Venous Thromboembolism and Mortality Associated With Recombinant Erythropoietin and Darbepoetin Administration for the Treatment of Cancer-Associated Anemia FREE

The erythropoiesis-stimulating agents (ESAs), erythropoietin and darbepoetin, are widely used to treat anemia in patients with cancer.¹ These drugs received approval from the US Food and Drug Administration (FDA) for cancer indications in 1993 and 2002, respectively, based on transfusion benefits among patients with nonmyeloid malignancies who developed anemia associated with chemotherapy.^2- 3 When erythropoietin received its approval, concerns over venous thromboembolism (VTE) and tumor progression were raised.⁴ Early trials evaluated serum hemoglobin levels and red blood cell transfusions when ESAs were administered to patients for treating chemotherapy-associated anemia. Two trials^5- 6 reported in 2003 evaluated the potential for ESAs to improve overall or progression-free survival. In these trials, poorer survival was identified for patients with breast cancer who were treated with ESA and chemotherapy, as well as patients with head and neck cancer who received radiotherapy. Published overviews from the Cochrane Collaboration, the National Institutes for Clinical Excellence, and the Blue Cross/Blue Shield Technology Evaluation Center of clinical trials published before mid-2005 included safety information for ESAs in the oncology setting.^7- 10 Two overviews found increased VTE risks associated with ESAs, but none identified mortality risks.^8- 10 The FDA has indicated that it will hold a meeting of its Oncologic Drug Advisory Committee (ODAC) in March 2008 to review safety concerns with ESAs that had been previously reported at ODAC meetings in 2004 and 2007.^11- 12 In this article, we review information for phase 3 trials that assessed ESA administration for treatment of anemia among patients with cancer. The primary data source, the 2006 Cochrane overview, was supplemented by reports for phase 3 trials reported more recently.^{9- 10,12- 24}

We reviewed results of trials evaluating ESAs for the treatment of anemia in the oncology setting (Figure 1). Data sources included reports of outcomes for phase 3 trials published between January 1, 1985, and April 1, 2005, as described previously,^7- 8 augmented by review of phase 3 trials identified in MEDLINE or EMBASE for ESA studies in the oncology setting published or presented between April 2005 and January 17, 2008 (key words: clinical trial, erythropoietin, darbepoetin, and oncology); presentations at the 2007 ODAC meeting on ESAs¹²; or phase 3 clinical trial summaries reported by health authorities, ESA manufacturers, or clinical investigators at national conferences.^{11- 13,16,19,25} Our review includes both independent and industry-sponsored studies.

Information on VTE and mortality rates was abstracted by independent researchers (A.T.S., K.J.G., and S.E.B.) and compared. Where disagreements were identified, the first author of the article resolved these differences. There was 100% agreement on abstracted results in the final data set.

The meta-analyses evaluated VTE rates and survival rates for all studies as well as according to prospective inclusion of survival as primary or secondary outcome measures. To test for differences of treatment, the test of interaction was used.²⁶ Effect estimates for relative risks (RRs), hazard ratios (HRs), and 95% confidence intervals (CIs) were derived from Stata version 9.1 (Stata Inc, College Station, Texas), calculated with random-effects models, and pooled by use of the Dersimonian and Laird method.²⁷ When mortality events were not available, HRs were calculated by using the inverse variance method to pool HRs. When VTE events were not available, a correction factor (0.5) was used to compute the RRs. All statistical tests were 2-sided. Effect estimates were deemed statistically significant when P ≤ .05.

In the 2006 Cochrane overview, 42 trials with 8167 patients were evaluated for overall survival. We excluded 4 trials that were included in the Cochrane overview because 1 study evaluated patients with myelodysplastic syndrome rather than cancer and 3 studies had HRs that were not estimable. We added 13 new trials (5369 patients).^12- 24 In total, survival was evaluated for 13 611 patients with cancer who were treated in 51 phase 3 trials and VTE was evaluated for 8172 patients with cancer who were treated in 38 phase 3 trials (Figure 1). Trials differed with respect to study drug, patient numbers, treatment duration, concomitant treatments, and cancer diagnoses (Table 1 and Table 2).^{2,5,11- 24,28- 61} For the mortality analysis, a median of 223 patients were included in the individual trials (range, 30-985). Epoetin alfa or epoetin beta were evaluated in 40 trials with 8878 patients and darbepoetin in 11 trials with 4733 patients. Twenty-six trials included 7384 patients with a single cancer diagnosis (lung cancer [7 trials], breast cancer [6 trials], head and neck cancer [4 trials], cervical cancer [3 trials], ovarian cancer [3 trials], lymphoma [2 trials], and multiple myeloma [1 trial]). Duration of ESA treatment ranged from 6 to 52 weeks and duration of follow-up ranged from 11 to 62 months. Treatment-related anemia was evaluated in 45 trials with 11 522 patients, and cancer-related anemia was evaluated in 6 trials with 2089 patients. Concomitant treatment varied between trials as follows: chemotherapy (28 trials), radiotherapy (3 trials), chemoradiotherapy (7 trials), mixed treatment (2 trials), palliative radiotherapy (1 trial), no treatment (7 trials), and treatment not reported (3 trials).

Eight trials (4062 patients), recently identified in 2 FDA safety advisories,^25,62 individually demonstrate significantly increased mortality risks and faster tumor growth among patients treated with ESA (Table 3).^{5,12,14,19,36- 37,40,48,58} These trials included patients with anemia and breast cancer (2 studies),^12,40,48 non–small cell lung cancer (1 study),⁵⁸ head and neck cancer (2 studies),^5,14 nonmyeloid cancers (1 study),¹⁹ lymphoma (1 study),³⁷ and cervical cancer (1 study).^36,36 Among these trials, sample sizes ranged from 70 to 985 patients, with a median of 533 patients. Five trials^{14,19,36- 37,48} administered the drug darbepoetin alfa, 2 trials^40,58 evaluated epoetin alfa, and 1 trial⁵ evaluated epoetin beta. Furthermore, in 3 trials,^37,40,48 patients received concomitant chemotherapy; 2 trials^5,14 included patients who received concomitant radiotherapy; in 1 trial,¹⁹ no concomitant treatment was used; in 1 trial,³⁶ concomitant chemoradiotherapy was used; and in 1 trial,⁵⁸ palliative radiotherapy was administered to some patients.

Survival was evaluated for 13 611 patients with cancer who were treated in 51 phase 3 trials (Figure 1). For survival, the HR for mortality was significantly higher for patients with cancer who were treated with ESA vs the control (placebo) group (HR, 1.10; 95% CI, 1.01-1.20; P = .03) (Figure 2). This association was not dominated by a small number of trials, with the largest 8 trials each contributing between 5% and 9% of the total number of clinical trial patients. Subgroup analyses evaluated mortality risks with ESA administration to patients with anemia and cancer for 6 trials with 2089 patients who were not receiving chemotherapy or radiation therapy (anemia of cancer) and 45 trials with 11 522 patients that evaluated patients with anemia and cancer with chemotherapy or radiation therapy–associated anemia (treatment-related anemia). Patients in the ESA group had nonsignificantly increased mortality risks in the anemia of cancer studies (HR, 1.29; 95% CI, 1.00-1.67; P = .05) and a nonsignificant increase in mortality risk in the treatment-related anemia studies (HR, 1.09; 95% CI, 0.99-1.19). There was no significant heterogeneity when evaluating all 51 trials as well as between trials evaluating anemia of cancer vs treatment-related anemia (heterogeneity χ² = 1.99; I² = 21.1%; P = .16).

Venous thromboembolism was evaluated for 38 trials that included 8172 patients (Figure 1). These trials identified a significantly increased risk of VTE among patients treated with ESA (334 events among 4610 patients treated with ESA vs 173 events among 3562 control patients; RR, 1.57; 95% CI, 1.31-1.87) (Figure 3). This association also was not dominated by a small number of trials.

We found a 1.57-fold increased VTE risk and a 1.10-fold increased mortality risk when ESAs were administered to patients with anemia and cancer. These risks are important given the prevalence of ESA use as a supportive care drug among patients with cancer as well as the dissemination of a series of safety advisories by the FDA and ESA manufacturers.

Our mortality results differ from those reported in published overviews based on trials reported before mid-2005 (Table 4).^{7- 11,62- 64} Only 23.8% of trials included in the 2006 Cochrane Review prospectively evaluated survival as a primary or secondary outcome. However, 8 recently reported trials with 4062 patients have individually identified increased rates of tumor progression or mortality with ESA use; all of these trials prospectively evaluated survival as a primary or secondary outcome (Table 3).^{5,14,19,36- 37,40,48,58} Although the mechanism by which ESAs may affect survival of patients with cancer is not completely understood, concern exists over the potential for ESAs to directly affect tumors.⁶⁵

Expression of erythropoietin and erythropoietin receptors has been demonstrated in a variety of human cancers.⁶⁶ Erythropoieitn stimulation of cancer cells in vitro activates signal transduction pathways, including phosphatidylinositol 3-kinase-Akt and JAK-STAT (Janus kinase-Signal Transducer and Activator of Transcription).^67- 68 Depending on the type of cancer, activation of the erythropoietin/erythropoietin receptor signaling axis results in measurable cellular effects, including proliferation, antiapoptosis, and invasion.^69- 72 Erythropoietin-mediated functions may result from autocrine/paracrine signaling or recruitment of both endogenous and exogenous erythropoietin by the tumor.^73- 75 Clearly, many issues remain to be clarified regarding the specific actions of ESAs in human cancer cells.

Our estimate of a 1.57-fold increased VTE risk with ESA administration to patients with cancer mirror findings reported in published overviews and by the ESA manufacturers at ODAC meetings.^7- 9 Increased risks of thrombovascular complications with ESAs have been noted in other clinical settings. The Normal Hematocrit Study⁷⁶ of patients with kidney and heart disease identified a 1- to 5-fold increased risk of myocardial infarctions and vascular access thrombosis when ESAs were administered with target hemoglobin levels of 14 g/dL. Similarly, the Correction of Hemoglobin and Outcomes in Renal Insufficiency (CHOIR) study⁷⁷ of patients with chronic kidney disease identified increased risks of mortality and congestive heart failure among patients who received ESAs targeted to hemoglobin levels of 13.5 g/dL vs 11.3 g/dL. In these settings, no relationship was identified between hemoglobin levels and increased VTE risks.

Additional research on ESA safety is clearly needed. At the prior ODAC meetings in 2004 and 2007, FDA reviewers indicated that they would monitor survival and tumor progression rates for patients with cancer participating in ongoing phase 3 trials.^11- 12 Data for 8 trials that have completed patient accrual since that meeting are currently being evaluated.^25,62 Four completed trials evaluated chemotherapy-associated anemia settings in which ESAs were withheld when hemoglobin levels reached 13 g/dL. Six studies, with 3000 patients, continue to evaluate the potential for ESAs to affect locoregional disease responses or survival among patients with cancer.^14,62 At a December 2007 workshop on ESAs and tumor growth convened by the National Cancer Institute, clinical and basic science researchers agreed that continued preclinical work is necessary to evaluate the precise role of erythropoietin and erythropoietin-receptor expression in human cancers, and future clinical trials should include the collection of tissue specimens to directly assess ESA effects in tumor cells.

Our analysis had some study limitations. We did not have access to original source data and therefore evaluated trial summaries reported in overviews, published literature, and FDA briefing materials. Although additional trials might exist, it is unlikely that neither the FDA nor the manufacturers would be aware of such studies. We pooled results of individual trials, most of which had limited ability to evaluate survival or tumor progression. The VTE definitions varied across trials and VTE rate was not a predefined primary outcome measure in any trial. Finally, we did not report separately on epoetin vs darbepoetin, because the American Society of Clinical Oncology/American Society of Hematology guidelines and the National Comprehensive Cancer Network guidelines, the FDA, and the ODAC considered the products as belonging to a single class.⁷⁸

In conclusion, we found that ESA administration to patients with cancer is associated with increased VTE and mortality risks. Safety concerns account in large part for changes observed in patterns of use, reimbursement policies, clinical guidelines, and FDA-approved package inserts pertaining to ESAs in the oncology setting.^25,78- 82 Our findings, in conjunction with basic science reports on erythropoietin and erythropoietin receptors in solid cancers, raise concern about ESA safety for patients with cancer.