National Cancer Institute NCI Cancer Bulletin: A Trusted Source for Cancer Research News
June 15, 2010 • Volume 7 / Number 12

A Closer Look

Looking at Older Chemotherapy Drugs through a Biological Lens

More often than not, identifying predictive biomarkers—biological aspects of a tumor that predict whether a drug will prove effective—is now a part of modern cancer drug development. This is because most new drugs are developed to target tumors with a specific molecular trait—a de facto biomarker—such as HER2 overexpression in breast cancer or BRAF mutations in melanoma.

But as the research community’s knowledge of the genetic underpinnings of cancer grows, it has become apparent that the efficacy of some older chemotherapy drugs also depends on the genetic makeup of tumors. Identifying biomarkers to help predict which patients will actually benefit from these older drugs would improve treatment planning by sparing patients the toxicity of drugs that would provide no therapeutic benefit.

When you have a drug that’s already in widespread use, it can be very difficult to go back and start doing randomized trials where that drug is withheld.

—Dr. Richard Simon

The ideal study of any predictive biomarker is a randomized trial in which some patients get the drug in question and some don’t and in which the candidate predictive biomarker is measured and defined in advance. Such a trial would allow researchers to understand whether there is any difference in the effectiveness of a treatment depending on the presence or absence of the biomarker. However, “when you have a drug that’s already in widespread use, it can be very difficult to go back and start doing randomized trials where that drug is withheld,” said Dr. Richard Simon, chief of NCI’s Biometric Research Branch in the Division of Cancer Treatment and Diagnosis (DCTD).

Looking Prospectively at Past Trials

To circumvent the near-impossibility of performing a randomized trial in which an accepted drug is withheld, some researchers have turned to a type of study that is sometimes called a “prospective-retrospective” analysis to explore whether specific biomarkers can predict the efficacy of anticancer drugs. In these studies, stored biospecimens from participants in past randomized trials are analyzed for a specific biomarker chosen based on independent experiments before the prospective-retrospective analysis is done.

Importantly, prospective-retrospective studies should not be “fishing expeditions,” in which samples are compared for differences in biomarkers without a prespecified hypothesis, explained Dr. Simon who, along with colleagues Dr. Soonmyung Paik and Dr. Daniel Hayes, recently published recommendations on this subject. Such studies often lead to false-positive results that do not stand up to replication. “You have to do a focused analysis,” said Dr. Simon.

Well-designed prospective-retrospective studies recently showed that patients with colorectal cancer who have mutations in a gene called KRAS will not be helped by drugs such as cetuximab (Erbitux) and panitumumab (Vectibix) that target the protein product of another gene, EGFR.

This type of study design is also being used to take a closer look at older chemotherapy drugs. Last month, Dr. Daniel Sargent, professor of biostatistics and oncology at the Mayo Clinic, and his colleagues published a study looking at a potential predictive biomarker for response to fluorouracil (5-FU), a drug that has served as the backbone of chemotherapy for colorectal cancer for decades.

Their research group had previously used a prospective-retrospective analysis to find evidence that patients with stage II or III colorectal cancer whose tumors have defective DNA mismatch repair (dMMR) mechanisms do not benefit from treatment with 5-FU. Researchers already knew that patients with dMMR have better prognoses than patients with tumors of the same stage and normal mismatch repair mechanisms.

“We had a finding supported by preclinical data that was potentially very important in terms of identifying a cohort of colorectal cancer patients who would not benefit from further treatment after surgery,” said Dr. Sargent. “Because the standard option after surgery for patients at high risk of relapse is 5-FU-based chemotherapy, we felt that we had to generate as much evidence as possible—and the best quality of evidence—to provide a very thorough recommendation in that regard.”

Toward a Molecular Definition of Risk

The researchers established an international collaboration to obtain archived biospecimens from five phase III clinical trials that had compared chemotherapy with 5-FU and either levamisole or leucovorin after surgery with surgery alone for patients with stage II or stage III colorectal cancer. The status of mismatch repair could be determined in 457 patients, 70 of whom had dMMR.

Overall, as in their earlier study, they found no benefit from such chemotherapy in preventing disease recurrence for patients with dMMR, whereas 5-FU-based treatment did benefit patients with normal DNA repair mechanisms. This finding also held for patients with stage II or stage III disease considered separately when the researchers pooled the data from the new analysis with the data from their previous analysis.

Clinical application of these results for patients with stage III disease is somewhat confounded, explained Dr. Sargent, by the fact that current regimens for stage III colorectal cancer, such as FOLFOX, include the drug oxaliplatin. “It’s possible that the addition of oxaliplatin to 5-FU could overcome the resistance that patients with defective mismatch repair have to 5-FU,” he explained.

But for patients with stage II disease, for whom oxaliplatin has not been shown to be beneficial, given the generally excellent prognosis of patients who have defective mismatch repair, “the lack of benefit from 5-FU-based chemotherapy in patients with dMMR tumors indicates that such patients should not receive 5-FU-based adjuvant chemotherapy,” the authors concluded.

“The prognostic importance of dMMR has been demonstrated convincingly in two additional large studies. Coupling this with the current observation of an apparent lack of treatment efficacy in the dMMR stage II patients, it would be reasonable to concur with Sargent et al. and suggest that the absolute benefits of chemotherapy are so marginally small in this good prognostic group that adjuvant chemotherapy should not be considered,” agreed Drs. David Kerr and Rachel Midgley from the University of Oxford in an accompanying editorial.

“For now, dMMR will likely be one of many factors patients and their physicians consider when determining whether to initiate adjuvant chemotherapy for stage II colorectal cancer,” said Dr. Jack Welch, head of Gastrointestinal and Neuroendocrine Cancers Therapeutics in DCTD. “Both assessment of biological risk factors and informed patient decision-making have a place in today’s treatment of stage II colorectal cancer,” said Dr. Welch. “I think eventually we’ll evolve a completely molecular definition of risk in stage II disease, and it might be that there are drugs besides 5-FU that would have some acceptable toxicity and give a reasonable boost to well-defined populations at relatively lower risk,” he concluded.

—Sharon Reynolds