Making Cancer Drugs Worth the Cost

Christin Melton

April 2011, Vol 2, No 2 - HOPA Conference


Salt Lake City, UT—When it comes to cancer, “how much is too much?” is a common question. At the John G. Kuhn Keynote Lecture at the 2011 Hematology/Oncology Pharmacy Association annual meeting titled “The Cost of Cancer Therapy,” speaker Tito Fojo, MD, PhD, with the National Cancer Institute (NCI), said cost is not the real issue. “If you frame the argument in terms of cost, you’re going to lose it,” he said. Dr Fojo said the real issue is “about how effective drugs are and how toxic drugs are.”

Dr Fojo offered examples of drugs used in clinical situations that not only fail to improve survival but actually harm patients, such as cetuximab (Erbitux) for patients with metastatic colorectal cancer (CRC) and bevacizumab (Avastin) for patients with breast cancer. He acknowledged that controversy surrounds these examples and not everyone shares his views, including Giuseppe Giaccone, MD, PhD, his supervisor at NCI and a member of the American Society of Clinical Oncology (ASCO) panel responsible for adding the combination of cetuximab and paclitaxel (Taxol) to the ASCO guidelines for first-line therapy of metastatic CRC. These therapies are also included in the National Comprehensive Cancer Network treatment guidelines.

The Unmet Promise of Personalized Medicine
Dr Fojo’s review of the pivotal trials used to support the US Food and Drug Administration (FDA) approval of cetuximab in metastatic CRC and bevacizumab in breast cancer underscores important considerations for various stakeholders when considering drug development and use. “Our drug development, drug approval, and drug consumption strategies need better focus,” he said, lamenting that despite the decade that has elapsed since imatinib (Gleevec) ushered in the “targeted therapy revolution,” too many treatments are still used indiscriminately in broad patient populations.

Dr Fojo sees room for improvement at all levels—academia, the pharmaceutical industry, government agencies such as the FDA, and clinicians treating patients with cancer. All are responsible for “our failure to deliver on the promise of personalized medicine.”

The example of cetuximab supports Dr Fojo’s contention that drugs must be analyzed prospectively to identify predictive markers before initiating trials. Four years after cetuximab’s 2004 FDA approval, a study published in the New England Journal of Medicine showed it was ineffective in patients with mutated KRAS. In 2009, ASCO issued its Provisional Clinical Opinion recommending KRAS testing before prescribing cetuximab, but left it as a voluntary option.

Cetuximab inhibits epidermal growth factor receptor (EGFR), and “as early as 1990, Bert Vogelstein at Hopkins had reported that as many as 50% of colorectal patients harbored KRAS,” Dr Fojo said. In 1998, KRAS was reported to be in the EGFR signaling pathway. “Despite this, we went ahead and conducted studies without looking at this prospectively.”

Continued investigation into the causes of poor response to cetuximab among some patients with wild-type KRAS will likely reduce the viable patient population even further. “Eventually, we’ll be able to identify the 80% of patients that do not benefit, and give it only to the 20% who benefit substantially, and a drug that was marginal to begin with will become highly effective,” said Dr Fojo.

What Constitutes an Acceptable Improvement?
“Increasingly, therapies that demonstrate at best marginal improvement are being used in the treatment of cancer,” he noted. With breast cancer, we may even be going backward, with the controversy over bevacizumab. He emphasized that his views did not represent those of the NCI.

Dr Fojo faulted the decision to approve bevacizumab based on data from the E2100 study, which suggested that adding it to paclitaxel significantly prolonged progression-free survival (PFS) but not overall survival (OS). Although the study showed nonsignificant improvement in OS favoring bevacizumab, “Nonsignificant improvement is no improvement,” he said, adding that it was unusual for a study to reflect major discordance between PFS and OS.

Dr Fojo and colleague published a review of major studies published since 1996 with a significant improvement in PFS or OS and found “tight correlation between the two” (Lancet Oncol. 2010;11:1117-1119). He therefore believes E2100 is an outlier.

Even if bevacizumab does improve PFS, “delaying symptoms is a worthwhile goal, but delaying PFS does not equal delaying symptoms,” he said.

Toxicities
Studies need to focus more on whether regimens used are harming patients. In E2100, patients treated with bevacizumab had more than twice as many adverse events as the control group. Dr Fojo believes that toxicity is unacceptable when a treatment fails to improve OS and is very expensive, but he acknowledged that many breast cancer specialists disagree. The FLEX trial, investigating cetuximab in metastatic CRC, also found high rates of grade 3 and 4 toxicities, particularly severe skin reactions.

With the growing use of oral therapies for cancer, it is important that studies not overlook the grade 1 and 2 toxicities, he said. “The scale was developed for chemotherapy, but when you are taking a drug every day for months on end, a grade 1 or 2 event can be very difficult.”

Although cost is relevant for drugs that cause serious adverse effects but demonstrate minimal efficacy, Dr Fojo said it comes down to the notion that you do not harm patients for little improvement in survival. Various attempts to combine therapies have hurt patients, without helping them; he called for an end to this “incremental approach” to treatment.

Finding better treatments requires determining the settings in which the drugs work best, but also the ones in which they do not work. This would require major journals to give greater weight to negative studies, he said, which are often relegated to secondand third-tier journals. It also requires expanding incentives for pharmaceutical companies to take a more personalized approach to drug development, where rewards are likely to be less lucrative