better datasets that are more quickly and cheaply available, as well as better methods. We also need to better engage our patients and manage conflicts of interest because this research is subject to the same kinds of conflict issues that clinical trials are,” she said.
Schottinger noted that Kaiser Permanente has a comparative effectiveness and safety research institute that uses the extensive information collected in EMRs. This health care system has also conducted several randomized controlled trials, demonstrating the importance of integrating palliative care services.
Several speakers emphasized the need for more biomarkers of tumor aggressiveness or responsiveness to make more effective use of cancer drugs. For example, sipuleucel-T (Provenge) is approved for prostate cancer, but it only provides a substantial benefit to a small proportion of prostate cancer patients, and there are currently no biomarkers for response to this drug, which costs $100,000 per patient, Peppercorn said.
McClellan noted that the variable responses cancer patients can have to the same drug suggests that instead of uniformly squeezing down prices of targeted therapies, they be priced differently for various patient subpopulations depending on how much value each of these subpopulations is likely to gain from the treatment. “We need to do a better job of encouraging high-value treatments that are individualized to particular patients,” he said. “Often diagnostic tests that could be highly valuable in influencing a treatment course for a patient are poorly reimbursed or not reimbursed at all. It is a misalignment between what matters most for patients with cancer or patients at risk for developing cancer, and the way that we pay for cancer care that is increasingly an issue as we have an increasingly broad array of treatment options available.”
Schrag emphasized the importance of embedding molecular correlates in cancer clinical trials. “We have to distinguish, when we do our trials, those treatments that provide a little bit of incremental benefit for many people versus those that provide an enormous benefit for a very distinct subset,” she said. She noted that crizotinib only benefits 8 percent of all lung cancer patients. However, Brawley cautioned against having subset analyses being the final word on a treatment’s effectiveness. “We need to be careful in using subset analyses to make clinical decisions. They usually should only be used to justify the next prospective randomized clinical trial,” he said.