorial Sloan-Kettering, lung adenocarcinoma samples get tested first for EGFR and KRAS mutations. If these are negative, the sample goes automatically to ALK FISH testing. “ALK-positive cases are a small percentage of lung adenocarcinoma specimens,” Dr. Ladanyi says. “And FISH is more laborious than a mutation analysis. If we can eliminate cases from ALK FISH testing, that is a big savings in technologist time and money.”

Abbott has established a Web site on which laboratories offering the Vysis kit will be listed (www.abbottALK.com). “Laboratories listed on our Web site will be those that we have confirmed are validating the use of the FDA-approved test,” Dr. Becker says. “When we agree to put a lab’s name on our Web site, we are confirming they are using the FDA-approved test that we sell according to the protocol in the package insert.”

Dr. Varella-Garcia’s laboratory in Colorado was one of the first three listed. She describes her validation process, which was fairly standard: “We took a number of lung cancer specimens with known genotype and started doing interpretation of patterns generated by the new reagent blind to the previous results. Any lab that wants to qualify to do this assay only has to follow CAP guidelines. I always try to be more stringent than the guidelines because there are multiple patterns for positivity. You have to get right at least one of each pattern before you call yourself proficient.”

While Dr. Varella-Garcia considers the Vysis test to be the best current screening test, she says it does have specificity issues. Regarding the EML4-ALK fusion, she says, “We do not yet know the molecular mechanisms underlying the rearrangement. Also, we know that ALK can have two other fusion partners. We do not know if that’s all that’s clinically important.” In this circumstance, it is a challenge to ensure that any test has 100 percent sensitivity and specificity. “From the clinical point of view,” she says, “the Vysis ALK FISH test is the gold standard, because the only patients treated were those selected by this test. It is a good test, but does it have 100 percent specificity? We don’t know.” For specificity, Dr. Varella-Garcia also relies on her own LDT, which detects the three known fusion partners.

ALK fusion proteins can be detected with immunohistochemistry, as Dr. Iafrate mentioned. Good IHC obviates the need for FISH in a majority of cases, he says. Dr. Iafrate’s group developed such IHC with 100 percent sensitivity and 99 percent specificity with “excellent” interobserver agreement (Mino-Kenudson M, et al. Clin Cancer Res. 2010;16:1561–1571). Because expression levels of ALK fusion protein in NSCLC are lower than in lymphoma, they had to use an especially sensitive antibody. If a company were to bring such IHC to the market, Dr. Iafrate says, “it would be easier to show equivalence to the Vysis FISH test. I don’t believe they would have to show that IHC predicts drug efficacy, as the first assay did.”

Dr. Sweetman agrees that the companion diagnostic “could evolve over time.” Pfizer is not tied to a specific method, he says. “Over the last few years IHC has improved.”

Dr. Ladanyi’s team cross-validated IHC employing the same investigational ALK antibody as Dr. Iafrate’s group against the Vysis ALK kit and RT-PCR for EML4-ALK. In 45 cases that were scored as positive by ALK FISH, RT-PCR missed 18 percent of FISH-positive cases, presumably due to variability in the precise structure of the EML4-ALK fusion or the fusion of ALK to another gene. Concordance between FISH and IHC was 97 percent; IHC would have missed only six percent of FISH-positive cases. “But that assumes that FISH is 100 percent specific,” Dr. Ladanyi says.

“Our take is that the ideal course would be to do both FISH and IHC on all cases,” Dr. Ladanyi says. “In the vast majority of cases they would agree. However, there would be specimens where the wrong area of the slide was scored by FISH. Under fluorescence, morphology is less obvious and sometimes it is hard to distinguish tumor from normal cells.” In those cases, IHC might detect ALK-positive tumor cells in another section of the biopsy, which would trigger rescreening by FISH. Dr. Ladanyi acknowledges there would be issues of cost with this approach. “But that would be the ideal scenario,” he says.

If the possibility of an IHC companion diagnostic for crizotinib is vague, so too is the prospect of an LDT. “The Xalkori PI [package insert] indicates that an FDA-approved ALK test must be used,” Dr. Becker emphasizes. “In the clinical data section the Abbott ALK test is specifically named.” Moreover, she says, “The FDA guidance document is pretty clear that the only way for a companion diagnostic to get approval is to be used in clinical trials. You can’t just show equivalence.”

Experienced molecular laboratorians are capable of validating a companion diagnostic, Dr. Iafrate believes. “A board-certified professional who has been running a lab for many years, who knows how to validate a test, whose lab has passed Joint Commission or CAP accreditation—to me that’s sufficient,” he says. “Professional societies, including AMP and CAP, have set guidelines saying how to validate tests.” However, he concedes, “It’s a question of how FDA will be involved.”

Aside from quality, Dr. Iafrate sees a financial aspect to this conflict. “There is some antagonism between the interests of companies making diagnostics and labs using LDTs,” he says. “I personally think that companies need to make high-quality kits that are easy to interpret and reasonably priced and then they’ll make money and develop market share. There is room, too, for LDTs in properly accredited laboratories.”

Dr. Ladanyi agrees that the FDA approval is written to say that laboratories have to use the Abbott ALK FISH test. “In this instance, it’s not so controversial,” he says. “Most labs were already using the Abbott ALK test.” The FDA’s approval of a single kit for a given companion diagnostic is more controversial for the approval of Roche’s melanoma drug vemurafenib paired with Roche’s BRAF mutation assay. “That assay is not so widely used,” Dr. Ladanyi says. “Many labs already have well-validated BRAF assays whose performance characteristics may be equivalent to or even exceed the performance of the FDA-approved assay.” The FDA’s approach may be understandable, Dr. Ladanyi says, “but it creates a problem for LDTs.”

As for the FDA’s apparent requirement to do a whole clinical trial to get an LDT approved, Dr. Ladanyi says, “That lacks common sense. If you have an FDA-approved kit, you can benchmark another kit against it. If it meets or exceeds the performance of the approved kit, then it seems excessive to require another randomized clinical trial.”

While the fate of LDTs as companion diagnostics is cloudy, two things are certain. The first is that diagnostic-treatment combos will proliferate. Dr. Iafrate’s laboratory has an automated system to seek genetic abnormalities that are susceptible to targeted treatments. “That is our paradigm,” he says, “to genotype as many patients as possible and triage them onto targeted therapy.”

The second certainty: Pathologists will have to become more involved in this field. “The team of pathologists and the clinical oncology group working closely together was essential for this trial to be effective,” Dr. Iafrate says. “Transferring this teamwork to community settings will be a bit more complicated. There are not so many molecular pathologists out there.” But as these “stories” continue to excite trainees about the future of molecular pathology, he says, “pathologists need to step up into that role.”