the quality and efficiency of execution, something that adds to the cost and is holding this field down,” Dr. Woodcock said. She mentioned that the FDA plans to form a public–private partnership to drive standardization of study execution.

Dr. Woodcock ended her talk by noting that the purpose of trials is to develop the evidence that products will save lives and improve health, as well as provide access to investigational products. Quality trials within this context mean those that meet the needs of patients and health-care providers, who are the “ultimate customers,” she said. “Regulatory agencies are supposed to be surrogates for what the patients and the providers actually want, and are not supposed to stand in the way,” Dr. Woodcock said. She urged the conferee to consider not only scientific problems to be addressed in improving the quality of cancer clinical trials, but also to think about whether the trials are meeting the needs of patients and providers—whether they are saving lives and improving the health of the population.

In the discussion following the presentations, Dr. Sullivan pointed out that the imaging agent industry believes radiotracers used in imaging studies have a risk/benefit profile similar to devices and should be regulated accordingly. Dr. Woodcock agreed that it was important for the FDA to consider this as a way to streamline the approval of such tracers. Later during the discussion, Dr. Parkinson noted that overcoming regulatory hurdles and gaining FDA approval for a drug or diagnostic does not guarantee insurance reimbursement for the product, especially when the product is used abroad. “A common problem globally is this disconnect between regulatory approval and reimbursement,” he said.

REPORTS FROM THE CASE STUDY DISCUSSION GROUPS

Adaptive Trial Design

Dr. Herbst presented a summary of his group’s breakout session on adaptive trial design. His group agreed on a number of issues, including the variability of adaptive trial designs and the notion that adaptive trials are not a new phenomenon—for a long time, many investigators have been using a few of these adaptive designs in their clinical trials, most notably sequential monitoring with early stopping boundaries. But what is new is the creation and expanded use of other forms of adaptive designs, such as sample size reestimation, adaptive dropping and adding of trial arms, as well as the other types delineated by Dr. Donald Berry in his presentation



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 89
 WORKSHOP SUMMARY the quality and efficiency of execution, something that adds to the cost and is holding this field down,” Dr. Woodcock said. She mentioned that the FDA plans to form a public–private partnership to drive standardization of study execution. Dr. Woodcock ended her talk by noting that the purpose of trials is to develop the evidence that products will save lives and improve health, as well as provide access to investigational products. Quality trials within this context mean those that meet the needs of patients and health-care provid- ers, who are the “ultimate customers,” she said. “Regulatory agencies are supposed to be surrogates for what the patients and the providers actually want, and are not supposed to stand in the way,” Dr. Woodcock said. She urged the conferee to consider not only scientific problems to be addressed in improving the quality of cancer clinical trials, but also to think about whether the trials are meeting the needs of patients and providers—whether they are saving lives and improving the health of the population. In the discussion following the presentations, Dr. Sullivan pointed out that the imaging agent industry believes radiotracers used in imaging studies have a risk/benefit profile similar to devices and should be regulated accord- ingly. Dr. Woodcock agreed that it was important for the FDA to consider this as a way to streamline the approval of such tracers. Later during the discussion, Dr. Parkinson noted that overcoming regulatory hurdles and gaining FDA approval for a drug or diagnostic does not guarantee insur- ance reimbursement for the product, especially when the product is used abroad. “A common problem globally is this disconnect between regulatory approval and reimbursement,” he said. REPORTS FROM THE CASE STUDY DISCUSSION GROUPS Adaptive Trial Design Dr. Herbst presented a summary of his group’s breakout session on adaptive trial design. His group agreed on a number of issues, including the variability of adaptive trial designs and the notion that adaptive trials are not a new phenomenon—for a long time, many investigators have been using a few of these adaptive designs in their clinical trials, most notably sequential monitoring with early stopping boundaries. But what is new is the creation and expanded use of other forms of adaptive designs, such as sample size reestimation, adaptive dropping and adding of trial arms, as well as the other types delineated by Dr. Donald Berry in his presentation

OCR for page 89
0 IMPROVING THE QUALITY OF CANCER CLINICAL TRIALS earlier in the day. The group agreed that although Bayesian methods are particularly well suited for building adaptive designs, traditional methods can also be adaptive. The group also agreed that it can be important to monitor relationships among early and late endpoints in adaptive studies, and to calculate operating characteristics, such as sample size distribution and frequency of incorrect conclusions. Such calculations usually require simulation. Dr. Herbst stressed the importance of noting that although adaptive designs can sometimes result in smaller trials, that is not always the case. Hopefully, adaptive trials provide more accurate conclusions, he said. Additional possible benefits of such designs are that more questions can be considered in a single trial; they can foster faster, more efficient drug development; they can lower costs of medical care; and they can lead to better treatment of participants. Because of its flexibility, another advantage of adaptive trials is the ability to work additional study parameters into an ongoing study as more knowledge is gained. For example, data collection on the predictive value of a newly discovered biomarker can be added into an ongoing trial of the predictive value of other biomarkers. Dr. Herbst also described the challenges involved in conducting adap- tive trials, such as additional logistical infrastructure needs. Adaptive designs must be prospective, which requires extra work before trials begin. Extra costs and potential delays in setting up adaptive trials may arise due to the need for increased communication with government regulators and local oversight boards, and the potential need for a larger drug supply. Data-flow needs increase the logistical burden, and it can be challenging to prepare for different alternatives—for managing a greater variety of treatment regimens that may change over time within a single trial. During the discussion, Dr. Richard Chappell pointed out the potential bias of response-adaptive randomization trials due to changes in patient characteristics during the accrual that can confound treatment effect with time effects. Quoting Peto (1985), he noted, for example, that patients entering the European Coronary Bypass Trial showed a statistically signifi- cant trend toward better prognosis at baseline as accrual continued. If there had also been a trend in allocation proportions toward the apparently better treatment, an appreciable bias might have been engendered. Although mod- els can be designed to address these issues, models tend to be subjective, based on sparse trend information, and, even when used, trials may be left with ambiguous and multiple adjusted answers—difficulties that should be prevented by randomization, Dr. Chappell stressed. A stratified group-

OCR for page 89
 WORKSHOP SUMMARY sequential approach (Karrison et al., 2003) can address this bias, he added. Although he recognized that results from response-adaptive randomized Phase II studies may be superior to those produced from nonrandomized designs, Dr. Chappell still thought fears of bias from confounding with time trends warrant not using response-adaptive randomization in confirmatory trials. During his summary, Dr. Herbst reiterated this concern about potential bias in response-adaptive randomized trials, but added that “although this needs to be looked at in any trial and analyzed and is a real concern, the belief was that we could go forward despite this.” He also expressed concern about the potential bias being introduced by revealing interim trial results to investigators, patients, and the investment community, as Dr. Ellenberg dis- cussed in her presentation. (See page 16.) But all agreed that this potential bias could be addressed, although it may limit the use of adaptive methods in some applications. Group members agreed that adaptive methods should be considered in designing clinical trials because they offer some benefit, although their use still presents challenges. Phase 0 Trials Dr. Giulio Draetta presented a summary of his group’s discussion of Phase 0/exploratory IND trials. This group discussed the advantages, dis- advantages, ethics, and costs of Phase 0 trials. The group agreed that Phase 0 trials were especially useful in doing compound triage when multiple compounds are being considered because they can indicate the compound with the most favorable pharmacokinetics or pharmacodynamics. But as Dr. Schilsky pointed out, generally only the large pharmaceutical companies, as opposed to academic investigators or researchers at small biotechnology companies, simultaneously have available several compounds that target the same pathway or disease. He also noted that Phase 0 trials add an extra step in the testing process for individual compounds. “At the end of the day, if you’re going to bring your drug through the full clinical development plan, you’re still going to need the full package of data. This is not a shortcut,” he said. But Dr. Collins pointed out the time savings by running Phase 0 and animal studies simultaneously. Dr. Schilsky expressed additional reserva- tions that the dose ranges in a Phase 0 study are more limited than those in a Phase I study and therefore may not reveal as much. He added that Phase 0 studies may not be appropriate for a drug with multiple targets or with an unknown target. They might also be unsuitable for a drug that is metabo-

OCR for page 89
 IMPROVING THE QUALITY OF CANCER CLINICAL TRIALS lized very differently in people than in animals, or for a drug that affects patients in a way that is substantially different from the effects observed in blood samples or other surrogate tissues used in the Phase 0 study. In addition to aiding compound triage, another advantage of Phase 0 trials pointed out by several discussants is that it can reveal the clinical pharmacodynamics of a compound early in the drug development process, so that compounds that appear worthwhile in preclinical studies—but are clinically irrelevant—are essentially thrown out early in the process. The group agreed that Phase 0 studies have unquestioned benefits in addressing early biological endpoints in patients, and impacts on target and potential downstream biology. But such studies require extensive and time-consuming efforts to develop and validate pharmacodynamic assays in animals and then in human tissues. Dr. Draetta said the group believed such effort is worthwhile, noting that, “You need to spend the time to know what you are doing. You cannot wait until Phase II to know whether you are hitting the tumor target.” Given the large time and expense that may be involved in developing and validating a pharmacodynamic or biomarker assay for a Phase 0 trial, there was some discussion over who would be willing to pay for such assay development. Dr. Schilsky suggested the NIH Clinical Center provide assay development and transfer the technology to interested parties. Dr. Doroshow agreed it would be a wise use of the Center’s resources and that this has already occurred for some assays, which are now publicly available. For academic researchers without drug sponsors, acquiring the resources to run a Phase 0 trial is also a problem because these studies are essentially being conducted outside the realm of routine medical care and therefore are not likely to be reimbursed by insurers, Dr. Schilsky pointed out. The group also discussed the ethical implications of a Phase 0 trial and noted that they are equivalent to what is seen in normal volunteer studies. But healthy volunteers are often paid to participate in clinical trials from which they personally receive no benefits, so Donna Przepiorka of the FDA raised the question of whether volunteers in Phase 0 trials should be paid as well. Dr. Mills noted that his long-term experience with imaging stud- ies in volunteers indicates their broad acceptance of the concept that their participation will not benefit them personally. Dr. Darlene Rosario of Mannkind Corporation added that even many patients participating in placebo-controlled randomized studies do not receive a personal benefit from their participation. All agreed that informed consent was essential in Phase 0 trials since they are for other clinical trials,

OCR for page 89
 WORKSHOP SUMMARY and that volunteers in Phase 0 trials should be made aware that they are not likely to personally gain anything from their participation. Dr. Doroshow pointed out that when the IRB reviewed his study of PARP inhibition in tumor biopsies, the IRB noted that because of the extensive preclinical work done by the investigators, the data generated by their tumor analyses would be more accurate than data typically garnered from a Phase I study. The low doses used in the study also meant that the risks would be lower to par- ticipating patients. As a result, the IRB considered the overall risk/benefit ratio to be lower than what might be seen in a typical Phase I trial, and the study was considered ethically sound as long as patients were made aware in writing that they personally may not benefit from the study. Katherine Meade, a patient advocate who volunteers for Us TOO International Pros- tate Cancer Education and Support Network, stressed the importance of involving patient advocates early in the process to gain feedback on the trial design and communication to volunteers. Imaging Drs. Hricak and Piwnica-Worms summarized their group’s discussion on imaging. Dr. Piwnica-Worms noted that imaging is used most often in preclinical or Phase I studies to determine if a drug is hitting its target (pharmacodynamics), to confirm the drug’s mechanism of action, to evalu- ate the clinical response, and to do pharmacokinetic analyses when labeled drugs are used as the imaged agent. The group agreed that there were opportunities to use imaging in Phase II or III trials to optimize patient selection and conduct treatment follow-up. “We have tools that we are not using,” said Hricak. The group spent some time discussing the differences between targeted diagnostic imaging and imaging of therapeutic response. Diagnostic imag- ing often involves imaging a single target at a single time point. Given that it can take between 100–200 million dollars to develop such a targeted diagnostic imaging agent (Nunn, 2006), which has a very limited use, this application of imaging may be too expensive to develop relative to the size of the market that would use it, Dr. Piwnica-Worms said. The group agreed that imaging targets that can be generalized to several different tumor types, such as imaging of apoptosis, proliferation, and tumor metabolism, are more broadly useful than tumor-type-specific targets. Discussants Drs. Tim McCarthy from Pfizer and Jeff Evelhoch from Amgen pointed out that imaging has and continues to have a growing

OCR for page 89
 IMPROVING THE QUALITY OF CANCER CLINICAL TRIALS impact on drug development. Pfizer has used PET assessment of whether a drug is reaching its target to make decisions about whether to stop a trial. They also used FDG and FLT PET imaging and DCE MRI measures to confirm mechanisms of action for novel therapeutics and to help make “go/no go” decisions regarding further drug development. Dr. Evelhoch noted that PET FDG measures of tumor metabolism demonstrated a metabolic effect even in the absence of a traditional clinical response, and enabled some of Amgen’s drugs to progress further in the clinical trial hier- archy than they would have if only standard radiologic measures of tumor response were used. CT volume imaging of tumors was also noted to be more accurate than RECIST. “This is just low-hanging fruit of today’s technology. It can have a significant impact on how you interpret the data, and is very accessible, straightforward, and can be executed today,” said Dr. Piwnica-Worms. By “killing failures early and fast,” imaging saves substantial drug development costs, Drs. McCarthy and Evelhoch noted. Some savings are due to avoiding Phase II or III trials of drugs likely to fail or repeated Phase II failures, while others were due to shortening the drug development time line. Even a delay of a few months can mean millions of dollars lost. Despite the recent explosion in U.S. clinical PET centers, as indi- cated by a slide Dr. Piwnica-Worms showed (Figure 19), there is a lack of FIGURE 19 Clinical PET centers in the United States in 2007. SOURCE: Piwnica-Worms presentation (October 4, 2007), reprinted, with permis- sion, from AMI Winter 00 News, 2007. Copyright 2007 by Academy of Molecular Imaging. figure 19

OCR for page 89
 WORKSHOP SUMMARY standardization, harmonization, and training for these and other imaging centers that can create inaccuracies in multisite clinical trials, the group pointed out. They suggested NCI-designated Comprehensive Cancer Cen- ters should have imaging and image analysis core laboratories for clinical trials. These core labs would be especially useful in the conduct of Phase I trials and would save money in the future by improving the quality of data, Dr. Piwnica-Worms said. Key to these core labs would be research protocol assistants, who ensure proper protocol execution, and Image Response Assessment Teams (IRATs), which ensure consistent interpretation of imaging. IRATs “are not widely distributed, but their value is integral, and with a modest investment could be integrated into the overall process and probably increase quality control substantially,” Dr. Piwnica-Worms said. For multisite trials, uniform central reading of images is key to forging an efficient path forward, the group agreed. The group also suggested that academic imagers (radiologists) be engaged at the starting point of the trial design process to produce better trial results. The group also discussed validation of imaging biomarkers. Such vali- dation requires precisely defining the question the biomarker is expected to answer, and using positive and negative controls in the protocol design, Dr. Piwnica-Worms said. The group requested that the FDA establish a path- way for qualifying imaging biomarkers akin to that used to qualify FDG PET—one that is mechanism-based and not limited to use on a specific organ or a similarly narrowed application. The group also suggested that the validation process for tracers used as research tools be differentiated from that needed for imaging agents used for clinical diagnostic purposes. Although some imaging technology, such as PET, is expensive, it can offer information that saves patients from undergoing surgeries or other invasive procedures, which are even more expensive, Dr. Piwnica-Worms explained. “PET was approved as a cheap alternative for surgery in the work-up of single pulmonary nodules 8 years ago,” he added. “The insur- ance companies loved it. They liked the chance of spending $2,000 to have a one-in-four chance of avoiding a $25,000 surgery. Can that same kind of logic optimize the use of more expensive bioimaging tests as long as the information they provide has value?” Other suggestions made by the group included that genomic and proteomic correlation studies in image-guided biopsies be standardized, and that researchers establish how serum, urine, and tumor biomarkers complement the information gained with imaging biomarkers. The group also suggested changing the culture of the imaging community so they are

OCR for page 89
 IMPROVING THE QUALITY OF CANCER CLINICAL TRIALS more inclined to participate in clinical research. This could be done by developing practice environments that encourage and reward imagers to engage in research. Dr. Piwnica-Worms concluded by reiterating the caveats of imaging that were described by other speakers. These caveats include that imaging may demonstrate that a target is being hit or confirm the expected mecha- nism of action, but it alone does not imply clinical benefit. He acknowl- edged the regulatory and financial barriers linked to imaging biomarker validation. These barriers hinder sponsors from running clinical trials in this country. Many imaging trials are moving overseas, he noted, and this poses a threat to the U.S. trial infrastructure. Use of Proteomics/Genomics to Assign Therapy in Lung Cancer Dr. Mendelsohn summarized his group’s discussion on the use of proteomics and genomics to assign therapy in lung cancer. This group’s discussion was focused on the presentations by Drs. David Carbone from Vanderbilt University and Mark Kris and William Pao of MSKCC. Dr. Carbone used the MALDI (matrix assisted laser desorption/ionization) mass spectrometry system to detect protein signatures associated with lon- ger survival in 139 advanced lung cancer patients following treatment with tyrosine kinase inhibitors gefitinib or erlotinib. In this retrospective study, he found the elevated production of eight key proteins in blood serum linked to longer survival. A second retrospective study in a different group of lung cancer patients found that the eight proteins did not correlate with longer survival in patients treated with standard chemotherapy, or surgery and radiation. This suggests that the eight-protein signature specifically predicts longer survival following treatment with a tyrosine kinase inhibi- tor and not merely in those patients likely to survive longer no matter what treatment they receive. He plans to do a prospective study on the usefulness of the protein signature in predicting lung cancer patients who will respond best to tyrosine kinase inhibitor treatments. Dr. Carbone also has used two- dimensional gel electrophoresis to find a more complex protein pattern, encompassing more than 1,000 proteins, that is present in the lung biopsies of cancer patients, but not in normal lung tissue biopsies. He is currently looking for candidate diagnostic markers among these proteins. Drs. Kris and Pao reported that EGFR is overexpressed in 45 percent of non-small cell lung cancers (NSCLCs), as measured by immunohisto- chemistry. Studies done by Dr. Pao and others (Lynch et al., 2004; Paez et

OCR for page 89
 WORKSHOP SUMMARY TABLE 4 Factors Predicting Sensitivity to Gefitinib (Iressa) Overall Response Rate 11% Women 18% Men 5% Never Smokers 29% Current/Former Smokers 5% Adenocarcinoma 12% Other Non-Small Cell Lung Cancer 7% SOURCE: Kris/Pao presentation (October 4, 2007). al., 2004; and Pao et al., 2004) detected four EGFR mutations associated with sensitivity to gefitinib or erlotinib, and indicated that the KRAS muta- tion predicts a lack of response to the same drugs. Seventy-five percent of lung cancer patients with these EGFR mutations responded to tyrosine kinase inhibitors, and only 1 percent of those with the KRAS mutations responded. By comparison, clinical predictors are not as informative in predicting the likelihood of response. Nonsmoker patients, for example, have the highest response rate, but only 30 percent of nonsmokers with NSCLC respond to gefitinib or erlotinib (Table 4). There are also molecular predictors for acquired resistance to tyrosine kinase inhibitor treatment, including second-site EGFR mutations in about 50 percent of the cases, and MET amplification in about 20 percent of patients. This suggests that MET inhibitor drugs may have a role in treating patients with acquired resistance to erlotinib or gefitinib. MSKCC is starting to screen lung cancer patients for KRAS or EGFR mutations and using this information, which can be acquired from the tissue removed in a needle biopsy, to decide which patients to treat with tyrosine kinase inhibitors. In conjunction with Dr. Varmus, Drs. Kris and Pao also developed a transgenic mouse lung cancer model, with the same EGFR mutations that cause lung cancer in humans, to screen for more effective tyrosine kinase inhibitors. “This is a nice example of a mouse model using transgenic technology in order to do preclinical studies,” Dr. Mendelsohn said. The cost of doing tumor biopsies and genomic and proteomic tests on the biopsied tissues was discussed. Drs. Kris and Pao noted that in their studies, this cost was about $5,000 per lung cancer patient. The cost was covered through grants or philanthropy and not by third-party payers. The New York researchers continue to do studies aimed at confirming the useful-

OCR for page 89
 IMPROVING THE QUALITY OF CANCER CLINICAL TRIALS ness of these predictive markers. “We still have work to do to convince the [insurance] companies that this is something that they should foot the bill for,” Dr. Mendelsohn pointed out. Dr. Kris stressed the need for tumor biopsies and molecular analyses on such biopsies in clinical trials, pointing out that EGFR and KRAS mutations play a critical role in one-fifth of lung adenocarcinomas. But Dr. Parkinson noted that patient subsets are a major problem for drug devel- opers, especially because so much redefining of those subsets is done each year. Discussant Dr. Sam Hanash of the Fred Hutchinson Cancer Center added that gene defects may not be so definitive, with the many overlap- ping and changing molecular pathways to cancer. One can target a pathway, he noted, but at some point it might not be the critical path. Dr. Carbone raised the question of when a biomarker is good enough to be predictive and prognostic, and said how good a biomarker is depends on what other alternatives there are, and how bad the outcome may be if a biomarker is not used. He said it might be appropriate to take risks when there are bad outcomes and poor alternatives. There also was some discussion on which regulatory pathways cancer biomarkers should follow to enter the clinical market. The lengthy FDA approval process is not necessarily required for some predictive biomarker tests performed in laboratories, for which only the laboratory is subject to CMS scrutiny under CLIA. The costs of such tests are likely to be reim- bursed if there is enough evidence for their usefulness in compendia or other published data, Dr. Mendelsohn noted. “It may be more important to have good data than to worry about all the criteria that would require the FDA to actually put this on the drug sheet in the package saying that this test is certified for this particular treatment,” he said. Dr. Mendelsohn summarized the discussion by saying, “We are finally taking what has been experimental and moving it into the clinic, and get- ting to the point of proof of principle. But it is painfully hard to prove the principle. These are 5-year projects, and they are still very much investiga- tive projects rather than in-clinical-practice projects.” Use of Genetics/Genomics to Assign Therapy Dr. Pierre Massion summarized his group’s discussion on the use of genetics/genomics to assign therapy. At this discussion, Drs. Sparano, Shak, Kucherlapati, and Chang presented their study results, which suggest that at least for breast, lung, and colon cancer, genetic or genomic tests for predict-

OCR for page 89
 WORKSHOP SUMMARY ing treatment response or prognosis are showing evidence that they could be useful for reducing the overtreatment of patients that is so common. “These still are primarily investigator-initiated trials and validation is on the way, but I think we have got strong evidence that this will be beneficial,” Dr. Massion said. Drs. Joseph Sparano of Albert Einstein College of Medicine and Steven Shak of Genomic Health noted that patients with ER-positive, lymph node–negative breast cancer comprise about half of all newly diagnosed breast cancers. More than three-quarters of these patients can be adequately treated with surgery and hormonal therapy, with or without radiation. Add- ing chemotherapy to these women’s treatment regimens provides an absolute benefit of only 5 percent or less, while adding significant toxicity. To better assess that risk of recurrence and aid the decision of whether to add chemo- therapy to the treatment, the Oncotype DX 21-gene test was developed. This test, which is already on the market, has been studied in patients treated with the hormonal therapy tamoxifen as well as in patients treated with both chemotherapy and hormonal therapy, and in patients who received no therapy. These studies show that the greater the “recurrence score” in the Oncotype DX test, the greater the likelihood of recurrence and death from breast cancer within 10 years. With the NCI-sponsored TAILORx study, researchers at 900 sites will be assessing prospectively in 4,000 women the benefit of adjuvant chemotherapy for those women with an intermediate Oncotype DX recurrence score. The women all have ER-positive, lymph node–negative, HER2/neu-negative breast cancers. Dr. Raju Kucherlapati from Harvard University described his much smaller iTarget trial that assessed the usefulness of EGFR mutation as a bio- marker test to predict response to gefitinib in first-line therapy of advanced NSCLC. “The results are quite impressive,” Dr. Massion said. “They move an overall 20 percent response rate for all NSCLCs, which is probably an optimistic estimate, to a 75 percent response rate for those patients with EGFR-mutated positive tumors.” The cost of the mutation analysis was absorbed by Harvard and not reimbursed by third-party payers, he added. Dr. David Chang reported the results of his prospective analysis of more than 400 archival tissues, which indicated that a wild-type KRAS in colon cancers predicted prolonged survival following treatment with an EGFR antibody, whereas patients with tumors that had KRAS mutations did not benefit at all from that therapy. “Genomic biomarkers have evolved in clinical decision making,” Dr. Massion summed up. “They are not on the horizon anymore but rather

OCR for page 89
00 IMPROVING THE QUALITY OF CANCER CLINICAL TRIALS right in front of us.” He stressed the need to continue to systematically discover and select across biological materials and molecules, new diagnostic biomarkers for cancer, including those that indicate posttranslational modi- fications, or other new classes of genetic biomarkers. Dr. Massion noted that the newly launched Cancer Genome Atlas project,21 which is sponsored by the NCI and the National Human Genome Research Institute, aims to find new classes of tumors and will focus on biomarker discovery and validation of those that are specifically related to the function of a target of interest. This project will involve comprehensive molecular analyses of a large series of tumor specimens at multiple sites. Lung, brain, and ovarian cancer will be the first three cancers that will be studied in the pilot phase. The goal is to test the feasibility of using large-scale genome analysis technologies to determine all of the important genomic changes involved in cancer. The group also recommended carefully accessing existing cohorts that are valuable resources for testing biomarkers. An example of such a resource is the blood samples from the Women’s Health Initiative study. “These studies need to be accessed for testing the performance of prevalidated biomarkers, and we also need guidelines as to how we can access them in a rational way,” Dr. Massion said. A major point raised in the discussion was the lack of funding to manage these clinical trial tissue repositories, which could be so useful for the discovery and validation of biomarkers. The group suggested finding a new funding mechanism to support the management of these tissue repositories. The group also discussed how to move new diagnostic tests and treat- ments more rapidly into clinical practice. Suggestions included incorporat- ing biomarker tests early in drug development, even at the level of preclini- cal models, Dr. Massion said. He also noted the need for both professional societies and regulatory agencies to establish guidelines for the evaluation and standardization of specific biomarkers, as well as for the standardiza- tion of the clinical elements associated with the biomarkers. “The biologi- cal nature of the study, the stage of the study, the performance of the tests, the mode of action, the stage validation, and the technologies—all these need to be clearly standardized, and guidelines would help the field move forward,” Dr. Massion said. Such standardization and guidelines, as well as incorporating biomarker tests early in the development process, should help to reduce the costs of biomarker development, provide stronger evidence for improved outcomes, and thus make biomarker test reimbursement by third- 21See http://cancergenome.nih.gov/about/index.asp.