National Academies Press: OpenBook

Improving the Quality of Cancer Clinical Trials: Workshop Summary (2008)

Chapter: New Clinical Trial Designs

« Previous: Introduction
Suggested Citation:"New Clinical Trial Designs." Institute of Medicine. 2008. Improving the Quality of Cancer Clinical Trials: Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/12146.
×
Page 3
Suggested Citation:"New Clinical Trial Designs." Institute of Medicine. 2008. Improving the Quality of Cancer Clinical Trials: Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/12146.
×
Page 4
Suggested Citation:"New Clinical Trial Designs." Institute of Medicine. 2008. Improving the Quality of Cancer Clinical Trials: Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/12146.
×
Page 5
Suggested Citation:"New Clinical Trial Designs." Institute of Medicine. 2008. Improving the Quality of Cancer Clinical Trials: Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/12146.
×
Page 6
Suggested Citation:"New Clinical Trial Designs." Institute of Medicine. 2008. Improving the Quality of Cancer Clinical Trials: Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/12146.
×
Page 7
Suggested Citation:"New Clinical Trial Designs." Institute of Medicine. 2008. Improving the Quality of Cancer Clinical Trials: Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/12146.
×
Page 8
Suggested Citation:"New Clinical Trial Designs." Institute of Medicine. 2008. Improving the Quality of Cancer Clinical Trials: Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/12146.
×
Page 9
Suggested Citation:"New Clinical Trial Designs." Institute of Medicine. 2008. Improving the Quality of Cancer Clinical Trials: Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/12146.
×
Page 10
Suggested Citation:"New Clinical Trial Designs." Institute of Medicine. 2008. Improving the Quality of Cancer Clinical Trials: Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/12146.
×
Page 11
Suggested Citation:"New Clinical Trial Designs." Institute of Medicine. 2008. Improving the Quality of Cancer Clinical Trials: Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/12146.
×
Page 12
Suggested Citation:"New Clinical Trial Designs." Institute of Medicine. 2008. Improving the Quality of Cancer Clinical Trials: Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/12146.
×
Page 13
Suggested Citation:"New Clinical Trial Designs." Institute of Medicine. 2008. Improving the Quality of Cancer Clinical Trials: Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/12146.
×
Page 14
Suggested Citation:"New Clinical Trial Designs." Institute of Medicine. 2008. Improving the Quality of Cancer Clinical Trials: Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/12146.
×
Page 15
Suggested Citation:"New Clinical Trial Designs." Institute of Medicine. 2008. Improving the Quality of Cancer Clinical Trials: Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/12146.
×
Page 16
Suggested Citation:"New Clinical Trial Designs." Institute of Medicine. 2008. Improving the Quality of Cancer Clinical Trials: Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/12146.
×
Page 17
Suggested Citation:"New Clinical Trial Designs." Institute of Medicine. 2008. Improving the Quality of Cancer Clinical Trials: Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/12146.
×
Page 18
Suggested Citation:"New Clinical Trial Designs." Institute of Medicine. 2008. Improving the Quality of Cancer Clinical Trials: Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/12146.
×
Page 19
Suggested Citation:"New Clinical Trial Designs." Institute of Medicine. 2008. Improving the Quality of Cancer Clinical Trials: Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/12146.
×
Page 20
Suggested Citation:"New Clinical Trial Designs." Institute of Medicine. 2008. Improving the Quality of Cancer Clinical Trials: Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/12146.
×
Page 21
Suggested Citation:"New Clinical Trial Designs." Institute of Medicine. 2008. Improving the Quality of Cancer Clinical Trials: Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/12146.
×
Page 22
Suggested Citation:"New Clinical Trial Designs." Institute of Medicine. 2008. Improving the Quality of Cancer Clinical Trials: Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/12146.
×
Page 23
Suggested Citation:"New Clinical Trial Designs." Institute of Medicine. 2008. Improving the Quality of Cancer Clinical Trials: Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/12146.
×
Page 24
Suggested Citation:"New Clinical Trial Designs." Institute of Medicine. 2008. Improving the Quality of Cancer Clinical Trials: Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/12146.
×
Page 25
Suggested Citation:"New Clinical Trial Designs." Institute of Medicine. 2008. Improving the Quality of Cancer Clinical Trials: Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/12146.
×
Page 26
Suggested Citation:"New Clinical Trial Designs." Institute of Medicine. 2008. Improving the Quality of Cancer Clinical Trials: Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/12146.
×
Page 27
Suggested Citation:"New Clinical Trial Designs." Institute of Medicine. 2008. Improving the Quality of Cancer Clinical Trials: Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/12146.
×
Page 28
Suggested Citation:"New Clinical Trial Designs." Institute of Medicine. 2008. Improving the Quality of Cancer Clinical Trials: Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/12146.
×
Page 29
Suggested Citation:"New Clinical Trial Designs." Institute of Medicine. 2008. Improving the Quality of Cancer Clinical Trials: Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/12146.
×
Page 30
Suggested Citation:"New Clinical Trial Designs." Institute of Medicine. 2008. Improving the Quality of Cancer Clinical Trials: Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/12146.
×
Page 31
Suggested Citation:"New Clinical Trial Designs." Institute of Medicine. 2008. Improving the Quality of Cancer Clinical Trials: Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/12146.
×
Page 32
Suggested Citation:"New Clinical Trial Designs." Institute of Medicine. 2008. Improving the Quality of Cancer Clinical Trials: Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/12146.
×
Page 33
Suggested Citation:"New Clinical Trial Designs." Institute of Medicine. 2008. Improving the Quality of Cancer Clinical Trials: Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/12146.
×
Page 34
Suggested Citation:"New Clinical Trial Designs." Institute of Medicine. 2008. Improving the Quality of Cancer Clinical Trials: Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/12146.
×
Page 35
Suggested Citation:"New Clinical Trial Designs." Institute of Medicine. 2008. Improving the Quality of Cancer Clinical Trials: Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/12146.
×
Page 36
Suggested Citation:"New Clinical Trial Designs." Institute of Medicine. 2008. Improving the Quality of Cancer Clinical Trials: Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/12146.
×
Page 37
Suggested Citation:"New Clinical Trial Designs." Institute of Medicine. 2008. Improving the Quality of Cancer Clinical Trials: Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/12146.
×
Page 38
Suggested Citation:"New Clinical Trial Designs." Institute of Medicine. 2008. Improving the Quality of Cancer Clinical Trials: Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/12146.
×
Page 39

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

WORKSHOP SUMMARY  that will develop consensus-based recommendations for moving the field of cancer clinical trials forward. The views expressed in this summary are those of the speakers and discussants, as attributed to them, and are not the consensus views of workshop participants or members of the National Cancer Policy Forum. New Clinical Trial Designs Phase 0 Trials The first session of the workshop was on new clinical trial designs. Dr. David Jacobson-Kram of the FDA began this session by giving his overview of the exploratory investigational new drug study and how it dif- fers from the traditional IND study. The main purpose of the exploratory IND is to assess the likely therapeutic effectiveness of a compound, based on whether it affects its target in people and how long it is active in the body. An exploratory IND study tests a new experimental drug on human subjects prior to a Phase I clinical trial, which is the first traditional test of a compound in humans to assess safety and the dosing of subsequent trials. For that reason, the exploratory IND study is also called a Phase 0 trial. Dr. Jacobson-Kram discussed the current problems in drug development and testing and how various types of exploratory IND studies might help assuage some of those problems. As Dr. Jacobson-Kram noted, we currently face a crisis in drug develop- ment with the number of drugs in the pipeline declining, the number of drug failures increasing, and the costs of developing drugs rising. The FDA finds that less than 20 percent of new molecular entities progress through clinical trials to the point where approval for them is sought so they can enter the market as drugs. Currently about half of drugs in Phase III clini- cal trials fail because of toxicity or a lack of efficacy, Dr. Jacobson-Kram reported using FDA data. “That is really a disaster, because by the time you are in a Phase III trial you have invested an enormous amount of money, resources, and time,” he said. The cost of developing a new molecular entity that makes it to the market is estimated to be nearly a billion dollars. To receive FDA approval for market, most drugs have to undergo three phases of clini- cal testing. Phase I testing determines safety and dose on a small number of individuals. Phase II testing is done on a larger group of volunteers to assess safety and effectiveness. If those tests are promising, a large-scale Phase III is usually done to confirm safety and effectiveness.

 IMPROVING THE QUALITY OF CANCER CLINICAL TRIALS To help abate this crisis in drug development, the FDA published its guidance on exploratory INDs in January 2006. According to this guidance, the goals of an exploratory IND are to • gain an understanding of the drug’s mechanism of action and whether it affects a target relevant to a disease process (pharmacodynamics), • provide information on how the drug is broken down by the body and how long it remains active (pharmacokinetics), • indicate the most promising lead product from a group of candi- date drugs designed to interact with a particular therapeutic target, and/or • reveal where the drug is distributed in the body using various imag- ing technologies. An exploratory IND study is done in a very small number of human subjects, with dosing up to 7 days, and is not designed to be therapeutic or assess the effectiveness of the experimental drug. “This is really important to keep in mind,” said Dr. Jacobson-Kram. “These trials are not designed to treat patients. These are simply experiments that are being done in human beings.” The FDA’s existing regulations are flexible in the amount of preclinical data it requires investigators to submit before conducting an exploratory IND. That data depends on the goals of the investigation, the testing being proposed, and the expected risks. For example, an exploratory IND that tests a single subpharmacologic drug dose would require a minimal data- set from a single animal species. More extensive data would be needed to conduct a repeated-dose clinical study designed to induce pharmacologic effects, but this expanded dataset would still be less than that required to initiate a traditional IND. Exploratory INDs allow sponsors to evaluate up to five experimental drugs simultaneously in the clinic so as to better choose the most promis- ing drug candidate to undergo traditional drug development and test- ing. Exploratory INDs can help to reduce the resources involved in drug development, including the amount of time and drug product needed to select promising drugs, and help to eliminate those that lack promise. The FDA guidance gives examples of several types of exploratory IND studies, including the microdose study, a study design developed by Pharmaceuti- cal Research and Manufacturers of America (PhRMA), and a study design

WORKSHOP SUMMARY  proposed by the National Cancer Institute (NCI) to specifically study experimental cancer drugs. The sole aim of a microdose study is to use imaging or other means to assess where in the body a compound is distributed and for how long it remains in these sites. A microdose is defined as less than 1/100th of the dose calculated to yield pharmacological effects, and less than 100 micro- grams. A microdose study is designed not to induce pharmacologic effects; rather, it can indicate whether an experimental drug reaches its target. The FDA assumes the risks of a microdose study are small and thus only requires a single study in a mammalian species, usually a rat, to assess safety prior to granting approval for a microdose exploratory IND study. The animals in this preclinical study would only have to be dosed a single time via the same route of administration that investigators would use in the exploratory IND study. The animal study must show a minimally toxic dose or show that the doses used in the microdose study would be well outside a toxic range. Genetic toxicology testing on the animals is not routinely needed. (The European Medicines Agency, in contrast, asks for additional safety data, including general toxicity studies using two ways of administering the compound, orally and intravenously, as well as in vitro genotoxicity studies.) Another example of an exploratory IND study is the paradigm pro- posed to the FDA in 2004 by PhRMA. This study tests, in healthy sub- jects or minimally ill patients, up to five compounds that have a common biological target, but might not be structurally related. These compounds are given in up to seven repeated doses to assess pharmacological response, but not a maximum tolerated dose, as is determined in traditional Phase I studies. The risks in the PhRMA paradigm are greater than in the microdose study, so it requires genetic toxicity studies, as well as a repeated-dose toxic- ity study in rodents and another mammal, usually a dog. If the dog shows toxicity at a dose level that does not cause toxicity in the rat, the compound is not included in the exploratory IND, under the assumption that its toxic- ity had not been adequately evaluated to be tested in humans. PhRMA used a database of 106 drugs tested in two species and in Phase I clinical trials to support the safety of its proposed exploratory IND using an analysis that assumed certain starting and stopping doses. That analysis revealed the trials would have been safe under the exploratory IND paradigm. In a presentation to the FDA, PhRMA discussed the advantages of its proposed exploratory IND versus a traditional IND (Table 1). The exploratory IND would accelerate discovery and development of new drugs,

 IMPROVING THE QUALITY OF CANCER CLINICAL TRIALS TABLE 1  Comparison of the PhRMA Exploratory IND and the Conventional IND Conventional IND PhRMA Exploratory IND Active Pharmaceutical • 1–3 kg • 10–300 g Ingredient (API) Preclinical Resources • 9–12 studies • 5–6 studies • 220 rodent and 38 • 170 rodent and 6 non-rodent non-rodent • 9–18 months • 3–6 months Benefits • Full toxicology • Predictable API profile requirement • Escalation to • Faster progression to maximum tolerated clinical trials dose (MTD) in • Capability to evaluate clinical trials candidates based on • Progression directly target activity to Phase II • Better development decisions made more quickly • Early and less costly attrition Disadvantages • Larger quantity of • MTD not established API • Potential delayed • Slower decisions progression to Phase II • Late and costly attrition SOURCE: Jacobson-Kram presentation (October 4, 2007). PhRMA claimed, because it would require a smaller number of animal studies that would take about one-third less time to perform using much less of the tested active ingredient of the drug. There would be a significant savings in non-rodent experimental animals, Dr. Jacobson-Kram pointed out. In addition, the exploratory IND would enable better development decisions to be made more quickly so there is early and less costly attrition of drugs that lack promise. This innovative IND would also give sponsors the ability to evaluate drug candidates based on target activity, and should enable faster progression to clinical trials. The only disadvantages cited for the exploratory IND were that it did not determine the maximum toler-

WORKSHOP SUMMARY  ated dose and thus could potentially delay the progression to Phase II trials, which entails closing the exploratory IND and opening a new, traditional IND, with the standard requirements for toxicology. The FDA used its own data to analyze clinical studies that were pre- ceded by 2-week or 4-week toxicology studies in two animal species and found that the PhRMA paradigm succeeded in identifying safe starting and stopping doses, but in many cases, the dogs or monkeys had lower no- observed-adverse-effect levels. In addition, an analysis of NCI data found that the toxicology data from the nonrodent species more closely resembled what is seen in humans than the data from rodents (Tomaszewski, 2004). “What this is saying is, at least for some cases, the dog better predicts the maximum tolerated dose in the clinical trial, so the exploratory IND might not be then as viable an option,” said Dr. Jacobson-Kram. He noted that the NCI developed its own version of an exploratory IND for oncology drugs. For what the agency termed “first-in-man” stud- ies, researchers should aim to assess the blood levels of the drug needed to induce the desired effect, instead of focusing on toxicity and basing doses for future studies on such toxicology findings. The NCI exploratory IND is used to select promising drugs for life-threatening diseases, primarily cancers, with up to 3 days of dosing in the clinic. Participants for these tests are terminally ill patients without therapeutic options; but because there is no therapeutic intent in the studies, the safety bar is the same as it would be for healthy volunteers. “The thinking is if you are just doing an experiment, why would you make sick people sicker?” said Dr. Jacobson-Kram. In a later presentation, Dr. James Doroshow of NCI added that researchers need to address the ethical issues linked to an exploratory IND by consulting with their research oversight committees, Institutional Review Boards, to develop a process to obtain the appropriate informed consent from patient volun- teers in these studies. According to Dr. Doroshow, because the exploratory IND is not considered therapy, participation in a Phase 0 trial should not preclude patient volunteers from then proceeding immediately to another clinical trial; the usual 3- to 4-week period between studies is not required in these cases. Despite these various Phase 0 study options, the FDA has received only a handful of exploratory INDs, Dr. Jacobson-Kram reported (although it was added later during the discussion that the agency’s recordkeeping of this may not be complete). “Although PhRMA was very excited about this pos- sibility, in the 2 years that this tool has been available it has been used very sparingly,” he said. He offered several reasons for why exploratory INDs are

 IMPROVING THE QUALITY OF CANCER CLINICAL TRIALS not being done more often by drug sponsors, including the slowness of the established drug industry to adopt a new paradigm and the potential that microdose studies do not accurately predict what is likely to be seen with doses in the therapeutic range. But perhaps the biggest stumbling block to more widespread use of an exploratory IND, according to Dr. Jacobson- Kram, is excessive optimism on the part of a drug development team. “No development team thinks that their drug is a loser. So they don’t want to use a tool that is going to kill their drug early, because they are convinced it is going to be a winner,” he said. In the discussion that followed Dr. Jacobson-Kram’s presentation, participants voiced more reasons for hesitancy to adopt exploratory INDs. Oncology researcher Dr. Giulio Draetta of Merck noted that although the exploratory IND is an excellent concept that he and his colleagues welcome, “no established clinical oncologist inside or outside the company would think of these Phase 0 trials as being important for reaching a go or no-go decision about a drug,” he said, implying that more knowledge is needed for such a decision. Dr. Jacobson-Kram countered that exploratory INDs offer more than decisions on whether to move a drug forward in the clinical testing hierar- chy. “With the current paradigm, from the tens of thousands of different structures you synthesize every day, you only take one into the clinic, and that is a big decision. But if you could take a handful of them in people and find the one that looks the most promising, based on clinical data, I think you have a much better chance of succeeding than just choosing that single one based on preclinical data,” he said. Another participant from Merck, Dr. John Wagner, concurred with Dr. Draetta that all of Merck’s exploratory INDs have been in oncology, and asked what can be done to improve the usefulness of an exploratory IND for oncology purposes. Dr. George Mills, who, when he was at the FDA, helped develop the agency’s guidance on exploratory INDs, responded by stressing the usefulness of an exploratory IND that uses imaging to determine which drug in a pool of candidates is the most promising. “All drugs will be promis- ing at some level,” he said. Further, Dr. Mills commented that clarification of which drug to focus on and thus accelerate the decision-making process for the group of drug candidates comes when rates and routes of clearances and target and non-target organ distribution are analyzed. Dr. Jerry Collins of NCI added that another advantage of the exploratory IND is that “it is an open invitation to a dialogue with the FDA.” He added that an exploratory IND reduces the number of toxicology studies needed, which is a distinct

WORKSHOP SUMMARY  advantage for academic researchers, many of whom lack the expertise or resources to conduct such testing. Dr. Jacobson-Kram summed up his talk by stressing the FDA’s commitment to improving the “critical path” to new medical products and sees the implementation of exploratory INDs as an important means for carrying out that commitment. Dr. Mills, Vice President of Perceptive Informatics, Inc., expanded on some of the points Dr. Jacobson-Kram made, but narrowed the focus of his talk to the use of molecular imaging and linked nanotechnology techniques in exploratory INDs. He noted that small pharmaceutical companies and biotechnology companies developing biologic drugs are particularly keen on using exploratory INDs that employ imaging because this approach literally enables investors to visualize the likely effectiveness of a potential drug com- pound by showing if it hits targets such as tumors, abscesses, or the amyloid plaques in Alzheimer disease patients, and whether it is relatively absent in the liver, kidney, or other organs where it could pose toxicity problems (Figure 1). “These companies have limited amounts of funds and need rapid proof-of-concept for the investment community,” Dr. Mills said. “You can take an image and show it to the investment banking industry person, who doesn’t understand our world, but understands from the image that this drug does go to colorectal cancer and the other ones don’t.” But particularly for oncology applications, it is not sufficient for a drug to just reach its target and be concentrated there. Its effectiveness or toxicity also depends on its duration in the target tissues as well as other parts or the body. An exploratory IND that uses imaging can show this effectively, he said. Radiation dosimetry studies can reveal rates and routes of clearance much more quickly and simply than the standard techniques used to determine these endpoints in Phase I studies, he claimed. After his presentation, audience participant Dr. Tim McCarthy from Pfizer pointed out that although an exploratory IND that uses imaging can reveal distribu- tion data to compare compounds, it does not provide information about specificity of the target. Dr. Mills responded that he expected new software and perhaps combination products that might provide that specificity information in the future. Dr. Mills added that the reduction in pharmacology and toxicology studies that an exploratory IND offers, especially one with an imaging component, is another incentive to drug companies. Many companies, he said, have several preclinically developed drug candidates, but are unwill- ing or unable to devote the financial resources to do the pharmacology and toxicology studies needed to take them to the next step. “The exploratory

10 IMPROVING THE QUALITY OF CANCER CLINICAL TRIALS FIGURE 1  Whole-body biodistribution imaging. Time points 1, 2, and 3 show a radio- labeled bio-distribution study of a therapeutic agent as it targets an abdominal tumor. Time point 1 shows no tumor localization in the mid-abdomen; time point 2 shows localization in the abdomen; and time point 3 demonstrates routine clearance of the labeled agent from the body. Brightness of signal corresponds to density of therapeutic agent. Source: Mills presentation (October 4, 2007).

WORKSHOP SUMMARY 11 IND allows those products to come off the shelf and come into human experience to be able to determine if they are going to be promising,” he said. A pre-IND development teleconference with the FDA can determine the exploratory IND’s minimum pharmacology, toxicology, chemistry, and manufacturing and control information needed for all products evaluated, Dr. Mills said. Dr. Mills also stressed the advantages of being able to make simultane- ous comparative assessments of competitive drug compounds in a single study. Drug developers can also do imaging studies to see how new drug compounds compare to standard therapies. Those that do not perform bet- ter than the standard treatment, in terms of distribution and persistence in various regions of the body, are not developed further. He pointed out that sequential assessments of competitive drug compounds can also be done with a series of exploratory INDs so that the first to perform adequately moves on to Phase I trials, and no further testing is done on other similar compounds. Smaller drug companies tend to pursue this vertical approach to Phase 0 testing because it is more cost- and time-effective than the hori- zontal approach where compounds are compared simultaneously, Dr. Mills said (Figure 2). Exploratory INDs can also address the concern recently raised by those pursuing nanotechnology that, when particle size is changed, the potential safety profile is changed as well. “With an exploratory IND, you can do comparative imaging to determine if particle size change will alter the distribution. It is very straightforward and immediate,” he said. Some companies are developing nanoparticles to carry both a therapeutic and an imaging marker, he added. Dr. Mills summarized his talk by concluding, “Exploratory INDs that use imaging can, in 5, 10, or 15 subjects, effectively let you make those business decisions that are so necessary and cost-effective in drug development.” Dr. Mills’ talk was followed by a presentation on how best to use Phase 0 clinical trials in cancer drug development, given by Dr. James Doroshow of the NCI. Dr. Doroshow discussed the recent shift in cancer drug devel- opment from traditional cytotoxic chemotherapies for cancers to drugs that act on specific molecular signaling targets. This shift has created a need early on in drug development for reliable and sensitive tests that reveal if the drug is affecting its target, as well as confirmation of this in people before initiating large clinical trials to assess the drug’s effectiveness. Phase 0 studies can address that need and establish standard operating procedures

12 IMPROVING THE QUALITY OF CANCER CLINICAL TRIALS 1 2 3 4 Horizontal Portfolio Analysis 1 2 3 4 Vertical Portfolio Analysis FIGURE 2  Exploratory IND assessment schemes of competitive drug compounds. The horizontal portfolio analysis assesses biodistribution of drug compounds simultaneously, regardless of performance and cost. The vertical portfolio analysis is a sequential, “top- down” assessment. In other words, “the first to perform, wins”; this is a cost-effective and time-effective approach. In the figure, 1–4 represent exploratory INDs dependent on chemistry, manufacturing, and controls (CMC) and pharmacology/toxicology. Source: Mills presentation (October 4,Figure 22007). needed to appropriately gather data in subsequent clinical studies, according to Dr. Doroshow. Researchers can also use findings from Phase 0 studies to closer approximate a safe, but potentially effective starting dose and limit the patient tissue sampling required in subsequent trials. “These are experiments that need to be performed to allow you to adequately inform the clinical trial, and even though they are not hypothesis driven, they are critical to the process,” he said. Dr. Doroshow pointed out that, for tests of a drug’s effectiveness on tumor cells (or surrogate markers in the blood), clinical researchers often

WORKSHOP SUMMARY 13 do not concern themselves with accurately duplicating in people how those samples were acquired and processed in animals for the same tests. But vari- ability in these standard operating procedures (SOPs) can affect the accuracy of the tests in clinical trials. Dr. Doroshow advocated using an exploratory IND to fine-tune SOPs for human subjects and create the appropriate bridge between what is done preclinically to what is done clinically. Dr. Doroshow gave an example of an exploratory IND he and his col- leagues conducted on the ability of a drug to inhibit the activity of the DNA repair enzyme poly(ADP-ribose) polymerase, known as PARP, in tumors, and how long that inhibition lasted. Before conducting this study, the researchers developed a sensitive test for PARP inhibition in tumor tissues and determined the SOPs for tissue removal, processing, and testing that were followed in the animal studies. “We tried to model the entire clini- cal experiment in a mouse—we had a veterinarian pretend that she was a radiologist and handle all the tissues the same way they would be handled in people,” Dr. Doroshow said later in response to a question posed by an audience participant. The exploratory IND study was done on only 13 patients, yet it gave the investigators the information needed to consider how best to combine the experimental drug with other cancer drugs in future clinical trials. Dr. Doroshow said this information was acquired much more quickly than in a traditional IND study that determines the maximum tolerated dose, yet lacks information on how long the drug affects its target. Dr. Doroshow pointed out that Phase 0 studies, such as the example he gave, are best done on targeted drugs with a fairly wide therapeutic index, as opposed to traditional toxic chemotherapy drugs that have a much narrower range of doses in which they can be safely used. He also noted that his enthusiasm for conducting such studies would be dampened for experimental drugs that lack an accurate and reliable test for the drugs’ effects on targets. “If you are going to go to the trouble of trying to do a proof-of-principle study, there has to be a principle to prove.” He also reiterated the importance of researchers using Phase 0 studies to fine-tune their methods in people prior to progressing to large clinical trials. “It makes sense to take a small number of patients, ask them to volunteer, and to evaluate and develop your methodology prior to using them on a broader scale,” he said. Much of the discussion that followed the Phase 0 presentations focused on how to fund exploratory IND studies. Dr. Richard Schilsky of the University of Chicago pointed out that the Phase 0 study example that

14 IMPROVING THE QUALITY OF CANCER CLINICAL TRIALS Dr. Doroshow gave required an extensive research team, including sur- geons who removed the tumor biopsies from patients and technicians who ran the tests on those samples (Figure 3). “An academic investigator who doesn’t have a drug sponsor to work with may be facing some formidable challenges in putting this kind of team together,” Dr. Schilsky said. Dr. Doroshow agreed and pointed out that because Phase 0 studies are done without therapeutic intent, health insurers are not likely to reimburse costs linked to the study, including computed tomography (CT) scans, biopsies, blood tests, etc. He estimated that, assuming a test to assess whether an experimental drug was affecting its target (pharmacodynamic assay) was already developed, the clinical costs of running an exploratory IND would be approximately $10,000 a patient. One way to address the funding issue would be to use General Clini- cal Research Centers (GCRCs) or other government-funded institutions to conduct Phase 0 studies, said audience participant Dr. David Parkinson from Nodality, Inc. He pointed out that even drug sponsors might balk at the high hospital and surgeon expenses linked to doing the tumor biopsies as was done for Dr. Doroshow’s exploratory IND study. “If you could take that into a GCRC-type mechanism, the surgeons become investigators, and it is out of the hospital billing system,” he said. Dr. Doroshow concurred that using GCRCs for Phase 0 studies would be appropriate, as would using resources of the newly established Clinical and Translational Science Awards consortium, which is funded by the National Institutes of Health’s (NIH’s) National Center for Research Resources. Dr. Parkinson also noted that if Phase 0 studies were viewed as screen- ing strategies to determine which patients could then benefit from a Phase I study of an experimental drug, they might be reimbursed by insurers. But Dr. Doroshow said later in the discussion that one cannot automatically proceed from a Phase 0 to a Phase I trial using the same patients for both without first having done the toxicology studies needed to proceed to a multidose investigation. “If you have already done that up front, that’s fine, but if you haven’t, you would have to now stop and do that as well. You can’t just roll over from one to the other without having the toxicology base for safety for a traditional Phase I study,” Dr. Doroshow said. He is actively involved in developing a molecule that will simultaneously be given in a For more information see http://www.ncrr.nih.gov/clinical_research_resources/ clinical_and_translational_science_awards/.

IND Sponsor • National Cancer Institute Repository • Pharmaceutical Industry Laboratory for Trial Monitor Pharmacodynamic • Investigator Analysis Pathology Laboratory REGULATORY Laboratory CLINICAL LABORATORY for Tissue AGENCY Handling and Processing Research Interventional Drug Bioethics Imaging Radiology Development Clinic Patient Laboratory for Education Review imaging Medical Oncologists Pharmacokinetic studies to determine Analysis feasibility of obtaining biopsies Research Nursing Documentation of patient understanding Clinical Nursing Schedule tumor of the nature of biopsies: the clinical trials Data Managers coordinate with times for drug administration Social Workers FIGURE 3  NCI integrated Phase 0 research team. 15 SOURCE: Doroshow presentation (October 4, 2007). Figure 03.eps

16 IMPROVING THE QUALITY OF CANCER CLINICAL TRIALS Phase 0–Phase I setting as a therapeutic, but with an imaging component to identify if the compound is hitting its target. Another issue raised by discussant Steve Litwin of Biologics Consulting Group was how applicable exploratory INDs are to drugs produced using biotechnology, which are termed “biologics.” His experience indicates that pharmacodynamic and pharmacokinetic data are not useful for biologics. Moreover, simultaneously testing a group of biologics aimed at the same substance may not work because there often are much larger differences between such compounds, he added. “The three licensed anti–tumor necrosis factor (anti-TNF) drugs have somewhat similar effects, but differ enormously and very importantly in the type of opportunistic infections the patients are prone to,” he said. Dr. Jacobson-Kram countered that one could still use an exploratory IND to compare changes in the primary sequence or formulations of biologics that could affect how they are distributed in the body. “I think there really is a role for these types of studies in biologics,” he said. Dr. Mills added that the horizontal drug testing model he presented was actually based on successful testing that was done on biologics. Adaptive Trial Designs In the next session of the conference, biostatisticians Dr. Donald Berry of the MD Anderson Cancer Center in Texas and Dr. Susan Ellenberg from the University of Pennsylvania gave presentations on adaptive trial designs. Dr. Ellenberg noted that “adaptive” simply means that one or more decision points are built into the trial design. How the trial proceeds following each decision point depends on the data observed up to that point. She and Dr. Berry described the many kinds of adaptive trials, including those which, during the course of the trial, adapt their stop date, range of doses tested, degree of randomization, and types of populations accrued and tested (Berry, 2005, 2006). One of the more commonly used adaptive trial designs is one that stops early or continues based on results that indicate how effective the treatment under study is after a limited number of patients have been tested. The stan- dard design for Phase II cancer trials has been of this type for many years, Dr. Ellenberg noted. It is also standard in Phase I cancer studies to have adaptive trials with dose escalation schemes dependent on observed toxic- ity at each stage, she said. Adaptive trial designs can also be used to make a seamless transition between phases in cancer clinical trials, Dr. Berry noted. The data from a Phase I portion of the trial, for example, determines the

WORKSHOP SUMMARY 17 design of the Phase II portion. The possibility of stopping a trial early on the basis of very positive results or very negative results is built into nearly every Phase III cancer trial. The so-called “stopping boundary” for superiority of an experimental agent is usually very conservative and therefore most Phase III trials accrue to their maximally targeted sample sizes. Less commonly used adaptive trials are those that have adaptive bor- rowing, adaptive randomization, adaptive study populations, or adaptive accrual rates. Adaptive borrowing incorporates historical control data or data from other studies in the final study’s conclusion. Dr. Berry noted that pharmaceutical companies may use adaptive borrowing to conduct, in a single trial, studies of a cancer drug in several different types of cancers. Dr. Berry mentioned several studies conducted at MD Anderson Cancer Center that have used an adaptive randomization design when testing vari- ous cancer drug combinations in a multiarm study. For example, in a study of treatments for acute myeloid leukemia, patients were initially random- ized into three different treatment arms. But rather than maintaining an equal number of patients in each arm, the data were analyzed continually and patients were assigned to the better performing arms of the study with higher probabilities. After only five patients had been tested in a poorly performing arm of the study, the assignment probability to that arm became 0 and so it was effectively dropped (Figure 4 and Table 2). The trial ended after only 34 patients had been tested, about half the number that would have been tested in a standard randomized trial in which 25 patients would have been assigned to each treatment arm, Dr. Berry said. Not everyone sees the value to such a study design, he added. Although one journal rejected the study because only five patients had been tested in the one treatment arm, another journal published the study and the journal editor compli- mented the study design. Dr. Berry noted that often with adaptive trials, investigators use math- ematical modeling and simulations to determine the likely relationships among various factors in a trial and trial results. For example, the likely relationship between patient biomarkers and response to experimental treatments is predicted based on data collected during a Phase I trial. This information is then used to determine what types of patients to enroll in the various treatment arms of a subsequent Phase II trial, which then transitions seamlessly from the Phase I study. Adaptive accrual ramps up the accrual rate of patients for a clinical study only after testing on an initial grouping of patients suggests the experi- mental therapy is likely to be effective and worth pursuing further. How-

18 IMPROVING THE QUALITY OF CANCER CLINICAL TRIALS RANDOMIZE Idarubicin Troxcitabine Troxcitabine + + + Cytarabine Cytarabine Idarubicin Standard design n = 25 n = 25 n = 25 Adaptive design Adaptive randomization to learn, while effectively treating patients in trial FIGURE 4  Study design of drug combinations in acute myeloid leukemia. In a stan- dard design, patients would be assigned in equal numbers to each treatment combina- tion. In the adaptive trial design, the data were analyzed continually and patients were assigned to the better performing arms of the study with higher probabilities. SourceS: Berry presentation (October 4, 2007) and Giles (2003). TABLE 2  Study Results—Drug Combinations in Acute Myeloid figure 4 Leukemia Drug Combination Complete Response by Day 50 Idarubicin/Cytarabine 10/18 = 56% Troxicitabine/Cytarabine 3/11 = 27% Troxicitabine/Idarubicin 0/5 = 0% SOURCES: Berry presentation (October 4, 2007) and Giles (2003).

WORKSHOP SUMMARY 19 ever, Dr. Berry did not know of any adaptive accrual trials that had been conducted. He believed such an adaptive trial would be popular among drug sponsors, but noted that it is contrary to their traditional approach, which rewards fast patient accruals from the start. Using adaptive clinical trials has several advantages. Dr. Ellenberg claimed that the primary rationale for doing adaptive designs has tradi- tionally been ethical, especially for cancer studies. “You need to modify or terminate a study when interim data suggest that patients aren’t being optimally treated,” she said. Dr. Berry pointed out that adaptive trials that use information collected early in a trial to better segregate patients into treatment arms likely to be most favorable for them—personalized therapy—result in more patients in a trial assigned to better therapies. His experience with a number of patient groups suggests that adaptive trials will encourage more patients to enroll in cancer clinical studies because patients perceive such trials as offering them better treatment in addition to provid- ing more efficient drug development. Both Drs. Berry and Ellenberg noted that adaptive designs have practi- cal advantages as well because they increase the likelihood that a study will be informative, and enable investigators to end studies early if they are not generating expected favorable results because the original design parameters were inaccurate. “If we see that things aren’t going the way we thought, we are going to end up with data that are uninformative. We want to be able to stop studies early when it looks like they are going nowhere,” Dr. Ellenberg said. Dr. Berry added that adaptive trials often enable faster, smaller, and more successful trials with substantial savings over nonadaptive trials. Dr. Ellenberg concurred that there is consensus that adaptive approaches are appropriate in all phases of clinical research. Although not all proposed adaptive designs are uniformly favored, she added, the concept of adapta- tion is universally accepted. She added that although adaptive trial designs have been used since the 1960s, new types of adaptive designs have appeared in recent years. Improved computing power has stimulated use of Bayesian statistical methods, which are of increasing interest in clinical trials, particu- larly adaptive trials. “These [Bayesian] designs were impractical in the 1950s when these computer models weren’t available,” she said. “There is a lot of excitement now about seeing whether [such designs can improve efficiency], and more people are learning about how to apply Bayesian methods.” Bayesian techniques are well suited to adaptive trials, Dr. Berry pointed out. They enable inferences based on observed data, continual updating, pre- dictive probabilities, and longitudinal modeling. “The Bayesian approach

20 IMPROVING THE QUALITY OF CANCER CLINICAL TRIALS allows you to say, here I am today, this is what I know, where do I want to go, and what are the probabilities associated with going there,” he said. Such analyses require prospective study designs. “It is a lot of work,” he said. “You have to think about what you would want to do for the various kinds of things that happen in the course of the trial. You have to simulate to see what effects various factors have on operating characteristics, includ- ing duration of the trial and sample size,” he said. He added that Bayesian analysis encourages modeling early and late endpoints. “One of the reasons for failure in Phase III trials is that in Phase II we use one endpoint, in Phase III we use another endpoint, and never the twain shall meet. We ought to be using both endpoints throughout and modeling the relationship,” Dr. Berry said. Even when early and late endpoints lack traditional statistical power, they are still useful in Bayesian analyses, he pointed out. “Twenty percent power is better than zero percent power. Even though you don’t learn a lot, you learn something,” he said. Dr. Berry pointed out that FDA Deputy Director Janet Woodcock stated at a recent conference that “improved utilization of adaptive and Bayesian methods” could help resolve the low success rate of and expense of Phase III clinical trials. He added that in the past 7 years, MD Anderson has had more than 200 adaptive trials that used Bayesian techniques. “Adap- tive design in the drug area has become what one pharmaceutical newsletter from Japan described as being a tsunami,” he said. “Virtually every major company is getting involved because they are attracted to the possibility of building efficient trials.” But Dr. Ellenberg pointed out that adaptive trials are not free from con- troversy, particularly when they involve sample size reestimation in Phase III trials based on interim data. There is concern that such reestimation can bias the trial by revealing information about the accumulating data that in turn can change how the trial is being conducted. That concern is based on historical precedents. She noted that in the “old days,” clinical trials were sometimes too adaptive. Often investigators and sponsors would review data as they came in and make decisions about stopping, continuing, or modifying a study based on emerging data. This led to studies that were inappropriately terminated early based on suggestive but not definitive data, and increased the chance of false-positive conclusions because investigators and sponsors would eye incoming data and “assume they had a winner” SPORE (Specialized Programs of Research Excellence) meeting, Baltimore, MD, July 18, 2006.

WORKSHOP SUMMARY 21 as soon as a p value crossed the “less than 0.05” threshold for determining statistical significance for study findings. At that point they would end the trial. This led to the recognition that when changes are influenced by interim data, statistical tests may lose their meaning. Statisticians recognized some structure was needed to allow some mid-course changes that still permit- ted valid inferences to be made about final data. and developed sequential designs in the 1970s and 1980s that allowed regular looks at the accumulat- ing data with the possibility of stopping the trial early, while maintaining the Type 1 error at an acceptably low level. Such designs are now routinely used in Phase 3 trials evaluating treatments for serious diseases. In the mid-1990s, a new approach to adaptive clinical trial design emerged. Unlike the group sequential designs that specified a full sample size but with the possibility of stopping early if results were more impressive than expected, these new designs “started small” and enlarged the number of patients to be included only if the effect seen appeared large enough to be worthwhile, but likely to be smaller than anticipated or hoped for. The start-small approach was attractive to many sponsors because it did not require committing extensive resources to support a large clinical trial at the time the trial was initiated. These new adaptive study designs, like the earlier generation of group sequential designs, also preserve the low false positive rate, a necessary condition for an acceptable design. “You can still have a meaningful statistical test at the end of a study,” said Dr. Ellenberg. According to Dr. Ellenberg, the two main criticisms raised about these new adaptive trial designs is that they don’t necessarily improve trial efficiency compared to standard designs, and that they create the potential for bias in trial conduct by providing information on emerging results to investigators and other interested parties. Because analyses of interim data in adaptive trial designs may reveal the need to enlarge a study population, they are not necessarily always more efficient. In fact, some investigators have shown that standard group sequential designs are always more efficient than the start-small type of adaptive design (Tsiatis, 2003). “This [does not necessarily mean there is no place for such a] design, but it is not so obvious that we are going to be able to do smaller studies on average if we go this route,” said Dr. Ellenberg. She is more concerned about the potential for bias in adaptive tri- als. When sample sizes are increased in adaptive trials, she said, sponsors, investigators, and even investment firms can back-calculate to figure out the interim data on which that change in size was based. This will

22 IMPROVING THE QUALITY OF CANCER CLINICAL TRIALS effectively “unblind” the trial so that investigators, for example, might be more inclined to notice favorable results in patients in the investigational treatment arm once they realize that the data are suggesting some benefit (albeit modest) for that treatment. Equally troubling, investigators may be uncomfortable continuing to enroll patients if there is evidence that one treatment is yielding better outcomes than the other. It is certainly true that use of stopping boundaries for more traditionally designed trials can also permit investigators and sponsors to make certain assumptions about the effectiveness of a treatment midstream in a trial. But those assumptions are much less precise than what can be inferred from an adaptive trial whose recipe for increasing sample size is prespecified, Dr. Ellenberg noted. A way to avoid this problem would be to keep confidential the aspects of the design relating to decisions to stop or enlarge the trial; but that would require sponsors to commit to (from their perspective) an open-ended trial, which is impractical, she said. She concluded by stating that concerns about study integrity should be addressed before adaptive designs become more widely used to change sample size in Phase III trials. “I haven’t seen any solution to this problem,” she said. “You are not going to be able to keep it a secret if the trial enlarges, and you are not going to be able to keep secret what the study design was that led to that enlargement.” In a later discussion, Dr. Berry countered that he thought solutions were possible for this problem. “I don’t have a universal solution, but there are ways of preserving confidentiality in some circumstances,” he said. One discussant revisited response-adaptive randomization designs, in which patients are equally randomized into treatment arms at the beginning of the study, but then preferentially placed in specific treatment arms based on preliminary results. He noted that this design inaccurately assumes that patient characteristics do not change during the course of a trial. Dr. Berry responded that patients do change during the course of a study. For example, at the beginning of a trial involving one or more intensive therapeutic regimens, investigators tend to present the trial to younger patients with more aggressive cancers than to older patients with less involved disease. However, as investigators become comfortable with the regimens and see that they can be given with a minimum of side effects they are more likely to offer the trial to older patients and to patients who have less aggressive disease. But Dr. Berry noted that, although patients change over time during a clinical study, the treatment benefits usually do not. The use of controls

WORKSHOP SUMMARY 23 throughout the duration of the trial provides information about patient drift. These control patients have the same confounding factors as partici- pants at every stage of the trial. “If you use the covariates and you have con- trols over time, you can at least partially resolve the issue,” Dr. Berry said. Targeting Multiple Pathways with Multiple Drugs Research is increasingly finding that a specific cancer can be dependent on more than one altered biochemical pathway. Therefore, treatments that target multiple pathways are more likely to be effective than those that only target a single pathway. However, numerous challenges are involved in determining the best targeted cancer therapies to combine for different cancer types, and in conducting clinical trials of those combination treat- ments. These challenges and ways to overcome them were discussed by two speakers in the session focused on targeting multiple pathways with multiple drugs. The first speaker in this session was Janet Dancey of the NCI’s Inves- tigational Drug Branch. She noted that combinations of more than one targeted therapy should be explored “earlier rather than later in the develop- ment of these agents,” but added that combining investigational drugs prior to their receiving FDA approval for marketing presents multiple challenges. These challenges are not just scientific or medical. Other challenges include sharing data and intellectual property among companies and academic institutions, the greater risk of failure of combination therapies, and regu- latory quandaries related to how best to show efficacy and safety for FDA approval of a combination therapy. Agreements often have to be forged among different industry partners and academic institutions for the development of combination treatments that target multiple cancer pathways. To aid this process and foster early clinical trials of investigational drug combinations, the NCI turned to its Cancer Therapy Evaluation Program (CTEP). This program supports early “proof of principle” trials, which identify the appropriate molecular contexts for effectiveness. CTEP provides template agreement language among NCI, industry, and academic investigators concerning the sharing of data and intellectual property that stems from combination studies. CTEP currently has collaborative development agreements with more than 80 industry partners for more than 100 experimental drugs. CTEP also has clinical trial agreements with academic institutions, consortia, and cooperative groups. The program has sponsored more than 100 trials combining investigational

24 IMPROVING THE QUALITY OF CANCER CLINICAL TRIALS agents, as well as agreements for sharing resources in preclinical studies of 75 investigational agent combinations. Commenting on the success of CTEP in increasing the number of investigation drug combination trials that have been initiated, Dr. Dancey noted that “Our rate-limiting step is not necessarily being able to put together drug combinations. It is actually more to prioritize between pos- sibilities.” Such prioritization depends on overcoming scientific challenges such as determining • which targets would be the most promising to combine, • the best agents that can act on those targets simultaneously, • the optimal types of patients likely to respond to the combination therapies, • the most appropriate dosing regimens for combination treatments, and • the best trial designs and endpoints that can reveal whether the combination is more effective than the treatments used alone. Such determinations are difficult to make given the incomplete under- standing of appropriate targets, agents, and likely responders, combined with the limits of what preclinical and clinical studies can reveal in that regard. Currently, nonclinical studies are the best means for determining mechanisms of action of investigational drugs and the development of use- ful biomarkers that can predict likely responders. These studies also can indicate which drugs have the most promising pharmacodynamics and work best in combination. These preclinical studies are usually done on animal models, but, Dr. Dancey said, “I think we would all agree that there are intrinsic differences between models and cancers in patients, and until we have models that look like patients or patients that look like models, we can’t really have good predictive value from preclinical experimentation.” The limited number of models used to test drug combinations or look for biomarkers that predict response is unlikely to reflect the heterogeneity that occurs in cancer patients, she said. In addition, doses used in preclinical tests and endpoints may not be relevant for the clinical situation. Cancer clinical trial endpoints are usually patient survival or progression-free sur- vival, but these endpoints are rarely used to evaluate for synergy in preclini- cal studies of drug combinations. The controls or standard treatments to For more information about CTEP, go to http://ctep.cancer.gov/industry/ipo.html.

WORKSHOP SUMMARY 25 which new drug combinations will be compared in a clinical trial also may not be appropriate for a preclinical study. To improve these preclinical models, Dr. Dancey suggested that system- atic efforts to molecularly characterize human tumors in nonclinical models “might help us get that much closer to that ideal of matching patients and models.” She also advocated testing drug combinations in multiple tumor models to see if there is consistency in observed effects. Ideally, such test- ing should be done with a dosing regimen that mimics what is clinically achievable. This may require conducting nonclinical studies after acquiring preliminary human data to determine exposures likely to give a desired outcome in the clinic. All these preclinical results should be explained within the molecular context of the models used to better understand why synergy or antagonism occurs, and how that might be promoted or avoided, respectively, in clinical trials. Another major scientific issue is which targets to aim for with combi- nation treatments. Common strategies are to combine agents that target a pathway at the same point to maximize inhibition of that pathway, such as combining an agent that acts on the vascular endothelial growth factor (VEGF) with one that acts on the receptor for VEGF. Another approach is to combine an agent that targets a specific cancer growth factor, such as human epidermal growth factor receptor 2, or HER-2, in breast cancer with another compound that plays a critical role downstream from the activation of that target, such as mTOR (mammalian target of rapamycin). Combinations that block parallel pathways and different cellular processes that underlie a cancer and its progression can also be effective. These include combinations that target both the VEGF receptor and the epidermal growth factor (EGF) receptor, both of which are believed to play a key role in cer- tain cancers. Other combinations include an agent aimed at a major cancer target and a second agent aimed at overcoming resistance to the first agent in the combination. (Dr. Gray expanded on this in his presentation, which is summarized below.) Once appropriate targets are determined, researchers have to select agents that can collectively counter those targets without causing significant overlapping toxicities. Dr. Dancey pointed out that when a drug combina- tion fails, it can fail for several reasons; perhaps the drugs individually or in combination did not effectively interact with their targets or the targets themselves singly or in combination were not relevant. “Therein lies the risk of evaluating combinations early on when you don’t know a lot about

26 IMPROVING THE QUALITY OF CANCER CLINICAL TRIALS the targets in human cancer and you don’t know a lot about the agents and their ability to effectively interact with those targets,” she said. Ideally, investigators should have biomarkers that predict response when testing targeted cancer drugs in clinical trials. But these biomarkers are often lacking, and without them, interpreting the results of clinical trials of combination targeted cancer drugs is especially problematic. To illustrate this point, Dr. Dancey showed a slide of possible outcomes with a three-drug combination therapy (Figure 5). If this combination of drugs Study Population A B A+B+C = 30% 10% 10% Patients benefiting from only 1 agent C 10% Study A+B+C < 30% Population • Unfavorable interaction A B • Overlap of 10% 10% sensitivities C • Cannot exclude 10% possible benefit in subset Study Population A B A+B+C > 30% >10% >10% (or there are cures) C True synergy >10% FIGURE 5  Possible outcomes with drug combinations in unselected patients. In the first example, patients respond to only one agent in the drug combination, so the overall response to the drug combination is additive. In the second example, combinations of agents result in fewer positive responses than if each patient was treated with only one agent. Possible reasons for the decrease in response are unfavorable interactions or that two agents target the same vulnerability. A positive response in a subset of the study population cannot be excluded, however. In5the third example, a greater number of Figure patients experience a positive therapeutic benefit or cure. This is due to synergistic activ- ity of the combined therapeutic agents. Source: Dancey presentation (October 4, 2007).

WORKSHOP SUMMARY 27 A, B, and C is tested in a population in which 10 percent are responsive to each of the three drugs, then a total response rate of 30 percent will be seen even if there is no benefit to the combination. In other words, patients would respond the same to the combination as they would to the individual drug in the combination to which they are responsive. A response rate of greater than 30 percent would occur if the combination is more favorable than the individual agents. But without predictive markers for response to each of the three drugs, as well as predictors of response to the combination, researchers cannot conclude who is likely to benefit from the combination treatment and whether it is more beneficial than individual agents. “Look- ing for predictive markers in the context of developing the combination is probably going to be very difficult, and even more difficult than doing it with the individual agents,” Dr. Dancey said. The search for such markers requires multiple assessments of tumor response to assess markers for initial response as well as markers for the development of resistance, which occurs later in treatment when tumors are enriched with resistant clones. Such assessments are best done preclinically because such evaluations in the clinic are more difficult, complicated, and expensive, Dr. Dancey noted. Determining the best dosing regimen for combination therapies also can be complex because of the need to consider multiple possibilities to discern the best dose and schedule. Often the optimal dose for individual agents within combination therapy is not the same as the dose for each drug when used alone. “This is one that we in particular have been wres- tling with for a number of targeted agents that we have been testing in combinations—because individual single agents are well tolerated, but the combinations induce toxicity,” Dr. Dancey said. For optimal effectiveness, the dosing schedule may need to be altered for the drugs used in combina- tion versus alone. “So if you do have to modify the dosing schedule, that means you have even more potential combinations that you might have to test,” Dr. Dancey said. An efficient approach for this is a multiarmed, controlled trial with an adaptive design, she noted. Dr. Dancey did not discuss the regulatory challenges involved in evalu- ating combination therapies. But in one of her slides, she noted that the FDA may not require clinical toxicology data for the combination if the individual agents have been tested in the clinic and their toxicity is known. However, the FDA does require sponsors to show the contribution of each component of a fixed combination regimen to its total effectiveness. Testing

28 IMPROVING THE QUALITY OF CANCER CLINICAL TRIALS of drug combinations requires early discussion with regulatory authorities, Dr. Dancey noted in her slide. In summary, Dr. Dancey asserted that the rapid emergence of hundreds of new agents on an expanding list of cancer-specific molecular targets offers tremendous hope to cancer patients, while presenting significant development challenges to the cancer research community. The major legal, regulatory, and scientific challenges involved in developing testing strate- gies for combination cancer treatments may be overcome with common agreements among industry and academic partners regarding intellectual property and data sharing; systematic evaluation of targets and agents in predictive nonclinical models; the development of biomarkers that predict response to individual agents or their combination; and controlled clinical trials that assess multiple combinations. Dr. Dancey was followed by Dr. Joe Gray of Lawrence Berkeley National Laboratory (LBNL) and the University of California, San Fran- cisco. He focused on how to model molecular heterogeneity to enhance multidrug clinical trial design. He summarized the efforts by the Greater Bay Area Consortium, which consists of investigators at the University of California, San Francisco, the University of California, Berkeley, LBNL, and SRI International, in collaboration with investigators at MD Ander- son Cancer Center and pharmaceutical company GlaxoSmithKline. He described the immense challenges involved in dealing with the heteroge- neous nature of cancer. Patients with the same type of cancer, or even with tumors that appear the same clinically, may differ in the molecular defects that underlie their cancers or fuel their growth. Dr. Gray stressed that “it’s not just the target, but everything that is going on in the tumor that is important.” Researchers have amassed enough data to provide “a good catalog” of the ensemble of molecular abnormalities that play key roles in the progression and response to treatment for most major cancer types, he added. A slide of his data on breast cancers revealed several portions of cancer cells’ genetic material (genome) that are abnor- mally activated due to duplications (Figure 6). These data indicate that at least 15 percent of the genes and the DNA that regulate their activity (the transcriptome) in breast tumors are activated abnormally. His research suggests that many, if not most, of these abnormally expressed genes play an important role in the progression of cancer and how well it responds to targeted therapeutics. Some of these genes are part of the same molecular pathway and are activated when a linchpin molecule in the pathway, such as the ErbB2

WORKSHOP SUMMARY 29 RAB25 Aberration frequency 14-3-3σ ZNF217 CCND1 FGFR1 MDM2 ERBB2 MYC S6K FIGURE 6  Recurrent copy number aberrations in breast cancer. Ten to fifteen percent of the transcriptome/proteome is deregulated by recurrent aberrations. Functional studies support the concept that many of these contribute to cancer pathophysiology. A: Frequencies of genome copy number gain and loss plotted as a function of genome location. Vertical lines indicate chromosome boundaries, and vertical dashed lines indi- cate centromere locations. Positive and negative values indicate frequencies of tumors showing copy number increases and decreases, respectively. B: Frequencies of tumors Figure 6 showing high-level amplification. Data are displayed as described in A. ACRONYMS: 14-3-3σ (SFN, stratifin), CCND1 (cyclin D1), ERBB2 (v-erb-b2 eryth- roblastic leukemia viral oncogene homolog), FGFR1 (fibroblast growth factor receptor 1), MDM2 (transformed 3T3 cell double minute 2), MYC (v-myc myelocytomatosis viral oncogene homolog), RAB25 (member RAS oncogene family), S6K (ribosomal protein S6 kinase), ZNF217 (zinc finger protein 217). Source: Gray presentation (October 4, 2007), reprinted from Cancer Cell, Volume 10, Chin, K., S. DeVries, J. Fridlyand, P.T. Spellman, R. Roydasgupta, W.-L. Kuo, A. Lapuk, R.M. Neve, Z. Qian, T. Ryder, F. Chen, H. Feiler, T. Tokuyasu, C. Kingsley­, S. Dairkee­, Z. Meng, K. Chew, D. Pinkel, A. Jain, B.M. Ljung, L. Esserman­, D.G. Albertson­, F.M. Waldman, and J.W. Gray, Genomic and transcriptional ­ aberrations linked to breast cancer pathophysiologies, pp. 529-541, Copyright 2006, with permission from Elsevier. receptor, which the drug Herceptin targets, is activated farther upstream. But even when ErbB2 is overexpressed, downstream genes are activated to different degrees in different tumors, according to Dr. Gray’s slide of gene expression in three different breast cancers (Figure 7). “We have to under- stand how these ancillary aberrations, co-acting with the target, condition response,” Dr. Gray said. Fortunately, recent large-scale “omics” technologies that enable simulta- neous assessment of all expressed genes or proteins with automated devices

30 IMPROVING THE QUALITY OF CANCER CLINICAL TRIALS FIGURE 7  Aberration combinations in the same pathway vary considerably among tumors—even in subsets having the same therapeutic target. ACRONYMS: 14-3-3σ (SFN, stratifin), CCND1 (cyclin D1), ERBB2 (v-erb-b2 erythroblastic leukemia viral oncogene homolog), MDM2 (transformed 3T3 cell double minute 2), MYC (v-myc myelocytomatosis viral oncogene homolog), RAB25 (member RAS oncogene family), S6K (ribosomal protein S6 kinase), ZNF217 (zinc finger protein 217). Source: Gray presentation (October 4, 2007). can reveal telltale molecular patterns relevant to specific cancers and how they are likely to respond to various targeted treatments. This information can be used to identify markers that indicate which drug combinations are most likely to be effective for individual cancer patients. But these markers are not usually available until late in the drug development process, so they are not often used to guide early trials or to prioritize which drug combina- tions should be tested preclinically based on the likelihood that they will have synergistic effects. Adding to the complexity is the fact that there are about 100 FDA- approved cancer drugs and more than 400 experimental cancer drugs in Phase II or III trials. The target specificities for most of these drugs are not well known, Dr. Gray pointed out, and clinical tests of these agents are not coordinated or guided by biomarkers. Unfortunately, the cost of molecularly characterizing all available cancer drugs and their effects on genes known to play a role in cancers would be enormous.

WORKSHOP SUMMARY 31 Another approach that Dr. Gray’s consortium and others are taking is to develop preclinical models for the molecular heterogeneity found in tumors that can be used to determine which drug combinations are the best to test clinically and which patients are likely to respond to these treat- ments. He and his colleagues have collected and characterized about 50 breast cancer cell lines that have enough molecular diversity to enable the detection of molecular abnormalities linked to response. These cell lines also seem to adequately mirror clinical findings. For example, the cell lines have the same patterns of gene expression (genetic signatures) that are seen in primary tumors (Figure 8). Even when the cell lines are broken down by type of breast cancer (e.g., luminal versus basal), they closely mimic the gene expression of the primary tumors for each type. Consortium investigators are using these cell lines to test large numbers of drug combinations in an automated fashion. Researchers can currently 1 3 5 7 9 11 13 15 17 19 21 x 1.0 Cell lines Frequency 0.0 -1.0 1.0 Primary tumors Frequency 0.0 -1.0 1 3 5 7 9 11 13 15 17 19 21 x Distance Along Genome FIGURE 8  Cell lines retain the recurrent genomic characteristics of primary tumors. A and B: Frequencies of significant increases or decreases in genome copy number are plotted as a function of genome location for 51 cell lines (A) and 145 primary tumors (B). Positive values indicate frequencies of samples showing copy number increases [Log2(copy number) > 0.3], and negative values indicate frequencies of samples showing fig 8 copy number decreases [Log2(copy number) < −0.3]. Source: Gray presentation (October 4, 2007), reprinted from Cancer Cell, Volume 10, Neve, R.M., K. Chin, J. Fridlyand, J. Yeh, F.L. Baehner, T. Fevr, L. Clark, N. Bayani­, J.-P. Coppe, F. Tong, T. Speed, P.T. Spellman, S. DeVries, A. Lapuk, N.J. Wang, W.-L. Kuo, J.L. Stilwell, D. Pinkel, D.G. Albertson, F.M. Waldman, F. McCormick, R.B. Dickson, M.D. Johnson, M. Lippman, S. Ethier, A. Gazdar, and J.W. Gray, A c ­ ollection of breast cancer cell lines for the study of functionally distinct cancer subtypes, pp. 515-527, Copyright 2006, with permission from Elsevier.

32 IMPROVING THE QUALITY OF CANCER CLINICAL TRIALS test hundreds of drugs and drug combinations simultaneously, and expect to use improved automation techniques to eventually boost such simultane- ous testing to as many as 10,000 drugs or drug combinations. This testing has already revealed various cancer drugs’ target specificities. For example, an AKT inhibitor appears to affect genes abnormally activated in luminal tumors, but not basal tumors. Some of these findings have been confirmed in clinical studies. For example, tests of lapatinib indicated it would only be effective in tumors that overexpress or phosphorylate ErbB2, and this was shown to be true when the drug was tested clinically (Di Leo et al., 2007). Dr. Gray’s studies have also revealed basic information about cancer pathways that will help to optimize targeted cancer treatments. There are two parallel molecular pathways relevant to breast cancer that are activated when ErbB2 is activated—the AKT pathway and the Raf-MAP kinase path- way (Figure 9). Research on the breast cancer cell lines reveals that luminal cancers have an activated Raf-MAP kinase pathway, whereas basal tumors have an activated AKT pathway. This suggests that using drug combina- tions that block the primary pathway activated by a mutation as well as the alternate “bypass” pathway could be effective. Dr. Gray summarized the strength of this modeling approach by point- ing out that cell lines can be characterized in exhaustive molecular detail, unlike patients or their tumor samples, and automated testing techniques can quickly indicate the most effective drug combinations to test clinically. In addition, the mechanism of action of an experimental drug can be easily assessed. For example, if it appears that the AKT pathway is important to a drug’s effects, it can be tested by altering the activity of that pathway in a cell line and seeing if it correspondingly affects the drug’s activity. The in vitro studies can also reveal promising new targets. Researchers have identified only about 20 percent of the genes in the abnormally duplicated regions of the breast cancer cell lines, Dr. Gray said in the discussion following his presentation. The main weakness of his cell line model is that more cell lines, including resistant cell lines, are needed to more completely represent the molecular heterogeneity of breast cancer. In addition, better modeling of the in vivo microenvironment is needed, and some culture-specific aberra- tions may accumulate over time such that the cell lines eventually may not adequately mimic what is seen clinically. Despite that potential problem, Dr. Gray is “fairly confident that this is at least a way forward of helping us

WORKSHOP SUMMARY 33 GFR RAS IRS1 Luminal RAF PI3Kp85 Basal PI3Kp110 PTEN Mek Torc1 Rictor PDK1 PDK2/mTOR erk AKT PKA LKB1 AMPK TSC2 ATP AMP TSC1 Glucose Rheb Amino Acids Torc1 mTOR Raptor Autophagy p27 S6K 4-EBP Cell cycle Motility S6 eI4FE Apoptosis FIGURE 9  Basal and luminal tumors may use different parts of the growth factor sig- naling network. Drug combinations can be selected to block activating mutations and alternate bypass pathways. The signaling pathways shown impact cell motility, growth, and survival. ACRONYMS: 4-EBP (translational repressor eukaryotic initiation factor 4E-binding Figure 9 protein), AKT (v-akt murine thymoma viral oncogene homolog), AMP (Adenosine Monophosphate), AMPK (AMP-Activated Protein Kinase), ATP (adenosine triphosphate), eI4FE (messenger RNA 5-cap binding protein), erk (extracellular-signal-­regulated kinase), GFR (Rap guanine nucleotide exchange factor 5), IRS1 (insulin receptor substrate 1), LKB1 (serine/threonine kinase), Mek (mitogen-activated protein kinase), mTor (Mammalian target of rapamycin), p27 (SSSCA1, Sjögren syndrome/scleroderma autoantigen 1), PDK1 (pyruvate dehydrogenase kinase, isozyme 1), PDK2/mTOR (pyruvate dehydrogenase kinase, isozyme 2), pI3kp110 (phosphatidylinositol 3-Kinase p110 subunit), pI3kp85 (Phosphatidylinositol 3-Kinase p85 subunit), PKA ­ (Protein Kinase A), PTEN (phosphatase and tensin homolog), RAF (a protein kinase), raptor (regulatory associated protein of mTOR), RAS (GTP-activated protein involved in cell growth regulation), Rheb (Ras homolog enriched in brain), rictor (rapamycin-insensitive companion of mTOR), s6 (ribosomal protein involved in translation), Torc1 (Target of rapamycin complex 1), TSC1 (tuberous sclerosis 1), TSC2 (tuberous sclerosis 2). Source: Gray presentation (October 4, 2007); pathways courtesy of Gordon Mills.

34 IMPROVING THE QUALITY OF CANCER CLINICAL TRIALS to prioritize these drugs and drug combinations for introduction into the clinic,” he said. In addition to using cell line models to indicate the most optimal drug combinations to test clinically, researchers can use them to select in vitro response biomarkers that are likely to work in a clinical setting. According to Dr. Gray, ideally, a clinically useful in vitro biomarker would be a genome aberration whose detection does not vary with culture condi- tion. The marker should also be the same in both the cell cultures and the primary tumors. Dr. Gray gives higher priority to transcriptional markers than to protein markers because the former are currently easier to measure, although he acknowledged that genomic markers won’t necessarily inform the biology as well as protein markers. Thus both approaches are ultimately needed. Dr. Gray and his colleagues are currently pursuing an innovative marker-intensive clinical trial of breast cancer treatment that uses a series of core breast biopsies and magnetic resonance imaging (MRI) to determine before-and-during-treatment responses, and the effectiveness of markers in predicting such responses. “We think this is a reasonable way of taking the drugs and markers that come out of our in vitro system and quickly evalu- ating them just for general efficacy in the neoadjuvant environment. Then for those things that seem to be behaving the way that we expect them to, we will introduce them into a later phase clinical trial to assess long-term outcome,” Dr. Gray said. This approach should lead to more efficient clinical trials, he noted, because the early trials would target patient subpopulations most likely to respond, and would be less likely to miss drugs effective against small subpopulations. The model system would also provide a rationale for use of drug combinations that may not show independent efficacy. In addition, patients would be more likely to participate in such trials because they would be given treatments tailored to be effective against their specific type of cancer. The end results would be lower costs due to testing in patients more likely to respond first, and increased patient participation. Trials that have a primary focus on biomarker development would also provide mate- rial to assess not just target response, but the presence of other molecular aberrations that affect treatment effectiveness, including those that contrib- ute to the development of resistance. In a discussion following Dr. Gray’s presentation, Dr. Roy Herbst from MD Anderson Cancer Center asked Dr. Gray about the role that animal models might play in modeling molecular heterogeneity to enhance multi-

WORKSHOP SUMMARY 35 drug clinical trial designs. Dr. Gray responded that the cell line model is just the first stage in the process, but that animal models can provide informa- tion that cell lines lack. “What our studies do is identify interacting aber- rations that look like they condition response to drugs. But we will never get to the point in vitro where we model all of the nuances of the micro- environment, so the next logical step is to go into the mouse models and complement it there,” he said. He added that the NCI’s Mouse Models of Human Cancers Consortium is developing a robust set of models that are genetically engineered to have many of the same molecular abnormalities linked to cancer progression or response to treatment that are seen in cell lines. (Dr. Anderson also discussed how to model the micro­environment in his presentation, which is summarized below.) Another discussant, patient advocate Kathy Needham, raised the ques- tion of whether cancers should be grouped according to their underlying genetic abnormalities rather than by the type of organ in which they occur when assessing drug effectiveness. Dr. Gray noted that the genetic abnor- malities in ovarian, prostate, and breast cancer are remarkably similar. But he added that the molecular conditioning abnormalities that affect response to treatment differ by organ site and tumor subtype. “So you have to pay attention to both. Clearly people are already pursuing targets, not organ types. But it is by organ site that the drugs get introduced into the clinic,” he said. Dr. Dancey added that although there hasn’t been a test case yet, the NCI has developed new clinical trial designs that enroll patients according to the molecular abnormalities in their tumors and not necessarily by where the tumors appear. Discussant Dr. Steven Shak from Genomic Health then raised the issue that solid tumors often have tens of thousands of mutations, prob- ably many of which are silent or biologically insignificant. But the large number of mutations makes it difficult to discern those mutations that do play a major role in the tumor. “Yes, there are a lot of mutations out there,” Dr. Gray responded, “but they tend not to be recurrent. What we need to do is identify the ones that are recurrently present.” He noted a recent journal article from researchers at Johns Hopkins University in which they catalogued mutations in 13,000 genes in breast and colorectal cancer. The researchers narrowed this list down to a few hundred genes that might be candidates for mutations that play an important role in the progression of these cancers (Sjöblom et al., 2006). See http://emice.nci.nih.gov/emice/MMHCC/mmhcc_organization.

36 IMPROVING THE QUALITY OF CANCER CLINICAL TRIALS Preclinical Model Systems The discussion was followed by a presentation on translation from preclinical model systems to the bedside (and back) in multiple myeloma by Dr. Kenneth Anderson of the Dana-Farber Cancer Institute at the Harvard Medical School. Dr. Anderson developed laboratory models for myeloma that researchers used to predict the effectiveness of several new cancer therapies, most of which are now FDA-approved and widely used in the treatment of myeloma. The thrust of his talk was that researchers can use preclinical modeling to collect the information needed to choose which drugs should be developed and to design clinical trials for those drugs. Although this concept was explored by previous speakers, Dr. Anderson went a step further by showing how best to model the tissue microenviron- ment in which myeloma tumors form so as to gather more clinically relevant information from preclinical studies. This microenvironment determines the expression of the genes that foster myeloma tumors or enable their resistance to treatment. “If one is going to make a preclinical model of cancer that is valid, one needs very strongly to reflect the microenviron- ment,” he said. Because of the recent extraordinary explosion of genetic findings, Dr. Anderson said, myeloma is now classified into seven groups based on the genes expressed in the tumors. Although researchers have detected hundreds of genes that are abnormally expressed in such tumors, studies to systemati- cally assess the effects of overexpression or deletion of these genes reveal a much smaller number of genes believed to play a major role in myeloma. But additional genes that strongly affect survival or metastasis of the tumor, or its resistance to treatment, are only expressed when myeloma cancer cells attach to particular bone marrow cells called stromal cells. Such attachment requires specific adhesion molecules. Some of the genes activated by attach- ment to the bone marrow stromal cells trigger the activity of a complex of proteins in the cells called proteasomes. By breaking down key proteins, proteasomes block normal cell death and enable cancer cells to live for a long time and actively divide. This understanding of the microenvironment of myeloma tumors explains why a proteasome inhibitor drug such as bortezomib is more effec- tive against myeloma cells with the preserved microenvironment of bone marrow stromal cells and adhesion molecules than in cell lines that lack this crucial microenvironment, Dr. Anderson pointed out. The microenviron- ment also explains why conventional myeloma therapies are not effective:

WORKSHOP SUMMARY 37 they are susceptible to cell adhesion-mediated drug resistance. “So testing the drug in the microenvironment is critical,” Dr. Anderson said. Dr. Anderson and his colleagues have developed both in vitro and ani- mal models that mimic the microenvironment of myeloma tumors. In their in vitro models, myeloma cell lines or patient tumor cells are bound to bone marrow stromal cells grown in the laboratory. They also have in vivo mouse models, including a mouse with a transplanted human bone chip in which fluorescent human myeloma cells have been injected. Researchers use this model to test drugs and assess the genes that confer resistance or sensitiv- ity to them, and how well that correlates with what is found from in vitro studies. “It is critical to look and see whether what you have proposed and observed qualitatively in vitro is reflected in vivo,” Dr. Anderson said. He also noted that the interplay between laboratory and clinical stud- ies can be bidirectional. For example, his genomic studies in patients with myeloma revealed a gene, XBP-1, which is overexpressed in all the patients. He used this finding to develop a mouse model in which the mice are genetically engineered to overexpress XBP-1 and have bone destruction and other features similar to that seen in patients with myeloma. “This is a genetic model of multiple myeloma which came from an observation made in patients by the new genomics,” he said. “We always think of bench-to- bedside research, but we can do it the other way around.” Dr. Anderson and his colleagues have used their preclinical models to screen many classes of drugs. They found that some drugs target the tumor and the microenvironment, while others target just one or the other. But whatever the mechanism, a drug must cause tumor cell death, even when the tumor is bound to the bone marrow stromal cells, in order to proceed further in the drug development and testing pathway. These studies led to four highly effective drugs receiving FDA approval in the past 3 years for the treatment of multiple myeloma, as well as several promising experimental drugs currently in clinical trials (Figure 10). Two of the approved drugs, bortezomib and lenalidomide, when used along with a steroid drug or, in the case of bortezomib, a steroid drug and various chemotherapy agents, Bortezomib (Velcade) received accelerated FDA approval as a single agent for relapsed, refractory multiple myeloma in 2003 (see http://www.fda.gov/bbs/topics/NEWS/2003/ NEW00905.html). FDA granted approval in 2006 to lenalidomide (Revlimid) for use in combination with dexamethasone in patients with multiple myeloma who have received one prior therapy (see http://www.fda.gov/cder/Offices/OODP/whatsnew/lenalidomide.htm).

38 IMPROVING THE QUALITY OF CANCER CLINICAL TRIALS Targeting Multiple Targeting Multiple Myloma Cell and Myloma Cell Bone Marrow Milieu IGF-1 inhibitors, CD40 Ab, Bortezomib, NPI0052, 17-AAG, PK11195, Smac Thalidomide, Revlimid, mimetics, Telomestatin, SAHA,Tubacin, CHIR 258, Rad 001 PTK787, Perifosine Bone Marrow stromal cells Targeting Bone Marrow Milieu IKK inhibitors, defibrotide p38MAPK inhibitors FIGURE 10  Novel agents targeting multiple myeloma (MM) cells and/or the bone marrow microenvironment. ACRONYMS: 17-AAG (17-(Allylamino)-17-demethoxygeldanamycin), CD40 ab (antibody to the CD40 integral membrane protein), CHIR258 (Tyrosine Kinase I­nhibitor), IGF-1 (Insulin-like growth factor 1), IKK (conserved helix-loop-helix ubiquitous kinase), NPI0052 (proteasome inhibitor), p38 MAPK (mitogen activated protein kinase 14), PK11195 (peripheral benzodiazepine receptor (PBR) ligand), figure 10 PTK787 (multi-VEGF receptor inhibitor), Rad001 (serine-threonine kinase inhibitor of mTOR), SAHA (Suberoylanilide hydroxamic acid), Smac (Second mitochondria- derived activator of caspase). Source: Anderson presentation (October 4, 2007). each produce remarkable and unprecedented response rates of 80 to 90 percent in newly diagnosed myeloma patients and about a 50 percent complete or near-complete response rate in some studies. Both bortezomib and lenalidomide take advantage of and overcome the growth, survival, and drug resistance potential that is conferred by the microenvironment,

WORKSHOP SUMMARY 39 Dr. Anderson noted. His preclinical models led to the bench-to-bedside development of lenalidomide in just 6 years—about half the typical amount of time needed for such development. Researchers also used Dr. Anderson’s preclinical models to determine the appropriate design of subsequent generations of myeloma drugs. Find- ings on the main proteasome activities that affect tumor cell growth and spread in the microenvironment led to the creation of a new type of pro- teasome inhibitor, NP10052, which inhibits a wider range of proteasome activities than bortezomib. Animal studies showed that about two-thirds of mice treated with NP10052 survived bortezomib-resistant myeloma, whereas all untreated mice died within 100 days. The drug is currently being tested in the clinic. “This drug came from preclinical lab and animal models that showed it was more effective to use broader inhibition of pro- teasome activities,” Dr. Anderson said. Dr. Anderson ended his talk by showing how his preclinical models help researchers discern which drugs to combine and how to test their combinations clinically. For example, these models revealed that proteasome inhibitors interfered with the ability of cultured myeloma cells to repair their DNA. This led the FDA to approve the use of the DNA-damaging agent doxorubicin combined with the proteasome inhibitor bortezomib for the treatment of multiple myeloma. Doxorubicin is not FDA approved as a single agent for myeloma, Dr. Anderson noted, but its combination with bortezomib extended time to progression by about 3 months, and increased the response rate and overall survival, one clinical study found. “This com- bination would not have gone forward if it were not for preclinical model- ing, which showed that this inhibitor of the proteasome has another feature inhibiting DNA repair,” he said. “The explosion in genetics and the ability to study the biology better than we have ever had before allows us to target the tumor directly, but as I hope we have illustrated for you, indirectly as well,” Dr. Anderson concluded. The kinds of genetic studies that have been mentioned and the modeling that I have stressed not only define targets, but also define and inform the design of clinical trials.” This new paradigm targeting the tumor cell in its microenvironment has great promise not only to change the natural history of multiple myeloma, but also to serve as a model for  Approved in May 2007. See http://www.cancer.gov/cancertopics/druginfo/ fda-doxorubicin-HCL-liposome.

Next: Molecular Imaging »
Improving the Quality of Cancer Clinical Trials: Workshop Summary Get This Book
×
Buy Paperback | $46.00 Buy Ebook | $36.99
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

Scientists and clinicians seek a new paradigm that could improve the efficiency, cost-effectiveness, and overall success rate of cancer clinical trials, while maintaining the highest standards of quality. To explore innovative paradigms for cancer clinical trials and other ways to improve their quality, the National Cancer Policy Forum held a workshop, Improving the Quality of Cancer Clinical Trials, in Washington, DC. The main goals of the workshop were to examine new approaches to clinical trial design and execution that would: (1) better inform decisions and plans of those responsible for developing new cancer therapies (2) more rapidly move new diagnostic tests and treatments toward regulatory approval and use in the clinic (3) be less costly than current trials The resulting workshop summary will serve as input to the deliberations of an Institute of Medicine committee that will develop consensus-based recommendations for moving the field of cancer clinical trials forward.

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!