Darbepoetin Alfa: No Benefit in Cancer Patients Not Receiving Chemotherapy and May Decrease Survival

LOS ANGELES—A phase III study evaluating the benefit of darbepoetin alfa (Aranesp) in preventing transfusions in cancer patients not undergoing chemotherapy failed to reach its primary endpoint and demonstrated worse survival in the treatment arm, according to findings presented at the 2007 American Association of Cancer Research (abstract LB-3). When the preliminary results showed a 29% higher mortality rate among patients receiving darbepoetin alfa, Amgen issued a warning to physicians and informed the FDA; the two entities agreed to a "black box" warning about drug safety in this population (see ONI April, page 5).

John Glaspy, MD, professor of medicine, David Geffen School of Medicine, UCLA, who presented the data at the meeting and in a press conference, noted the difference between this study and those showing benefits in patients undergoing chemotherapy. "There are no obvious reasons for this discrepancy," he said.

He emphasized that the finding of a lower observed survival in patients receiving darbepoetin alfa "is of interest to all. A huge amount of data suggests, in the aggregate, that there is not a negative impact on survival in patients who are receiving chemotherapy. Our study was large enough that the findings are in the range where they become statistically significant. For now, we have to conclude that these agents may reduce survival in patients who are not receiving chemotherapy. We should restrict their use to those patients for whom there is a very good reason to give the drug, and not use them routinely."

Large International Trial

The randomized, double-blind, placebo-controlled study, conducted in 21 countries, evaluated darbepoetin alfa 6.75 &mgr;g/kg or placebo given every 4 weeks to 989 patients with active (not in remission) nonmyeloid cancer who were not receiving chemotherapy. Target hemoglobin (Hb) was 12 g/dL. Follow-up occurred at week 19 and at 2 years to evaluate survival.

Did Not Meet Primary Endpoint

The occurrence of all red blood cell (RBC) transfusions from weeks 5 to 17 was higher in the placebo group (215 vs 176 for darbepoetin), but this difference was not significant. "Hence," Dr. Glaspy said, "the study did not meet its primary endpoint." He added that the 391 transfusions given to the 989 study patients highlights the transfusion problem in this patient population.

The incidence of first RBC transfusion in weeks 5 to 17 (or Hb , 8 g/dL) was significantly reduced in the darbepoetin arm, he noted, with a 33.5% risk of transfusion in the placebo arm vs 26.9% in the darbepoetin arm (P = .022).

Patients receiving darbepoetin alfa were significantly more likely to achieve a hemoglobin response (48.4% vs 28.7%); hemoglobin correction (41.6% vs 26.7%); and hematopoietic response (52.2% vs 37.5%). Mean increase in hemoglobin was 0.28 g/dL with placebo vs 0.72 g/dL with darbepoetin (P < .0001).

Interestingly, and perhaps relevant to the findings, Dr. Glaspy noted, the transfusion practice in Europe was different from the rest of the world: 75 patients were not transfused when Hg dropped below 8 g/dL, the transfusion threshold specified in the protocol; the majority of these nontransfused patients were from Central and Eastern Europe. "If the analysis is adjusted to include a transfusion for these patients, the reduction in transfusions in the darbepoetin group is statistically significant," he said.

The total number of deaths during the study was 94 on the placebo arm (20%) and 136 on the darbepoetin arm (26.4%). Deaths on study and during the long-term follow-up were 216 (45.9%) and 250 (48.5%), respectively, for a hazard ratio (HR) of 1.29 for darbepoetin (P = .006).

In a post-hoc analysis adjusting for stratification factors at randomization and sex, stage IV disease, prior chemotherapy use, and prior radiotherapy use, HR was 1.18 (P = .082). Hazard ratio and statistical significance diminished when the analyses were further adjusted for known prognostic factors, including baseline Hb, performance status, tumor type, tumor stage, and baseline FACT-Fatigue score(HR 1.17, P = .11), he added.

Altogether, adverse event rates were similar, reported by 76.4% of placebo patients and 77.5% of darbepoetin alfa patients; treatment-related adverse events were reported by 3.4% and 2.1%; and study discontinuation due to adverse events occurred in 9.1% and 9.9%, respectively. However, cardiovascular and thromboembolic events were more frequent with darbepoetin, 9.7% vs 7.7%.

Dr. Glaspy added that the findings "pose a particular puzzle to cancer researchers" as to the mechanisms underlying any survival risk, and this will be actively investigated.