BETHESDA, Md-The Oncologic Drugs Advisory Committee (ODAC), citing problems with the study data presented to it, voted not to recommend to the FDA that it approve Neomark (broxuridine for injection, NeoPharm) “for use as a cell proliferation marker to determine the Labeling Index in breast cancer.”
BETHESDA, MdThe Oncologic Drugs Advisory Committee (ODAC), citing problems with the study data presented to it, voted not to recommend to the FDA that it approve Neomark (broxuridine for injection, NeoPharm) for use as a cell proliferation marker to determine the Labeling Index in breast cancer.
The companys request for marketing approval led ODAC members into territory unfamiliar to both the FDA and the panel, the issue of how to forecast the likely outcomes of cancer patients. Committee members expressed concern about the small number of patients, the statistical interpretations of the data, and the value of approving a test whose value to individual patients appeared uncertain.
An intense discussion ensued between committee members and Robert Temple, MD, director of the FDAs Office of Drug Evaluation I, over the value of prognostic tests and what kinds of studies might prove useful in determining their effectiveness.
In the end, committee members generally agreed that prognostic tests shown to be effective by better studies than those presented for Neomark would be valuableand could win approvaleven without providing therapeutic direction to physicians. People who are pursuing this need to know what hurdles they must overcome, Dr. Temple said.
NeoPharm president and chief executive officer William C. Govier, MD, PhD, noted that Neomark is not a therapeutic agent or diagnostic tool, and is not meant to direct a physician to use any specific therapy. Instead, he said, the drug is intended to determine a tumors Labeling Index, which clinical trials reveal is indicative of the probability that a cancer will recur.
Neomark is an important tool to rapidly obtain prognostic information about a breast carcinoma, he said. NeoPharm argued that the Labeling Index is as good or better at predicting recurrence than current prognostic approaches and that the drug is easier to use in determining a Labeling Index than radioactive 3H-thymidine.
Neomark, a thymidine analog, is infused shortly before surgery and incorporated into the DNA of dividing cancer cells. After surgery, the tumor tissue is treated with a monoclonal antibody and reagents, which stain the dividing cells brown. Counting the total number of cancer cells in the tumor specimen and dividing it by the number of dividing cells gives the Labeling Index.
Value of the Labeling Index
NeoPharm presented combined data from two prospective studies in support of its drug: A 207-patient study at the University of California, San Francisco, and a 33-patient study at the State University of New York in Syracuse.
The company argued that the data show that women with a Labeling Index greater than 8 have a 16-fold greater risk of death and a fourfold greater risk of recurrence than women below the 8 cut-off mark. If the Labeling Index is high, our data indicate the cause for alarm is high, Dr. Govier told the panel.
He also argued that the drug is safe, with only three known instances of adverse effectsone case of mild hypertension, one headache, and one incidence of vomiting. Three university-based on-cologists who had participated in the Neomark studies voiced their support for the drug as a prognostic tool.
Nonetheless, the companys presentation left panel members skeptical. Im concerned about drawing strong conclusions from a minimally small data set, remarked Richard M. Simon, DSc, chief of the Biometric Research Branch at the National Cancer Institute.
Concerns were strengthened by the findings of the FDA review team, presented by Karen Johnson, MD. In the San Francisco study, for example, she said that there was not a methodical basis for assessing recurrent disease status in these patients.
FDA reviewers also concluded that only 198 patients had received infused Neomark and that the sponsor had excluded 35 of these from the analysis for various reasons.
Small Study Size
The small size of the Syracuse study, its median follow-up of only 2.3 years, and the exclusion of five patients from analysis does not allow determination of whether the two studies are compatible for merging, as the company did in its presentation, Dr. Johnson said.
Thus, the FDA based its analysis on data from only 163 patients in the San Francisco trial. Reviewers did find a correlation between survival and the Labeling Index as determined by Neomark infusion. However, Dr. Johnson said, the study procedures for assessing relapse-free survival were not sufficiently defined to warrant the use of broxuridine-labeling for prognostication of relapse.
The ODAC Vote
Questions and discussion by the committee clearly indicated that members felt uncomfortable with the Neomark data. This was reflected in their eventual 9 to 0 vote, with one abstention, that the drug did not provide clinically meaningful information for physicians and breast cancer patients, and their vote without dissent against recommending Neomark.
I just have a lot of trouble with the data and extrapolating it to clinical use, said Sandra Swan, MD, medical director of the Comprehensive Breast Center of Greater Washington (DC).
My guess is that they are probably on to something, but our role is not to guess, remarked Derek Raghavan, MD, PhD, chief of solid tumor oncology and investigation therapeutics, Roswell Park Cancer Institute.
Several FDA officials at the hearing seemed particularly concerned with getting guidance from the panel with regard to future trials of prognostic products in cancer patients. Committee members generally seemed to agree with Dr. Raghavan when he said: There are all sorts of clinical and personal decisionsmore personal than clinical onesthat would make such information useful to a patient.
Dr. Temple finally told the panel that, based on the discussion, I am going to understand that most of you think an effective, good prognosticator indicator would be acceptable.