To the Editor:

Erythropoiesis-stimulating agents are commonly used to treat anemia resulting from malignancy and its treatment. However, these agents are associated with adverse effects that include venous thromboembolic events, tumor progression, and mortality.^1,2 The magnitude of these risks in specific cancer subgroups remains uncertain, and the use of epoetins in oncology has become a controversial practice. We report the results of a randomized clinical trial sponsored by the Ontario Clinical Oncology Group (Hamilton, Ontario, Canada) that closed prematurely. It is important that data from negative trials be published and included in meta-analyses. This report also illustrates how safety concerns arising in one trial can influence accrual in a separate trial in which the same agent is used.

The primary objective of the trial was to evaluate the effect of epoetin alfa on anemia-related quality of life in men with anemia and castration-resistant prostate cancer by comparing Functional Assessment of Cancer Therapy–Anemia (FACT-An) subscale scores at baseline and 16 weeks.³ Change in FACT-An total score (sum of FACT-General subscales and FACT-An subscale), hemoglobin level, transfusion requirements, adverse effects, and overall survival were also evaluated. Men were eligible for this randomized, double-blind, placebo-controlled trial if they had a hemoglobin level ≤ 120 g/L, and progressive metastatic castration-resistant prostate adenocarcinoma. Patients with Eastern Cooperative Oncology Group performance scores greater than 2, known or suspected CNS metastasis, prior use of epoetins, uncontrolled hypertension, other active malignancy, or anemia due to bleeding, hematinic deficiency, or hemolysis; who had had blood transfusions within 14 days; or who were unwilling or unable complete quality-of-life questionnaires were ineligible. Eligible patients were randomly selected via central telephone by the Ontario Clinical Oncology Group—with stratification by the presence or absence of bone metastases and pretreatment hemoglobin level (< 106 v 106 to 120 g/L)—to receive either epoetin alfa (Eprex; Janssen-Ortho Inc, Toronto, Ontario, Canada) starting at 40,000 IU or matching placebo (supplied by Janssen-Ortho Inc) subcutaneously thrice weekly. FACT assessments, hemoglobin level, transfusion use, changes in cancer treatment, and adverse effects were assessed at baseline and every 4 weeks to week 20. The drug dose was increased to 60,000 IU if no hemoglobin response was observed, and the drug was withheld and then restarted at a reduced dose if hemoglobin levels exceeded 140 g/L. The planned sample size of 260 patients provided 80% power to detect a mean improvement in six units over baseline FACT-An score. Changes in FACT-An scores and hemoglobin levels in the epoetin alfa and placebo groups were compared using a two-sample t test. Survival was summarized using Kaplan-Meier plots and Cox proportional hazards modeling. All reported P values are two-sided. Independent ethics approval of the protocol was obtained at each site, and informed consent was obtained in writing from all patients participating in the study.

Between December 2002 and March 2005, 56 men were enrolled at 11 Canadian centers, and all are included in the intention-to-treat analyses. A decision to terminate the trial was made in April 2005 because of low accrual. The main contributors were emergent widespread availability of epoetin alfa therapy through provincial drug plans and third-party payers, and the early closure of a concurrent trial of epoetin alfa in patients with advanced non–small-cell lung cancer conducted by the Ontario Clinical Oncology Group. Both trials underwent unplanned safety analyses because of increased thromboembolism reported in other trials. The lung cancer study was closed due to increased mortality in the patients receiving epoetin alfa.⁴ The prostate cancer treatment trial was interrupted, letter of information revised, and dose-reduction threshold modified to 130 g/L.

At the time of statistical analysis in May 2007, 51 deaths (91%) had occurred, with a median time to death of 35 weeks. All patients are included in analyses. Patients had a median age of 71 years (range, 50 to 87 years), and 57% had a hemoglobin level ≤ 105 g/L (Table 1). Median FACT-An scores were slightly better in the epoetin alfa group and fewer patients had symptoms, but no other imbalances in baseline characteristics were apparent.

There was a trend to improved FACT-An scores at 16 weeks, compared with baseline in patients receiving epoetin alfa therapy, but this did not reach statistical significance (Table 2). A statistically significant improvement in hemoglobin levels was observed in patients receiving epoetin alfa therapy compared with those receiving the placebo (median change at 16 weeks, 16.8 g/L v 2.9 g/L; P < .05). A trend to reduced need for transfusions was also observed in the epoetin alfa group compared with the placebo group (mean number of units transfused by 16 weeks, 0.6 v 1.2; P = .066). All 27 deaths in the placebo group were attributed to prostate cancer; 21 of 24 deaths in the epoetin alfa group were attributed to prostate cancer, and three deaths were attributed to other causes (pneumonia, renal failure, and natural causes). Most adverse events were considered to be results of prostate cancer progression; nine adverse events in three patients were considered to be related to study therapy, and all three patients were receiving the placebo. Two patients experienced venous thromboembolic events, and both were receiving the placebo. Median overall survival was 29 weeks with the placebo versus 49 weeks with epoetin alfa (hazard ratio, 0.58; 95% CI, 0.33 to 1.05; Fig 1).

This trial confirms the ability of epoetin alfa to increase hemoglobin levels and reduce transfusion requirements in this population of men with castration-refractory prostate cancer and anemia. Beneficial effects on health-related quality of life, and detrimental effects including venous thromboembolism and mortality, were not confirmed. However, the small number of patients studied should provide little confidence that these outcomes were not affected. The rate of venous thromboembolic events in this population was low (3.6%), and all such events occurred in patients receiving placebo therapy. Cancer progression was not specifically evaluated in this trial. The epoetins provide an excellent example of the value of randomized trials—and meta-analyses of these—for the detection of subtle and unexpected adverse effects of new therapies in cancer patients. Our trial also provides an example of how the early adoption of apparently innocuous treatment, and safety concerns arising from other trials, can undermine the successful completion of ongoing clinical trials.

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

The author(s) indicated no potential conflicts of interest.