The use of recombinant erythropoietin for the treatment of chemotherapy-induced anemia in patients with ovarian cancer does not affect progression-free or overall survival


  • Leigh A. Cantrell MD,

    1. Department of Obstetrics and Gynecology, Division of Gynecologic Oncology, University of Virginia, Charlottesville, Virginia
    Search for more papers by this author
  • Shannon N. Westin MD,

    1. Department of Gynecologic Oncology, The University of Texas M. D. Anderson Cancer Center, Houston, Texas
    Search for more papers by this author
  • Linda Van Le MD

    Corresponding author
    1. Department of Obstetrics and Gynecology, Division of Gynecologic Oncology, University of North Carolina, Chapel Hill, North Carolina
    • Department of Obstetrics and Gynecology, Division of Gynecologic Oncology, University of North Carolina, CB 7572, Chapel Hill, NC 27599-7572
    Search for more papers by this author
    • Fax: (919) 966-2646



Studies have suggested that erythropoietin-stimulating agents (ESAs) may affect progression-free survival (PFS) and overall survival (OS) in a variety of cancer types. Because this finding had not been explored previously in ovarian or primary peritoneal carcinoma, the authors of this report analyzed their ovarian cancer population to determine whether ESA treatment for chemotherapy-induced anemia affected PFS or OS.


A retrospective review was conducted of women who were treated for ovarian cancer at the corresponding author's institution over a 10-year period (from January 1994 to May 2004). Treatment groups were formed based on the use of an ESA. Two analyses of survival were conducted to determine the effect of ESA therapy on PFS and OS. Disease status was modeled as a function of treatment group using a logistic regression model. Kaplan-Meier curves were generated to compare the groups, and a Cox proportional hazards model was fit to the data.


In total, 343 women were identified. The median age was 57 (interquartile range, 48-68 years). The majority of women were Caucasian (n = 255; 74%) and were diagnosed with stage III (n = 210; 61%), epithelial (n = 268; 78%) ovarian cancer. Although the disease stage at diagnosis and surgical staging significantly affected the rates of disease recurrence and OS, the receipt of an ESA had no effect on PFS (P = .9) or OS (P = .25).


The current results indicated that there was no difference in cancer-related PFS or OS with use of ESA in this cohort of women treated for ovarian cancer. Cancer 2011. © 2010 American Cancer Society.

Erythropoietin (EPO) is a renal hormone that increases the production of erythrocytes in response to hypoxia. Recombinant human EPO was introduced initially 20 years ago to treat anemia diagnosed in renal dialysis patients. Two preparations of recombinant EPO-stimulating agents (ESAs) currently are marketed in the United States (epoetin alfa and darbepoetin alfa), and both have similar clinical results and safety profiles; a third preparation, epoetin beta, is available outside of the United States. ESA use in patients with cancer was initiated to reduce the need for erythrocyte transfusion. When administered to patients with chemotherapy-induced anemia, treatment with an ESA consistently decreases the need for erythrocyte transfusion by 50%.1 Additional studies have correlated the use of ESAs with improvements in fatigue and productivity in patients with cancer.2-4 Initial labeling or treatment guidelines for ESAs included use for chemotherapy-induced anemia (CIA) and anemia associated with cancer and allowed for the initiation of ESA treatment at hemoglobin (Hgb) levels ≤11 g/dL.

After the use of ESAs became commonplace for the treatment of CIA, the safety of that use came into question. A few studies suggested that prognosis was compromised with use of ESAs, specifically with regard to disease progression and survival. A meta-analysis of randomized, controlled trials (N = 13,933 patients) reported that ESA use in patients with cancer was associated with increased mortality and shortened survival; however, when ESAs were evaluated using only studies that treated CIA (N = 10,441 patients), there was no difference in survival.5 This led to class labeling revisions in 2007 and 2008. In 2010, an updated meta-analysis (N = 15,323 patients) confirmed that there was no significant affect of ESAs on survival or disease progression in patients who received ESAs for the treatment of CIA or anemia from cancer.6 Current US Food and Drug Administration recommendations state that ESAs are not to be used if there is intent to cure and should be used only in patients who are receiving chemotherapy with Hgb levels <10 g/dL.7 Recent literature indicates that there also may be an undefined association with decreased survival.8

It is estimated that 21,500 American women were diagnosed with ovarian cancer in 2009.9 Standard ovarian cancer therapy includes surgical cytoreduction followed by platinum-based chemotherapy, which has known myelosuppressive and nephrotoxic effects and is associated with a high rate of anemia. In addition, women with ovarian cancer generally are postmenopausal, elderly and are less tolerant of anemia. A recent Gynecologic Oncology Group study of 5 different chemotherapy regimens noted a 20% incidence of grade 3 or 4 anemia.10 In a review of the role of epoetin beta in patients with gynecologic malignancies, Cornes and Boiangiu opined that women with gynecologic cancer may comprise a subgroup of cancer patients who potentially benefit the most from treatment with epoetin.11 We sought to examine this cohort of women with ovarian cancer who received chemotherapy with or without an ESA before the ESA labeling changes and to determine whether there was a difference in progression-free survival (PFS) or overall survival (OS).


After obtaining an institutional review board exemption, all women who were diagnosed with ovarian or primary peritoneal cancer and received chemotherapy between January 1994 and May 2004 were identified from medical records at the University of North Carolina at Chapel Hill. Charts were reviewed, and demographic, treatment, and survival data were abstracted. Demographic data were collected and included age at diagnosis, race, body mass index (BMI), tobacco use, and pertinent comorbidities. Disease and treatment data included disease stage at diagnosis, histology, type of chemotherapy, blood transfusion, and use of ESA (epoetin alfa or darbepoetin alfa). Survival data included dates of recurrence and death and the date of last follow-up or discharge to hospice.

ESA use during the study period was initiated when Hgb values fell below 11 g/dL at the discretion of the treating physician. All patients had a workup for correctable causes of anemia (ie, iron studies). No patient was treated who had an Hgb level>11g/dL. Patients received either weekly epoetin alfa (40,000 IU dosing) or darbepoetin alfa (2.25 mcg/kg dosing), which was added to ESA treatment regimens when it became available in 2002. During the study period, epoetin and darbepoetin alfa were analyzed together as “ESA received.” Many patients in this cohort received chemotherapy with curative intent and would not have been compliant with current ESA labeling. Various chemotherapy regimens were used and the majority contained a platinum agent (Table 1).

Table 1. Chemotherapy Regimens Used
Chemotherapy RegimenNo ESA, N=253ESA, N=90
  1. ESA indicates erythropoietin-stimulating agent.

Platinum alone20
Taxane alone11

We measured PFS from the date of diagnosis to the date of first recurrence. Disease-specific OS was measured from the date of diagnosis to the date of death from disease or entry into hospice. Patients who had recurrent disease or were disease free were censored at the date of their last visit (through March 30, 2009) to calculate OS in months. Covariates that were included in the analysis included age at diagnosis (as a continuous variable), race, BMI, disease stage at diagnosis, histology, tobacco use, and diagnosis of hypertension. BMI was defined according to World Health Organization criteria.12

For survival analyses, patients were divided into 2 groups: those who received an ESA (Group 1) and a control group of patients who did not receive an ESA (Group 2). To determine PFS, disease status was modeled as a function of treatment group with the covariates mentioned above using a logistic regression model. All covariates were included in the initial model, and backward selection (the elimination of variables analyzed one by one that were not statistically significant) was used to remove covariates that were nonsignificant at the 0.05 level. Kaplan-Meier curves were generated to compare the groups, and a Cox proportional hazards model was fit to the data using SAS software (SAS Institute, Inc., Cary, NC). Treatment with ESA status was forced to remain in the model as the variable of interest.


In total, 343 women were identified using the inclusion criteria. The median age of our population was 57 years (interquartile range, 48-68 years). The majority of patients were Caucasian (255 women; 74%), and African-American women composed the next most common patient group (78 women; 23%). The majority of patients were diagnosed with stage III (210 women; 61%), epithelial (268 women; 78%) disease. The median follow-up of both groups was similar (35 months in the ESA-treated group and 37 months in the untreated group). Most women were normal weight (134 women; 39%), had never used tobacco (237 women; 69%), and were normotensive at diagnosis (213 women; 62%). No patient was treated if their Hgb level was >11 g/dL. Ninety patients (26%) received an ESA. Notable differences between treatment groups included older age at diagnosis, a greater proportion of women with hypertension, fewer patients who used tobacco, and a greater proportion of stage IV disease in the ESA treatment group. Table 2 describes the demographics of the study group in detail.

Table 2. Demographic and Clinical Characteristics of the Study Population
FactorNo. (% of Total)P
All Patients, N=343No ESA, N=53ESA, N=90
  1. ESA, erythropoietin-stimulating agent; IQR, interquartile range; BMI, body mass index; FIGO, International Federation of Gynecology and Obstetrics.

Median age at diagnosis [IQR], y57 [48-68]55 [46-67]60.5 [53-69]<.001
Median follow-up, mo 3537 
Race   .57
 Caucasian255 (74)186 (74)69 (77) 
 African-American78 (23)59 (23)19 (21) 
 Other10 (3)8 (3)2 (2) 
BMI   .42
 Underweight: <19 kg/m252 (15)   
 Normal weight: 19-24.9 kg/m2134 (39)   
 Overweight: 25-30 kg/m253 (15)   
 Obese: >30 kg/m2104 (30)   
FIGO stage at diagnosis   .001
 I39 (11)35 (14)4 (4) 
 II23 (7)18 (7)5 (6) 
 III210 (61)153 (60)57 (63) 
 IV37 (11)19 (8)18 (20) 
 Unstaged34 (10)28 (11)6 (7) 
 Epithelial268 (78)   
 Nonepithelial75 (22)   
Tobacco use   .04
 Never237 (69)   
 Current or previous106 (31)   
Hypertension   .006
 Yes130 (38)85 (34)45 (50) 
Median hemoglobin, g/dL
 Preoperative 12.612.2 
 Postoperative, prechemotherapy 10.510.5 

Disease stage at diagnosis and surgical staging status had significant effects on disease recurrence. Patients who had stage III or IV ovarian cancer had greater odds (OR, 10.5; P<.0001 and OR, 32.5; P<.0001, respectively) of developing recurrent disease compared with women who had early stage disease (OR, 2.3; P = .15). Similarly, women who did not undergo surgical staging had greater odds of developing recurrent disease (OR, 7.9; P = .0003). Although a difference in the recurrence rate by age was noted, older women trended toward more recurrent disease; the difference did not reach statistical significance.

Our variable of interest, treatment with an ESA, had no effect on PFS or the disease recurrence rate (OR, 1.04; P = .9). Twenty-six percent of the study population received an ESA. Among the women who had no evidence of disease, 20% received an ESA. Among the women who remained alive with disease, 36% had received an ESA. The full logistic regression model and the reduced logistic regression model, which contained only the significant variables and ESA, are presented in Table 3. Backward selection was used to remove variables from the model in the following order: hypertension, histology, race, BMI, and tobacco use.

Table 3. Logistic Regression Model Results for Progression-Free Survival of the Study Cohort of Women With Ovarian Cancer (N=343)
PredictorFull ModelReduced Model
  1. OR indicates odds ratio; ESA, erythropoietin-stimulating agent; BMI, body mass index; FIGO, International Federation of Gynecology and Obstetrics.

Use of ESA0.959.89801.040.9027
Age at diagnosis1.022.04561.022.0215
 African American1.879.0983
FIGO disease stage
Tobacco use
 Current smoker0.607.2254
 Previous smoker0.572.0994
Epithelial histology1.393.3518

The receipt of treatment with an ESA had no effect on disease-specific OS. (HR, 0.82; P = .25) A higher disease stage at diagnosis and the lack of surgical staging significantly increased the risk of death. It is noteworthy that African-American patients also experienced a significant increase in the risk of death (HR, 1.4; P = .039) (Table 4).

Table 4. Results of the Cox Proportional Hazards Full and Reduced Models for Overall Disease-Specific Survival for the Study Cohort of Women With Ovarian Cancer (N=343)
PredictorFull ModelReduced Model
  1. HR indicates hazard ratio; ESA, erythropoietin-stimulating agent; BMI, body mass index; FIGO, International Federation of Gynecology and Obstetrics.

Use of ESA0.851.35450.824.2546
Age at diagnosis1.027<.00011.024<.0001
 African American1.375.08741.424.0390
FIGO disease stage
Tobacco use
 Current smoker0.753.2999
 Previous smoker0.893.5683
Epithelial histology0.930.7140

Kaplan-Meier analysis was used to generate survival curves, which also demonstrated the lack of an association between receiving an ESA and disease-specific OS; the median survival of patients who received an ESA was 35 months compared with 37 months for patients who did not receive an ESA (Fig. 1).

Figure 1.

Survival in months (SurvMos) is illustrated for women who did and did not receive an erythropoietin-stimulating agent (ESA).


In the current retrospective study of patients with ovarian cancer, there was no difference in PFS or disease-specific OS between patients who did and did not receive ESAs. It is important to note that our patients received ESAs for the treatment of CIA, and the targeted Hgb levels were consistent with product labeling and recommendations at that time. Although higher disease stage and age were associated with worse outcomes, we did not observe that the receipt of an ESA had an impact on either PFS or OS.

Concerns regarding the safety of ESAs and their possible association with disease progression started to emerge in 2003. Studies of patients with head and neck cancer who were receiving radiation therapy suggested that locoregional PFS was compromised in patients who received ESAs. ESAs were added to radiation study protocols initially, because it was hypothesized that increasing Hgb concentrations would raise tumor oxygenation, resulting in improved treatment.13, 14 Several breast cancer studies also demonstrated that survival was decreased for patients who received an ESA with their chemotherapy.15 Criticisms of those studies have included inadequate analysis of prognostic factors, targeting Hgb levels >12 g/dL, and imbalanced control and treatment groups.15 However, several studies with targeted Hgb levels <12 g/dL have also suggested a compromised outcome. In a study of 156 patients with lung cancer, survival was decreased in patients who received ESAs with a target Hgb level <12 g/dL; however, that study was not designed to determine differences in survival.16

This group of studies collectively led to concern regarding compromised disease outcome, however there are studies that did not demonstrate compromised PFS or overall survival. Many of those studies used a target Hgb level consistent with product labeling. Unfortunately, 1 limitation of this group of studies, is that they were not designed prospectively to study survival as an outcome.17-20 In a recent meta-analysis of darbepoetin α administered to patients with CIA, no association was observed with increased disease progression or death. It is noteworthy that, although there has been concern that negative results, from off-label studies using Hgb targets of >12 g/dL, were caused by using the higher Hgb level, in that analysis, PFS reportedly was not effected when Hgb levels were 12 g/dL or 13 g/dL.3, 6 To our knowledge, no studies of ESAs used in accordance with labeling guidelines (treatment of CIA with a starting Hgb level <11 g/dL) have demonstrated a negative impact on survival.

To date, there have been only 3 studies of ESA treatment in patients with gynecologic cancers. The results from 2 studies of patients with cervical cancer suggested that ESAs have a negative impact on this particular patient population. In a retrospective review of women with cervical cancer who received radiation therapy as their primary treatment, ESA treatment was correlated with disease recurrence and death.21 In a prospective Gynecologic Oncology Group study of 114 patients, an ESA was given concurrent with radiation plus weekly platinum, and a target Hgb level of 14 g/dL was used. The hypothesis for that trial was that increasing oxygen-carrying capacity would improve the therapeutic effect of radiation. The study was terminated early because of a higher incidence of thrombotic events in the ESA-treated group; and, at 30 months of follow-up, there was no difference in PFS or OS. In both of those studies, ESA was administered with elevated target Hgb levels.22 Wilkinson et al prospectively studied patients with ovarian cancer who received ESA treatment to determine the impact on quality of life and hematologic changes. Those authors incidentally observed a nonsignificant trend toward worsening disease among patients who received ESAs and concluded that this finding was associated with disease stage.23 To our knowledge, no studies have evaluated the impact of ESAs on the outcome of ovarian cancer patients who received chemotherapy.

The mechanisms by which ESAs may affect cancer behavior are multiple and complex. EPO receptor expression has been observed in a variety of tumor cell lines in vitro (breast, lung, central nervous system, head and neck) and in a variety of tissues, including cardiomyocytes, smooth muscle, and vascular endothelium. These observations have raised concern that there may be direct stimulation of tumor growth through these receptors. However, in a recent, extensive study of EPO receptor expression in tumor cell lines, no ESA-induced activation of the EPO receptor was observed.24 In addition, in studies by Sinclair et al, EPO messenger RNA was detected in all cell types without evidence of subsequent signaling or expression.25 Thus, evidence demonstrating the reliable function of these receptors and signaling and lack of evidence of their functionality in mouse models have not supported this route as a clear mechanism for tumor growth. Direct stimulation of endothelial cells by ESAs is another consideration for promotion of tumor growth.26 It is hypothesized that stimulation of endothelial cells would lead to further neovascularization of tumors, resulting in increased growth or reduced sensitivity to chemotherapy or radiation. Finally, some investigators believe that ESA stimulation of platelet activation may stimulate tumor growth. Studies currently are underway to determine whether the latter 2 hypotheses have biologic validity.

The recent controversy surrounding the use of ESAs has made many physicians wary of prescribing them and has limited their utility. In the current study, we observed no correlation between ESA use and the rate of ovarian cancer progression or survival in patients who had received an ESA within labeling guidelines. We acknowledge the weaknesses inherent to retrospective studies and do not propose that our results should lead to prescribing changes. However, we do believe that these results should continue to stimulate relevant discussion and research regarding the true risks and benefits of ESAs, and we caution against the broad rejection of ESA use in a patient population that may find symptomatic relief from the benefits of ESA therapy.


The authors made no disclosures.