Chemotherapy drugs

Rapid advances in both molecular diagnostics and the understanding of cancer biology are transforming the cancer research and drug development enterprise, which experts say will lead to new clinical trial paradigms and testing protocols aimed at better matching patients to studies and therapies. The oncology community largely is viewing these events with enthusiasm and optimism for their potential to bring cancer drugs to market sooner, improve patient outcomes, and to make real the dream of personalized cancer therapy. Such efforts, which depend heavily on validated biomarkers, also are expected to transform the role of clinical laboratorians in oncology diagnostics and treatment.

"We're at a very interesting point. We've now spent the last several decades developing cancer therapeutics, but it's always been evident that even very successful drugs work in only a fraction of patients and all too often for a limited time. The reasons for that have become much clearer with the introduction in the last five-to-10 years of new technologies that allow us to characterize in detail cancer biology," said David Parkinson, MD, a venture partner at New Enterprise Associates, a venture capital firm in Menlo Park, Calif. "We now understand that selection of therapeutic tools ought to match the biological characterization of the patient's tumor at each particular point in the natural history of the patient's disease. This means that if we're going to move toward more efficient cancer therapy we have to have parallel development of biologically targeted therapeutic agents and diagnostics which characterize patients accurately enough for the efficient use of those tools." Parkinson for 5 years led Nodality, a molecular diagnostics start-up, and he co-chairs the cancer steering committee of the Biomarkers Consortium, a public-private biomedical research partnership managed by the Foundation for the National Institutes of Health.

He went on to explain that the old model of single biomarker test results being used to select oncology treatments, monitor the effects of those therapies, and switch therapies as treatment resistance emerges is no longer viable. "It's becoming clear that in order to be able to predict whether or not a cancer agent is going to work in a particular tumor setting is going to require a much more complex series of measurements, and the opportunity is that those measurements taken in whole can much more accurately predict outcomes related to therapeutics."

Cancer: A Chaotic Brew

A substantial body of evidence has dashed forever the notion that cancer is a monolithic disease with a straightforward path to cure. "It now is quite clear that diseases 10 years ago we considered one disease are really a compilation of different tumor types, each with a different driving biology. We now realize that in each tumor type if we just treat with a targeted agent, only a fraction of patients are going to genuinely derive benefit," said Funda Meric-Bernstam, MD, professor of surgical oncology and medical director of the Khalifa Bin Zayed Al Nahyan Institute for Personalized Cancer Therapy at the University of Texas MD Anderson Cancer Center in Houston.

Oncologists now recognize the disease as a chaotic brew of genetic rearrangements, mutations, deletions, and amplifications. Toss in epigenetic influences on gene expression and evolution under therapeutic pressure, and the recipe for devising effective treatments becomes considerably more complicated, according to John Mendelsohn, MD, co-director of the Khalifa Institute for Personalized Cancer Therapy and former president of MD Anderson.

"There are unresolved questions that require aggressive research. The heterogeneity of the cancer. The plasticity of the cancer cells. They can convert from being stem cell-like mesenchymal cells to epithelial-like cells and back and forth. The stem cell-like cells have different responses than the epithelial-like cells," he said. "The number of genes we're picking up sometimes is in the hundreds for a single cancer. Most are important for growth and survival of the cancer, but only a few are driving the cancer. We don't know how to identify those yet."

A Sketchy Track Record

Cancer's convoluted pedigree is reflected in the sketchy track record of the oncology therapeutics industry. Up to 70% of late-stage cancer trials ultimately fail to show benefit, and arrive at this unsatisfying conclusion after involving thousands of patients, costing as much as $1 billion, and taking 10 years of rigid adherence to strict clinical trial regulations and protocols.

Yet recent breakthroughs twinning molecular diagnostics and therapies are spurring the industry to work smarter in weeding through candidate drugs and better matching new and existing therapies to patients more likely to benefit from them. Within just the past few years, mutation analysis as a precursor to targeted therapy has become the standard of care for certain tumor types. For example, the U. S. Food and Drug Administration (FDA) has approved trastuzumab for breast cancer patients with HER2 overexpression, and cetuximab and panitumumab for patients with KRAS-wild type metastatic colon cancer (See Table, below).

Selected List of Cancer Treatment-Related Companion Diagnostics
CDx Drug Indication
Biomarker Type Year CDx approval
KRAS Mutation LDT 2006 Cetuximab, Panitumumab Colorectal Cancer
BRAF V600E Mutation PMA 2011 Vemurafenib Melanoma
ALK Fusion PMA 2011 Crizotinib NSCLC
EGFR Mutation LDT 2003 Gefitinib, Erlotinib NSCLC
HER2 Amplification PMA 1998 Trastuzumab Breast Cancer
PMA 2008
BCR-ABL Translocation PMA 2005 Imatinib,

CML: Chronic Myeloid Leukemia; LDT: Laboratory Developed Test in CLIA certified clinical laboratory; NSCLC: Non-Small Cell Lung Cancer; PMA: Pre-Market Approval by the FDA

Source: Clin Chem 2013;59:198–201.

Adding to the handful of tests used to guide clinical decision-making around chemotherapy, two developments in 2011 really captured the industry's attention. In a first for the agency, FDA granted landmark approvals for two drugs and companion diagnostic testing required for patients to receive the drugs. First came approval for BRAF V600E mutation testing in patients with metastatic melanoma as a precursor to receiving vemurafenib, and then days later FDA approved ALK gene rearrangement testing in patients with late-stage non-small cell lung cancer (NSCLC) as a condition of receiving crizotinib.

As significant as these approvals were, they illustrate just how far the industry has to go to achieve personalized treatment based on the patient's tumor biology, and why new drug and biomarker development paradigms are needed, according to Parkinson.

"We have applications of all these new technologies and glimpses of just how useful they can be in matching patients with their therapeutics. Yet we have many challenges, including developing new methodologies and standardizing them if they're going to be released into clinical medicine." In an opinion piece, Parkinson and his co-authors noted just how rare these parallel diagnostic and therapeutic regulatory approvals were, and in a nod to cancer's complexity, they suggested that the value of selecting patients with either the BRAF V600E or ALK mutations is limited because "not all patients respond even within these marker-selected enriched patient groups, and the responses achieved vary significantly in extent and duration" (Clin Cancer Res 1012;18:619–24).

As many as 60% of melanoma patients have the BRAF V600E mutation, but at most only about 7% of lung cancer patients have the ALK gene rearrangement. The latter suggests that conventional drug development strategies, based on measuring the effects of the drug in unselected patients with the type of cancer in question, probably would not have arrived at this biomarker-drug connection, because such a large number would have had to have been enrolled to detect a preferential response in this subset of patients.

New Models for Clinical Trials

Given the caveats of not only the BRAF V600E-vemurafenib and ALK-crizotinib stories but also other molecular diagnostic tests used in oncology clinical decision-making, researchers are pursuing new clinical trial designs. The goals of these efforts are many, including learning earlier in the drug development pipeline which agents will fail or succeed, speeding approvals for the drugs that succeed, predicting which patients will benefit from therapy, and personalizing the use of biomarkers and therapeutics.

Without new and different paradigms, the oncology field will never speed up its knowledge turn, the time it takes for experiments to proceed from hypothesis to results and on to new hypotheses, according to Laura Van't Veer, PhD, Angela and Shu Kai Chan endowed chair in cancer research at the University of California San Francisco. "There are more than 800 targeted investigational agents in development and if we keep doing the traditional way of testing drugs in Phase 1–4 trials where the Phase 3 trial would be a large randomized adjuvant trial, we will continue to need several thousands of patients to prove that a new drug is better than the standard. We'll also continue to wait five-to-10 years for each drug to come to its endpoint. We need to screen drugs more quickly," she explained.

Van't Veer is an investigator with I-SPY 2, one of the most notable in a new breed of clinical trials utilizing an adaptive design based on Bayesian statistics. This groundbreaking $26.5 million, 5-year study involving 20 major cancer centers and sponsored by the Biomarker Consortium, is testing whether adding novel agents to standard chemotherapy in the neoadjuvant setting improves outcomes in women with high-risk, fast-growing breast cancer. The trial, which was launched in 2010, is testing 10 biomarkers and up to 12 different drugs from multiple companies, and is designed to enable investigators to incorporate knowledge gained during the trial into the trial while it is still ongoing.

Patients' estrogen receptor (ER), progesterone receptor, and HER2 receptor status, and their MammaPrint scores are being used to enroll and randomize them initially, and to stratify them within each arm of the study. When the patients have surgery, their tumor response is assessed and evaluated for biology-specific associations. At this point, the adaptive part of the trial comes into play. "Let's say we find that type 1 breast cancer based on biomarker signatures responds particularly well to drug 2, and type 2 cancer biology responds well to drug 1. As the trial continues, patients with those particular biology types will be preferentially randomized into the trial arms where there's a high likelihood of response while the control arm still receives all biological types. This is all based on pre-defined statistical significance," explained Van't Veer. "Our end point is pathological complete remission with a threshold of 85 percent predicted likelihood of success in a 300 patient Phase 2 trial. This means 100 to 200 patients are needed for each arm with a minimum of 60 to find successful drug-biomarker combinations or a failure."

Building Evidence for Biomarkers

At the same time that I-SPY 2 is helping identify which investigational drugs lead to pathological complete response, the research team also expects the trial will advance evidence around biomarkers, potentially moving some from strictly experimental status to being on the pathway for FDA clearance (See Table, below). For instance, Van't Veer's lab developed a gene expression signature that in experimental systems predicted response to a PARP inhibitor, and I-SPY 2 will test whether that experimental finding holds in patients.

Biomarker Categories in the I-SPY 2 Trial

The landmark I-SPY 2 trial is testing 10 biomarkers and up to 12 different drugs. The trial's novel adaptive design aims to enable investigators to incorporate knowledge gained during the trial into the trial while it is still ongoing. In addition to determining which therapies work best with particular cancer biology types, I-SPY 2 investigators also expect the trial to advance evidence around biomarkers, potentially moving some from strictly experimental status to being on the pathway for FDA clearance and widespread clinical use.

Biomarker Category Purpose in I-SPY 2
Established biomarkers*
IDE biomarkers**
Trial stratification, randomization
Qualifying biomarkers Hypothesis testing
Exploratory biomarkers Hypothesis generating

*FDA cleared or approved
**Biomarker investigational device exemption (IDE) by the FDA as part of investigational new drug application facilitates companion diagnostic pre-market approval

Source: adapted from and used with permission of Laura Van't Veer, PhD

Efforts like this will be crucial going forward as more mutations are found with putative associations to various malignancies and therefore certain treatments, according to Vijay Modur MD, PhD, chief medical officer at HTG Molecular Diagnostics in Tucson. "It's exceedingly important to generate clinical evidence so that we don't pocket every mutation in the same space and say the same drug works with all. For example, we know that in melanoma, treatment of BRAF mutations with BRAF-targeted therapy results in a dramatic response. But for colorectal cancer, at least in preliminary studies, the response doesn't seem to be the same, even if the mutation is the same.

In an opinion article in Clinical Chemistry's January 2013 special issue on cancer, Modur argued that in addition to a need to demonstrate clinical utility, both FDA-cleared and lab-developed companion diagnostics have room for analytical improvement. Even FDA-cleared tests, like ER, HER2, and BRAF, have large inter-lab and patient variabilities. Modur and his co-authors called for improved and more flexible approaches to developing oncology companion diagnostics to respond to how quickly evidence is being generated around matching the right patients with the right therapy.

The Downsides of Adaptive Trials

Given the urgent need to press on with companion diagnostic development, the oncology field can't—and isn't—relying only on the adaptive trial model to make leapfrog advances toward personalized cancer therapy. That's because I-SPY 2 and trials like it, including BATTLE in NSCLC, are logistically challenging and perhaps not as well-suited to less prevalent cancers, according to Meric-Bernstam.

"This is a really outstanding model; however, it's not doable on a small scale and it's logistically difficult. If you want to give more than one drug you need to have multiple drugs available potentially from multiple different companies that may or may not be enthusiastic about comparison of their compounds to each other. One also needs a large enough pool of patients to draw from, as well as the commitment to prioritize this trial over all else to ensure that it is successfully recruiting so that all arms remain relevant while the trial is ongoing," she explained. "Those are not easy things to accomplish, and that's why only a handful of trials will be done like this."

I-SPY 2 also required special involvement from the FDA to consider this new paradigm, including using pathological complete response rather than long-term outcomes as the primary endpoint. The agency has acknowledged that diagnostic tests used to measure biomarkers can have a role in the selection of patients who are likely to respond to specific therapies. It also has recommended using analytically validated biomarkers that have strong evidence of being fit for purpose to evaluate patient response to therapy, toxicity, and drug resistance.

The agency's commitment to revisit existing regulatory requirements given recent scientific advances is encouraging, but Parkinson contended that still more regulatory clarity and even new business models for the molecular diagnostics industry are needed. "The FDA, not inappropriately, has said these tests are going to drive the use of regulated therapeutics, therefore we intend to use our enforcement discretion and regulate them. We're starting to see some guidance, but there still is confusion with regard to regulatory interfaces and the need for them or not. This is leading to a lot of uncertainty around business models and hesitancy about putting new financing into molecular diagnostic test companies."

The Vanguard of Trial Enrollment

To continue advancing molecular diagnostics and drug development for patient populations and cancer types not suited to adaptive clinical trial models, leading cancer centers, such as MD Anderson and Memorial Sloan-Kettering Cancer Center in New York City, have implemented protocols for performing molecular subtyping on at least certain populations, and using this information to enroll patients in other clinical trials. MD Anderson has two such initiatives, its clearinghouse and unusual responder protocols.

Under the former, patients with advanced disease undergo molecular testing with a 46-gene panel looking for common cancer mutations. This information is used to assign patients to any relevant clinical trials the institution is involved in. If this process fails to uncover anything significant, the next step is targeted exome sequencing. "This is done in the research setting. The protocol is set up so that if an abnormality is identified and there's a clinical trial that would be relevant, then the treating oncologist can order CLIA validation testing for clinical trial selection," said Meric-Bernstam.

MD Anderson's unusual responder protocol is for patients such as those experiencing an unexpected rapid progression. Here, the focus is on deep characterization of patients' tumors to identify mechanisms of resistance and predictors of response.

Clearly these approaches are on the vanguard and not ready for widespread adoption. Indeed, sequencing performed in MD Anderson's research labs for both its clearinghouse and unusual responder protocols are supported through philanthropy. However, clinical laboratorians across the country will do well to keep abreast of these efforts as this is the direction in which oncology is moving, according to Marc Ladanyi, MD, William J. Ruane endowed chair in molecular oncology and attending pathologist on the Molecular Diagnostics Service at Memorial Sloan-Kettering Cancer Center.

"It's a rapidly evolving field. Everybody assumes we'll end up with these all-purpose assays that will sequence all the major cancer genes. Things are moving very quickly and it's quite possible that within one-to-three years labs that haven't gotten into tumor sequencing yet might be able to leapfrog some of these growing pains and go from multiple single-gene tests to more comprehensive testing platforms that are being developed," he said.

Modur sees reasons for all clinical labs, even those not currently performing tumor sequencing, to keep their dials tuned to this ever-changing scene. "Laboratorians should be able to have discussions with oncologists to determine what the laboratory test menu is going to look like, both for established and qualifying markers. Many oncologists already are requesting that their patients be tested for qualifying biomarkers because the level of evidence for some of them is pretty high," he explained. "Then there are exploratory biomarkers that are driving clinical trials and may be performed without clinical laboratory oversight right now. The field is moving away from that because many trial protocols pre-specify that even exploratory biomarkers have to be done in a CLIA regulatory environment. So laboratorians need to understand the spectrum of biomarkers, categorize them, and have an educated discussion with oncologists on where each one falls and how the clinician's demands for testing can be met."

For Further Information