Abstract

Background: Anemia associated with cancer and cancer therapy is an important clinical and economic factor in the treatment of malignant diseases. Methods: We conducted a systematic literature review to assess the efficacy of erythropoietin to prevent or treat anemia in cancer patients with regard to red blood cell transfusions, hematologic response, adverse events, and overall survival. We searched the Cochrane Library, Medline, EMBASE, and other databases for relevant articles published from January 1985 to December 2001. We included all randomized controlled trials that compared the use of recombinant human erythropoietin (plus transfusion, if needed) with no erythropoietin treatment (plus transfusion, if needed). Relative risks (RRs) and 95% confidence intervals (CIs) were calculated under a fixed-effects model. Clinical and statistical heterogeneity were examined with sensitivity analyses and meta-regression. Statistical tests for effect estimates were two-sided. Results: We identified 27 trials involving 3287 adult patients. Patients treated with erythropoietin had a lower relative risk of having a blood transfusion than untreated patients (RR = 0.67, 95% CI = 0.62 to 0.73). Erythropoietin-treated patients with baseline hemoglobin levels lower than 10 g/dL were more likely to have a hematologic response than untreated patients (RR = 3.60, 95% CI = 3.07 to 4.23). The relative risk for thromboembolic complications after erythropoietin treatment was not statistically significantly increased (RR = 1.58, 95% CI = 0.94 to 2.66) compared with that of untreated patients. There is suggestive but inconclusive evidence that erythropoietin may improve overall survival (adjusted data: hazard ratio [HR] = 0.81, 95% CI = 0.67 to 0.99; unadjusted data: HR = 0.84, 95% CI = 0.69 to 1.02). Conclusions: Erythropoietin treatment may reduce the risk for blood transfusions and improve hematologic response in cancer patients. However, our favorable survival outcome is in contrast to two large (N = 351 and 939) recently published randomized controlled trials in which erythropoietin-treated patients had statistically significantly worse survival than untreated patients. Possible reasons for the disparity with our results include differences in study population and design, higher target hemoglobin levels and higher risk of thromboembolic complications, and concerns that erythropoietin may stimulate tumor growth.

Anemia, defined as an inadequate number of hemoglobin-containing red blood cells, is a widely prevalent complication among cancer patients and varies by type of neoplasia and cytostatic treatment ( 1 ) . Apart from the physical symptoms ( 2 ) and diminished quality of life ( 3 ) patients with anemia experience, there is some evidence that anemia, with the consequence of increased tumor hypoxia, might result in a poorer response to radiotherapy or chemotherapy ( 48 ) . On the basis of these observations, researchers have hypothesized that strategies to diminish cancer-related anemia might not only alleviate anemia-related symptoms and improve quality of life but also improve tumor response and possibly extend overall survival time. However, randomized controlled trials testing this hypothesis have generated conflicting evidence. Results of a phase III trial showed that patients treated with erythropoietin had statistically significantly improved disease-free survival compared with untreated patients ( 9 ) ; however, two recently published trials reported statistically significantly worse tumor control and survival rates ( 10 , 11 ) .

Historically, blood transfusion has been the treatment of choice for severe cancer-related anemia. Severe anemia, which is defined as a hemoglobin level less than 8 g/dL, is usually treated, whereas mild-to-moderate anemia (hemoglobin level of 8–11 g/dL) is left untreated in most patients. Although blood transfusion is the fastest means to alleviate symptoms associated with anemia, there are short- and long-term risks associated with this treatment, such as transmission of infectious agents, transfusion reactions, alloimmunization, and overtransfusion ( 12 ) . The development of increasingly more aggressive antineoplastic treatments that may lead to anemia has increased the need for blood transfusions and has prompted oncologists to weigh the advantages and disadvantages of transfusion. Two forms of recombinant human erythropoietin (rHuEpo), epoetin alfa and epoetin beta, both with similar clinical efficacy ( 13 , 14 ), are available to treat anemia and have been tested in randomized controlled trials. Recently, a novel long-acting erythropoietin variant (novel erythropoiesis stimulating protein [NESP] or darbepoetin alfa) has been introduced into clinical practice ( 15 , 16 ) .

Evidence-based guidelines and several systematic reviews on erythropoietin in cancer patients that concentrate on specific underlying malignancies, such as myelodysplastic syndromes ( 17 ) and solid tumors ( 18 ) or specific clinical outcomes such as quality of life ( 19 ) or methodologic issues ( 20 ) , have been published. A systematic review of erythropoietin treatment published by the Agency for Health Care Research and Quality (AHRQ) provides the most comprehensive summary of clinical trials to date ( 21 , 22 ) . Although the AHRQ report provides evidence that erythropoietin significantly reduces blood transfusion requirements, the critical question, whether erythropoietin affects overall survival, could not be addressed using the published data available at that time. In collaboration with authors of the AHRQ report, we here present the results of our systematic review of erythropoietin treatment with respect to hematologic response, red blood cell transfusion need, adverse events, and overall survival. This report is part of the Cochrane Review, in which additional outcomes (tumor response, quality of life, and fatigue) are being addressed. An updated Cochrane Review that includes randomized studies of darbepoetin alfa is being planned ( 23 ) .

M ETHODS

Literature Search

Trials were identified by searching the Cochrane Controlled Trials Register, Medline, EMBASE, and Internet databases of ongoing trials. We manually searched the conference proceedings of the American Society of Hematology, the American Society of Clinical Oncology, and the European Society of Medical Oncology for clinical trial information. Experts in the field at academic institutions and at pharmaceutical companies were contacted to provide information about their study. Citations of all trials identified in the search were checked for additional references. We searched for articles published from January 1985 through December 2001. No language restrictions were used. The full search strategy is published in the Cochrane Library ( 23 ) .

Inclusion Criteria

Only randomized controlled trials among patients of any age with a histologically or clinically proven malignancy were included in this analysis, regardless of type or stage of disease or previous therapy. Other causes of anemia, such as hemolysis or iron deficiency, had to be ruled out. Epoetin alfa or epoetin beta had to be administered subcutaneously or intravenously at doses of at least 300 U/kg of body weight per week and given for at least 4 weeks to prevent or reduce anemia in cancer patients treated with or without nonmyeloablative antineoplastic therapy. Dose adaptation of erythropoietin, depending on hematologic response, was allowed. The control group had to receive identical antineoplastic and supportive treatment, e.g., iron supplementation and placebo or no experimental treatment. We included published and unpublished data. We excluded ongoing studies, interim analyses, crossover studies, quasi-randomized studies, and studies with 10 or fewer patients per study arm. Studies on long-acting substances, such as darbepoetin alfa, were not included in this review.

Study Selection, Quality Assessment, and Data Extraction

Study selection, quality assessment, and data extraction were carried out independently by two reviewers (S. Langensiepen and J. Bohlius). Any disagreement between the reviewers was resolved by discussion involving a third party (A. Engert, G. Schwarzer). Assessment of study quality included randomization, concealment of allocation, masking of patients and clinicians, documentation of dropouts and withdrawals, and intent-to-treat analysis. To obtain unreported data, we contacted the first author of the included trials. The investigators were also asked for details about the study design as well as aggregated patient data with respect to baseline characteristics, and several study outcomes, including hematologic response, initial and final hemoglobin levels, the number of transfused patients, the number of red blood cells transfused, and individual patient data regarding survival duration. Hematologic response was defined as an increase in hemoglobin level of 2 g/dL or more or an increase in hematocrit of 6% or more, unrelated to blood transfusion.

Data Analysis and Statistical Methods

Analyses were performed using Review Manager (RevMan, version 4.2.7 for Windows; Oxford, England): The Cochrane Collaboration 2004; the statistical software package R ( 24 ) was used for additional analyses not possible with RevMan. A fixed-effects model was assumed in all meta-analyses. For binary data, the relative risk was used to measure treatment effect; the Mantel–Haenszel method was used to pool relative risks. The estimated overall relative risk and a plausible value for baseline risk were used to estimate numbers of patients needed to treat. For continuous data, weighted mean differences were calculated. Overall survival was calculated as hazard ratios (HRs) and was based on individual patient data when possible. If individual patient data were not available, the hazard ratio was calculated from data obtained from published reports, using methods described by Parmar et al. ( 25 ) or derived from binary mortality data. The number of patients needed to treat for overall survival was calculated with data from trials with available individual patient data at arbitrarily chosen time points (at 60 and 150 days), based on methods described by Altman et al. ( 26 ) . The P value of the homogeneity test was used only to describe the extent of heterogeneity inherent in a meta-analysis. Potential causes of heterogeneity were explored by performing sensitivity and subgroup analyses ( see below). The influence of a single large study on the pooled estimates was tested in a sensitivity analysis by including and excluding it. In meta-analyses with at least four trials, a funnel plot was generated and a linear regression test ( 27 ) was performed to examine the likely presence of publication bias in meta-analysis. A P value less than .1 was considered statistically significant for the linear regression test. Potential causes of heterogeneity were explored by performing sensitivity analyses to evaluate the effects of hemoglobin level at study entry, type of tumor, antineoplastic therapy given, duration of study, study quality, source of data, and the influence of single large studies on the effectiveness of erythropoietin treatment. In addition to subgroup analyses, a fixed-effect metaregression ( 28 ) was conducted for the outcome “patients receiving red blood cell transfusions.” All covariates showing a statistically significant effect in univariate analysis were included in the multivariable analysis. A backward selection method was used for the model; the covariate with the largest P value was consecutively removed until only statistically significant covariates, according to the Akaike Information Criterion ( 29 ) , remained in the model. Statistical tests for heterogeneity were one-sided; statistical tests for effect estimates were two-sided.

R ESULTS

Description of Studies

A total of 1592 potentially relevant trials were screened for inclusion. Eighty-one studies were retrieved for more information. Of these, 54 were excluded because they did not meet the inclusion criteria (n = 30) or were still ongoing (n = 24). A flow diagram of the process used to identify and evaluate potentially relevant trials is displayed in Fig. 1 .

Fig. 1.

Identifying and evaluating randomized controlled trials (RCTs).

Fig. 1.

Identifying and evaluating randomized controlled trials (RCTs).

A total of 27 randomized controlled trials involving 3287 patients were included in this systematic review ( Table 1 ). Although the search was limited to studies published in year 2001 or earlier, we included one report that was published in 2002 ( 30 ) because unpublished data from that trial were available in 2001. Because the full-text report for that study was published shortly thereafter, it was used as the principal data source. All studies recruited adult patients. Thirteen studies included patients with solid tumors only, six studies focused on patients with hematologic tumors, two studies included patients with myelodysplastic syndrome, and six trials evaluated patients with various malignancies. All trials compared the effectiveness of erythropoietin treatment initiated at study entry (plus transfusions if necessary) compared with no erythropoietin treatment (plus transfusions if necessary). Transfusion of red blood cells was given when the patient's hemoglobin (Hb) level fell below a defined threshold or at the discretion of the treating physician. In most studies, the effectiveness of erythropoietin was measured with regard to hematologic response, transfusion requirements, and adverse events. Several studies also addressed quality of life. Studies were grouped by mean or median baseline hemoglobin level at study entry in intervention studies (Hb ≤ 10 g/dL, 16 studies; Hb 10–12 g/dL, six studies) and preventive trials (Hb > 12 g/dL, five studies). For 17 studies, final hemoglobin levels were reported or submitted by the investigators. In the erythropoietin-receiving groups, the hemoglobin level at week 12 ranged from 9.82 g/dL (standard deviation [SD] = 2.10) ( 31 ) to 13.9 g/dL (SD = 1.85) ( 32 ) . None of the trials directly compared the outcomes of initiating erythropoietin treatment at alternative hemoglobin thresholds. Epoetin alfa was given in 15 studies, and epoetin beta was given in eight studies. In four studies, the specific erythropoietin preparation given could not be clarified. In several studies, different erythropoietin dosages and schedules of administration were compared with one control group ( 3136 ) ; for each of these studies, we randomly assigned control patients to the corresponding number of separate control groups. If the erythropoietin dose was less than 300 U/kg of body weight per week in a single experimental arm, the data were excluded from analyses because of the possibility of an incomplete response ( 33 ) . Duration of treatment ranged from 6 weeks to more than 20 weeks.

Table 1.

Summary of size of trial, patient characteristics, interventions, concealment of allocation, and publication form *

Study Disease Erythropoietin Antineoplastic therapy Allocation concealed Source of data 
Baseline Hb < 10 g/dL       
    Abels 1993 ( 42 )  124 Various Alfa None Yes  Full text  
    Cascinu 1994 ( 42 )  100 Solid tumors Alfa Pb-CT Yes  Full text  
    Case 1993 ( 44 )  157 Various Alfa CT Yes  Full text  
    Cazzola 1995 ( 33 )  146 MM, NHL Beta CT Unclear  Full text  
    Coiffier 2001 ( 38 )  262 Various Beta NR Yes  Abstract  
    Dammacco 2001 ( 55 )  145 MM Alfa CT, some Pb-CT Unclear  Full text  
    Henry 1994 ( 47 )  132 Various Alfa Pb-CT Yes  Full text  
    Italian 1998 ( 48 )  87 MDS Alfa None Yes  Full text  
    Kurz 1997 ( 49 )  35 Gynecolog. Alfa Pb-CT Yes  Full text  
    Littlewood 2001 ( 50 )  375 Various Alfa CT Yes  Full text  
    Oberhoff 1998 ( 51 )  218 Solid Beta Pb-CT Yes  Full text  
    Österborg 1996 ( 31 )  144 MM, NHL, CLL Beta CT Yes  Full text  
    Österborg 2002 ( 30 )  349 MM, NHL, CLL Beta CT Yes  Full text  
    Rose 1994 ( 40 )  221 CLL Alfa CT Unclear  Abstract  
    Silvestris 1995 ( 56 )  54 MM Alfa CT Unclear Full text 
    Thompson 2000 ( 52 )  66 MDS Alfa None Yes  Full text  
Baseline Hb 10–12 g/dL       
    Carabantes 1999 ( 37 )  35 SCLC, Ovarian Alfa Pb-CT Unclear Abstract 
    Henke 1999 ( 34 )  50 Head and neck Alfa or beta RT Unclear Full text 
    Quirt 1996 ( 39 )  56 Various Alfa NR Unclear Abstract 
    Ten Bokkel 1998 ( 32 )  122 Ovarian Beta Pb-CT Yes  Full text  
    Throuvalas 2000 ( 41 )  55 Cervix, bladder NR Pb-CT + RT Yes  Abstract  
    Wurnig 1996 ( 54 )  30 Osteosarcoma Beta CT, some Pb-CT Unclear Full text 
Baseline Hb > 12 g/dL       
    Del Mastro 1997 ( 45 )  62 Breast ca. NR CT Yes  Full text  
    Dunphy 1999 ( 46 )  30 NSCLC, head and neck NR Pb-CT Unclear Full text 
    Kunikane 1997 ( 35 )  72 NSCLC Beta Pb-CT Yes Full text 
    Thatcher 1999 ( 36 )  130 SCLC Alfa Pb-CT Yes  Full text  
    Welch 1995 ( 53 )  30 Ovarian Alfa Pb-CT Unclear Full text 
Study Disease Erythropoietin Antineoplastic therapy Allocation concealed Source of data 
Baseline Hb < 10 g/dL       
    Abels 1993 ( 42 )  124 Various Alfa None Yes  Full text  
    Cascinu 1994 ( 42 )  100 Solid tumors Alfa Pb-CT Yes  Full text  
    Case 1993 ( 44 )  157 Various Alfa CT Yes  Full text  
    Cazzola 1995 ( 33 )  146 MM, NHL Beta CT Unclear  Full text  
    Coiffier 2001 ( 38 )  262 Various Beta NR Yes  Abstract  
    Dammacco 2001 ( 55 )  145 MM Alfa CT, some Pb-CT Unclear  Full text  
    Henry 1994 ( 47 )  132 Various Alfa Pb-CT Yes  Full text  
    Italian 1998 ( 48 )  87 MDS Alfa None Yes  Full text  
    Kurz 1997 ( 49 )  35 Gynecolog. Alfa Pb-CT Yes  Full text  
    Littlewood 2001 ( 50 )  375 Various Alfa CT Yes  Full text  
    Oberhoff 1998 ( 51 )  218 Solid Beta Pb-CT Yes  Full text  
    Österborg 1996 ( 31 )  144 MM, NHL, CLL Beta CT Yes  Full text  
    Österborg 2002 ( 30 )  349 MM, NHL, CLL Beta CT Yes  Full text  
    Rose 1994 ( 40 )  221 CLL Alfa CT Unclear  Abstract  
    Silvestris 1995 ( 56 )  54 MM Alfa CT Unclear Full text 
    Thompson 2000 ( 52 )  66 MDS Alfa None Yes  Full text  
Baseline Hb 10–12 g/dL       
    Carabantes 1999 ( 37 )  35 SCLC, Ovarian Alfa Pb-CT Unclear Abstract 
    Henke 1999 ( 34 )  50 Head and neck Alfa or beta RT Unclear Full text 
    Quirt 1996 ( 39 )  56 Various Alfa NR Unclear Abstract 
    Ten Bokkel 1998 ( 32 )  122 Ovarian Beta Pb-CT Yes  Full text  
    Throuvalas 2000 ( 41 )  55 Cervix, bladder NR Pb-CT + RT Yes  Abstract  
    Wurnig 1996 ( 54 )  30 Osteosarcoma Beta CT, some Pb-CT Unclear Full text 
Baseline Hb > 12 g/dL       
    Del Mastro 1997 ( 45 )  62 Breast ca. NR CT Yes  Full text  
    Dunphy 1999 ( 46 )  30 NSCLC, head and neck NR Pb-CT Unclear Full text 
    Kunikane 1997 ( 35 )  72 NSCLC Beta Pb-CT Yes Full text 
    Thatcher 1999 ( 36 )  130 SCLC Alfa Pb-CT Yes  Full text  
    Welch 1995 ( 53 )  30 Ovarian Alfa Pb-CT Unclear Full text 
*

Hb = hemoglobin; MM = Multiple myeloma; NHL = non-Hodgkin lymphoma; MDS = myelodysplastic syndrome; CLL = chronic lymphatic leukemia; gynecol = ovarian and cervical carcinoma; NSCLC = non–small-cell lung cancer; SCLC = small-cell lung cancer; NR = not reported; CT = chemotherapy; RT = radiotherapy; Pb-CT = platinum-based chemotherapy.

Additional unreported data from personal communication.

Study Quality

Details of the studies are shown in Table 1 . All included studies were described by the authors as randomized. In 17 of 27 trials, which covered 2490 (76%) of the patients included in the analysis, the method for concealing allocation was judged to be adequate. In 10 studies, the method for concealing allocation could not be determined. Fourteen trials were placebo controlled. Most of the studies included intent-to-treat calculations or excluded fewer than 10% of the participants from the analyses. Twenty-two studies were published as full text, and five were abstracts ( 3741 ) . For 19 of the 27 trials, which covered 2930 (89%) of the patients, additional unpublished data were provided by investigators.

Patients Receiving Red Blood Cell Transfusions

Twenty-five trials with 3069 patients reported the percentage of patients who received red blood cell transfusions ( 3033 , 3554 ) . Investigators provided unpublished aggregated results for 1525 (49.7%) of the included patients ( 30 , 31 , 38 , 40 , 43 , 44 , 48 , 51 ) , whereas data from the remaining 1544 (50.3%) patients were taken from published reports ( 32 , 33 , 35 , 36 , 39 , 41 , 42 , 4547 , 49 , 50 , 5254 , 55 ) . Patients treated with erythropoietin had a 33% lower risk of transfusion than untreated patients (relative risk [RR] = 0.67, 95% CI = 0.62 to 0.73, 25 studies; n = 3069, Fig. 2 ).

Fig. 2.

Meta-analysis of the relative risk to receive red blood cell transfusions for cancer patients receiving erythropoietin or standard care. Risk estimates for the single studies ( solid squares ). The size of the squares is proportional to the sample size and the number of events. Horizontal lines denote 95% confidence intervals. Wide confidence intervals were truncated with an arrow . The confidence intervals for the pooled relative risks are shown ( diamonds ). Negative values indicate a relative risk reduction for red blood cell transfusions favoring the erythropoietin group. *Additional unreported data from personal communication. Cazzola 1995c: patients in treatment arm received 5000 IU daily; Cazzola 1995d: patients in treatment arm received 10 000 IU daily; Österborg 1996a: patients in treatment arm received 10 000 IU daily; Österborg 1996b: patients in treatment arm received 2000 IU daily, if Hb did not increase after 8 weeks, dose was increased to 5000 IU and 10 000 IU daily after 12 weeks; Ten Bokkel 1998a: patients in treatment arm received 3 × 150 IU/kg three times a week; Ten Bokkel 1998b: patients in treatment arm received 3 × 300 IU/kg three times a week; Thatcher 1999a: patients in treatment arm received 3 × 150 IU/kg three times a week; Thatcher 1999b: patients in treatment arm received 3 × 300 IU/kg three times a week; Kunikane 2001a: patients in treatment arm received 3 × 100 IU/kg three times a week; Kunikane 2001b: patients in treatment arm received 3 × 200 IU/kg three times a week. Test for overall effect: z = 9.73, P <.001 (two-sided), test for heterogeneity chi-square = 57.8, degrees of freedom = 29, P <.001 (one-sided).

Fig. 2.

Meta-analysis of the relative risk to receive red blood cell transfusions for cancer patients receiving erythropoietin or standard care. Risk estimates for the single studies ( solid squares ). The size of the squares is proportional to the sample size and the number of events. Horizontal lines denote 95% confidence intervals. Wide confidence intervals were truncated with an arrow . The confidence intervals for the pooled relative risks are shown ( diamonds ). Negative values indicate a relative risk reduction for red blood cell transfusions favoring the erythropoietin group. *Additional unreported data from personal communication. Cazzola 1995c: patients in treatment arm received 5000 IU daily; Cazzola 1995d: patients in treatment arm received 10 000 IU daily; Österborg 1996a: patients in treatment arm received 10 000 IU daily; Österborg 1996b: patients in treatment arm received 2000 IU daily, if Hb did not increase after 8 weeks, dose was increased to 5000 IU and 10 000 IU daily after 12 weeks; Ten Bokkel 1998a: patients in treatment arm received 3 × 150 IU/kg three times a week; Ten Bokkel 1998b: patients in treatment arm received 3 × 300 IU/kg three times a week; Thatcher 1999a: patients in treatment arm received 3 × 150 IU/kg three times a week; Thatcher 1999b: patients in treatment arm received 3 × 300 IU/kg three times a week; Kunikane 2001a: patients in treatment arm received 3 × 100 IU/kg three times a week; Kunikane 2001b: patients in treatment arm received 3 × 200 IU/kg three times a week. Test for overall effect: z = 9.73, P <.001 (two-sided), test for heterogeneity chi-square = 57.8, degrees of freedom = 29, P <.001 (one-sided).

The funnel plot was asymmetric ( P <.001), suggesting that negative results (i.e., no reduction of the proportion of patients transfused) were underreported. Whether studies with more than one experimental arm were analyzed with separated experimental arms or merged into one experimental arm did not influence the overall result (data not shown). There was statistically significant heterogeneity among the trials ( P = .0012). Results of the meta-regression analysis showed that the treatment effects for hematologic malignancies and myelodysplastic syndrome were similar, whereas the treatment effect for solid tumors was markedly better. In addition, full-text publications and unpublished data yielded similar results, whereas publications restricted to abstracts only reported larger treatment effects than full-text publications. For each combination of type of allocation concealment, publication, and underlying disease, we used the information in Table 2 to calculate the relative risk of receiving a blood transfusion as follows. For example, the logarithm of the relative risk for an adequately concealed trial using unpublished data on patients with solid tumors is: Intercept + concealment adequate + unpublished + solid tumor = −.60 + 0.08 + 0.20 − 0.28 = −0.60. Accordingly, the relative risk is 0.55. For hematologic malignancies and myelodysplastic syndrome, analogous calculations yield relative risks of 0.86 and 0.80, respectively. Applying the overall relative risk of 0.67 to a hypothetical population with an estimated risk of 50% for transfusion, the number of patients needed to treat is 6.06 (95% CI = 5.26 to 7.41). Thus, approximately six patients would have to be treated with erythropoietin to spare one patient from transfusion. In a hypothetical population with an estimated risk of 70% for transfusion, the number needed to treat is 4.33 (95% CI = 3.76 to 5.29). Thus, in this group, four to five patients would have to receive erythropoietin to spare one patient from transfusion. These results show that the absolute effectiveness of erythropoietin depends on the baseline risk for transfusion.

Table 2.

Patients receiving red blood cell transfusions: results of the meta-regression analysis *

Category Effect (log) SE 95% CI P 
Intercept −0.60 0.11 −0.81 to 0.38 <.001 
Concealment adequate 0.08 0.058 −0.034 to 0.19 .17 
Concealment unclear −0.08 0.058 −0.19 to 0.034 .17 
Full-text publication 0.22 0.12 −0.0083 to 0.45 .059 
Abstract publication −0.42 0.22 −0.86 to 0.012 .057 
Unpublished data 0.20 0.13 −0.042 to 0.45 .10 
Solid tumors −0.28 0.075 −0.43 to 0.13 <.001 
Hematologic malignancies 0.16 0.075 0.016 to 0.31 .03 
Myelodysplastic syndrome 0.097 0.075 −0.049 to 0.24 .19 
Mixed tumors 0.02 0.073 −0.12 to 0.16 .78 
Category Effect (log) SE 95% CI P 
Intercept −0.60 0.11 −0.81 to 0.38 <.001 
Concealment adequate 0.08 0.058 −0.034 to 0.19 .17 
Concealment unclear −0.08 0.058 −0.19 to 0.034 .17 
Full-text publication 0.22 0.12 −0.0083 to 0.45 .059 
Abstract publication −0.42 0.22 −0.86 to 0.012 .057 
Unpublished data 0.20 0.13 −0.042 to 0.45 .10 
Solid tumors −0.28 0.075 −0.43 to 0.13 <.001 
Hematologic malignancies 0.16 0.075 0.016 to 0.31 .03 
Myelodysplastic syndrome 0.097 0.075 −0.049 to 0.24 .19 
Mixed tumors 0.02 0.073 −0.12 to 0.16 .78 
*

For each combination of type of allocation concealment, publication, and underlying disease, the relative risk can be calculated from the table. SE = standard error; CI = confidence interval.

P values (two-sided) were determined using the Wald test.

Number of Red Blood Cell Units Transfused

Thirteen studies with 2056 patients were included in the analysis ( 3033 , 38 , 40 , 4244 , 47 , 49 , 51 , 55 ) . All included data were unpublished aggregated results provided by the investigators. The overall weighted mean difference between the amount of blood transfused in treated and untreated patients showed a statistically significant benefit for patients receiving erythropoietin (−1.00, 95% CI = −1.31 to −0.70, 13 studies, n = 2056). In other words, the erythropoietin-treated group received 1.00 unit of blood less on average than the control group, in which patients received an average of 3.57 units of blood. There was no statistically significant heterogeneity among the trials ( P = .91), and sensitivity analyses did not show statistically significant differences. Whether studies with more than one experimental arm were analyzed with separated experimental arms or merged into one experimental arm did not influence the overall result (data not shown).

Hematologic Response

Data from 2347 patients with a baseline hemoglobin level below 10 g/dL from 14 trials were analyzed ( 30 , 31 , 33 , 38 , 40 , 42 , 44 , 4752 , 55 ) . Investigators provided unpublished aggregated results for 1359 (58%) of the patients ( 30 , 31 , 33 , 38 , 40 , 48 , 51 ) , whereas data from 988 (42%) of the patients was taken from published reports ( 42 , 44 , 47 , 49 , 50 , 52 , 55 ) . Hematologic response was observed in 690 of 1338 patients (median = 48%, range = 9%–70%) in the erythropoietin group compared with 142 of 1009 patients (median = 11%, range = 0%–27%) in the control group, corresponding to a relative risk for hematologic response for erythropoietin versus control of 3.60 (95% CI = 3.07 to 4.23, 14 trials, n = 2347). The funnel plot was asymmetric ( P = .01), suggesting that negative results (i.e., no hematologic response) were underreported. Whether studies with more than one experimental arm were analyzed with separated experimental arms or merged into one experimental arm did not influence the overall result (data not shown). In addition, data from investigators' personal communications were statistically significantly ( P = .04) more conservative (RR = 3.18, 95% CI = 2.61 to 3.88, seven trials, n = 1359) than data from full-text publications (RR = 4.23, 95% CI = 3.31 to 5.64, seven studies, n = 988). Other sensitivity analyses did not show statistically significant differences between the subgroups compared.

Adverse Events

Based on 1738 patients in 12 trials, thromboembolic events, such as transient ischemic attacks, stroke, or myocardial infarction, were observed in 43 (4%) of 1019 of the erythropoietin group and in 14 (2%) of 719 of the control group ( 3032 , 36 , 41 , 43 , 44 , 47 , 48 , 50 , 52 , 53 ) . The pooled relative risk was increased by 58% in the erythropoietin-treated group (RR = 1.58, 95% CI = 0.94 to 2.66, 12 trials, n = 1738), but the increase was not statistically significant. There was no statistically significant heterogeneity among the trials ( P = .99). A funnel plot analysis revealed statistically significant asymmetry ( P = .003), suggesting that negative results (i.e., no thrombotic event) were underreported. We excluded the study published by Littlewood et al. ( 50 ) , which contributed 46.6% to the weight of the overall result in a sensitivity analysis to explore the influence of this single large study. Exclusion of this study did not change the overall result (data not shown). No statistically significant differences were observed between subgroups with different hemoglobin levels at baseline (data not shown). Whether studies with more than one experimental arm were analyzed with separated experimental arms or merged into one experimental arm did not influence the overall result (data not shown).

Hypertension data were reported for 1656 patients from 12 studies ( 31 , 32 , 35 , 36 , 40 , 43 , 44 , 47 , 50 , 53 , 55 , 56 ) . The relative risk of developing hypertension was 19% higher in erythropoietin-treated patients than in untreated patients, but the increase was not statistically significant (RR = 1.19, 95% CI = 0.96 to 1.49, 12 studies, n = 1656, Table 3 ). There was no statistically significant heterogeneity among the trials ( P = .35). Funnel plot analysis revealed statistically significant asymmetry ( P = .02), suggesting that negative results (i.e., no hypertension) were underreported. Whether studies with more than one experimental arm were analyzed with separated experimental arms or merged into one experimental arm did not influence the overall result (data not shown).

Table 3.

Adverse events reported most often: hypertension, thromboembolic and thrombopenic events, rash, and itching *

Outcome No. of trials No. of patients  rHuEPO group   Control group  Pooled RR (95% CI) 
Thromboembolic events 12 1738 43/1019 14/719  1.58 (0.94 to 2.66)  
Hypertension 12 1656 138/1009 64/647  1.19 (0.96 to 1.49) § 
Hemorrhage, thrombocytopenia 1082 74/670 32/412  1.26 (0.85 to 1.86)  
Rash, irritation, itching 675 21/395 11/280  1.17 (0.63 to 2.18)  
Outcome No. of trials No. of patients  rHuEPO group   Control group  Pooled RR (95% CI) 
Thromboembolic events 12 1738 43/1019 14/719  1.58 (0.94 to 2.66)  
Hypertension 12 1656 138/1009 64/647  1.19 (0.96 to 1.49) § 
Hemorrhage, thrombocytopenia 1082 74/670 32/412  1.26 (0.85 to 1.86)  
Rash, irritation, itching 675 21/395 11/280  1.17 (0.63 to 2.18)  
*

rHuEPO = recombinant human erythropoietin; RR = risk ratio; CI = confidence interval. Statistical tests for treatment effects were two-sided; statistical tests for heterogeneity were one-sided.

Events/sample size.

Test for overall effect z = 1.73, test for heterogeneity chi-square = 3.29, df = 12; P = .99.

§

Test for overall effect z = 1.56, test for heterogeneity chi-square = 15.45, df = 14; P = .35.

Test for overall effect z = 1.16, test for heterogeneity chi-square = 4.80, df = 9; P = .85.

Test for overall effect z = 0.50, test for heterogeneity chi-square = 8.32, df = 8; P = .4.

No statistically significant differences between patients treated with erythropoietin and untreated patients were detected for other adverse events analyzed (hemorrhage/thrombocytopenia and rash, Table 3 ).

Overall Survival

Overall survival was compared among 2805 randomized patients from 19 studies ( 3033 , 36 , 38 , 4047 , 4952 , 55 ) . Studies included patients with solid tumors only ( 32 , 36 , 41 , 43 , 45 , 46 , 49 , 51 ) , patients with hematologic malignancies only ( 30 , 31 , 33 , 40 , 55 ), patients with both solid and hematologic malignancies ( 38 , 42 , 44 , 47 , 50 ) , or patients with myelodysplastic syndrome ( 52 ) . Fourteen studies included anemic patients with baseline Hb less than 10 g/dL, two studies assessed patients with baseline Hb from 10 to 12 g/dL ( 32 , 41 ) , and three prevention trials analyzed only patients with baseline Hb level greater than 12 g/dL ( 36 , 45 , 46 ) . For seven studies, including 1235 (44%) of 2805 of the data points analyzed, individual patient data were provided by the authors; for the other studies, the hazard ratio was either calculated as described by Parmar et al. ( 25 ) or from binary mortality data. For the Littlewood et al. ( 50 ) study, which contributed 50.9% weight to the overall analyses, both adjusted (Cox regression) and unadjusted (Kaplan–Meier) aggregated survival data were available, but individual patient data were not available. To test the influence of the two statistical methods used, we used both estimates in the analysis. The pooled hazard ratio was statistically significant (HR = 0.81, 95% CI = 0.67 to 0.99, 19 trials, n = 2805, Fig. 3 ) when we included the adjusted survival data from Littlewood et al. ( 50 ) , but not when we included the unadjusted survival data (HR = 0.84, 95% CI = 0.69 to 1.02, data not shown). There was no heterogeneity among the trials ( P = .6). Whether studies with more than one experimental arm were analyzed with separated experimental arms or merged into one experimental arm did not influence the overall result (data not shown). On average, the median observation time of the included studies was 84 days [range = 42 days ( 46 ) to 7.17 years ( 45 ) ]. Although the included studies were clinically heterogenous, sensitivity analyses for baseline hemoglobin level, tumor entity, different antineoplastic therapies, study quality, duration of follow up, and published versus unpublished data and funnel plot analysis did not reveal statistically significant differences among the studies. For seven studies with available individual patient data, survival probabilities in the control group were estimated by the Kaplan–Meier method for 60 and 150 days after treatment onset. Survival probabilities in the control group ranged from 0.82 to 1.00 at 60 days after treatment onset and from 0.73 to 100 after 150 days. Accordingly, to calculate the number of patients needed to treat, survival probabilities of 0.90 and 0.80 were assumed for 60 and 150 days, respectively. Given this information and the estimated hazard ratio of 0.81, the number of patients with a need to treat after 60 days was 55 (95% CI = 31.4 to 1,054). If we assume this calculated effect is real, 55 patients would have to be treated to prevent one death within 60 days of treatment onset. For the same patient population, the number needed to treat was 29 (95% CI = 16.4 to 559.6) after 150 days. Therefore, treating 29 patients with erythropoietin would prevent one death within 150 days of treatment onset.

Fig. 3.

Meta-analysis of the hazard ratio for overall survival for cancer patients receiving erythropoietin or standard care. Risk estimates for the single studies ( solid squares ). The size of the squares is proportional to the sample size and the number of events. Horizontal lines denote 95% confidence intervals. Wide confidence intervals were truncated with an arrow . The confidence intervals for the pooled hazard ratio are shown ( diamonds ). Negative values indicate a hazard ratio reduction favoring the erythropoietin group. Studies in which individual patient data were available are in bold. *Additional unreported data from personal communication. Österborg 1996a: patients in treatment arm received 10 000 IU daily; Österborg 1996b: patients in treatment arm received 2000 IU daily, if Hb did not increase after 8 weeks, dose was increased to 5000 IU and 10 000 IU daily after 12 weeks; Thatcher 1999a: patients in treatment arm received 3 × 150 IU/kg three times a week; Thatcher 1999b: patients in treatment arm received 3 × 300 IU/kg three times a week. Test for overall effect: z = 2.1, P = .04 (two-sided); test for heterogeneity chi-square = 15.88, degrees of freedom = 18, P = .60 (one-sided).

Fig. 3.

Meta-analysis of the hazard ratio for overall survival for cancer patients receiving erythropoietin or standard care. Risk estimates for the single studies ( solid squares ). The size of the squares is proportional to the sample size and the number of events. Horizontal lines denote 95% confidence intervals. Wide confidence intervals were truncated with an arrow . The confidence intervals for the pooled hazard ratio are shown ( diamonds ). Negative values indicate a hazard ratio reduction favoring the erythropoietin group. Studies in which individual patient data were available are in bold. *Additional unreported data from personal communication. Österborg 1996a: patients in treatment arm received 10 000 IU daily; Österborg 1996b: patients in treatment arm received 2000 IU daily, if Hb did not increase after 8 weeks, dose was increased to 5000 IU and 10 000 IU daily after 12 weeks; Thatcher 1999a: patients in treatment arm received 3 × 150 IU/kg three times a week; Thatcher 1999b: patients in treatment arm received 3 × 300 IU/kg three times a week. Test for overall effect: z = 2.1, P = .04 (two-sided); test for heterogeneity chi-square = 15.88, degrees of freedom = 18, P = .60 (one-sided).

D ISCUSSION

This systematic review analyzed the effectiveness of erythropoietin in the treatment of cancer patients. The primary findings of our study are that cancer patients treated with erythropoietin had a reduced need for red blood cell transfusion and an increased hematologic response compared with untreated patients. There is insufficient evidence to conclude that erythropoietin increases the risk of hypertension or thromboembolic events and suggestive but inconclusive evidence that erythropoietin may improve overall survival.

There is strong and consistent evidence that treatment with recombinant human erythropoietins may reduce the need for red blood cell transfusion and increase hematologic response rates. Patients treated with erythropoietin had a statistically significantly lower risk of blood transfusion and, on average, received statistically significantly fewer blood transfusions than control patients. Similarly, there was consistent evidence that patients with a mean hemoglobin level below 10 g/dL at study entry who were treated with erythropoietin had an increased frequency of hematologic responses than untreated patients, independent of transfusion. From a clinical point of view, sparing patients from red blood cell transfusion is meaningful. Among patients with a 50% risk of blood transfusion, the number needed to treat was 6.06 (95% CI = 5.26 to 7.41), suggesting that six patients would have to be treated with erythropoietin to prevent one patient from undergoing blood transfusion. The overall reduction in red blood cell units transfused seems to be less meaningful. It is unlikely that patients would care whether they received 3.6 or 2.6 units of blood on average. The relevance of this reduction for other health services, such as blood banks, will vary between institutions, depending on their case patient population. Results of a metaregression analysis for the risk of transfusions suggested that erythropoietin might be more effective in patients with solid tumors than in patients with hematologic malignancies or myelodysplastic syndrome. However, there was no evidence that baseline hemoglobin levels at the initiation of erythropoietin treatment (<10 g/dL versus 10–12 g/dL versus >12 g/dL) had a statistically significant influence on the risk of transfusion or the quantitative benefit of erythropoietin. Results of laboratory analyses of predictive factors, such as low baseline serum erythropoietin level, or a decreased ratio of observed-to-predicted erythropoietin levels ( 31 ) , were unavailable to permit a meta-analysis based on individual patient data.

The randomized controlled studies included in our analysis were of good methodologic quality. For example, the methods of concealing allocation were judged to be satisfactory for most of the studies, which included 76% of the patients randomly assigned. The meta-analysis of the specific outcomes showed a high proportion of studies with proper allocation concealment that ranged between 79.1% (transfusion risk) and 94.3% (overall survival). There was no evidence that performance bias affected the overall study results. We obtained additional unreported data for 19 of the 27 trials, representing 89% of the patients included. When we compared results obtained using published versus unreported data, the unreported dataset yielded more conservative pooled estimates than those based only on published data. Thus, any bias introduced into this meta-analysis by unreported data would have decreased rather than increased the apparent effectiveness of erythropoietin.

Studies with more than one experimental group created some methodologic problems because of the specific requirements of the software we used (RevMan version 4.2.5). In studies with more than one erythropoietin dose arm, the control arm had to be artificially divided into as many control arms as treatment arms, which decreased the apparent size of the control arm, reduced some groups to very small numbers, and thus affected the weighting of these studies. However, an alternative analysis (data not shown) that instead merged experimental arms yielded similar results.

This meta-analysis demonstrated suggestive but inconclusive evidence for improved overall survival among patients treated with erythropoietin (HR = 0.81, 95% CI = 0.67 to 0.99). The relevance of this finding is unclear because none of the trials included in the analysis of overall survival was designed or had adequate statistical power to determine whether erythropoietin increases overall survival. Also, this result was not robust to changes in the statistical methods used. Although the pooled hazard ratio was statistically significant when we included data from the largest study ( 50 ) , analyzed by Cox regression, and adjusted for confounders, it was no longer statistically significant if the same study was analyzed with the unadjusted raw data (HR = 0.84, 95% CI = 0.69 to 1.02). The computed effects would be clinically meaningful if they are real. On the basis of the available data and the relatively short follow-up, we estimated that it would be necessary to treat 55 (95% CI = 31.4 to 1054) patients for 60 days with erythropoietin to prevent one death. With a longer follow-up period of 150 days, the estimated number needed to treat was 29 (95% CI = 16.4 to 559.6). However, the confidence intervals around these estimates are wide, indicating substantial uncertainty surrounding these results.

These provocative findings on overall survival from our meta-analysis differ from results of two recently published clinical trials. Henke et al. ( 10 ) reported that among 351 patients with head and neck cancer undergoing radiotherapy, those who received erythropoietin had a worse overall survival (RR = 1.39; 95% CI = 1.05 to 1.84, P = .02) than those who did not. Similarly, a multicenter trial that investigated the use of erythropoietin as an adjunct to chemotherapy among 939 patients with metastatic breast cancer undergoing first-line therapy was terminated early because survival at 12 months was worse in the group that received erythropoietin than the group that did not (70% versus 76%; P = .017), although the survival curves converged at 19 months ( 11 ) . Because our review included only studies published from 1985 through 2001, neither of these two studies ( 10 , 11 ) was included in this analysis. In the study by Leyland-Jones et al. ( 11 ) , the mortality rate during the first 4 months of study was explained in part by an increased incidence of thrombotic and vascular events in the erythropoietin group versus control (1% versus 0.2%) and in part by an increase in incidence of disease progression in the erythropoietin group versus control (6% versus 3%). In the study by Henke et al. ( 10 ) , vascular disorders, including hypertension, hemorrhage, venous thrombosis, pulmonary embolism and cerebrovascular events, were observed in 11% of patients in the erythropoietin group and in 5% of the placebo group when the data were analyzed on an intent-to-treat basis. Taken together, these studies raised concerns about the safety of erythropoietins, particularly when used in clinical trials aimed at raising the hemoglobin at a target level of 12–14 g/dL and higher. These safety questions were discussed during a comprehensive Food and Drug Administration hearing on May 4, 2004 ( 57 ) . The discussion at that meeting, at which results of company-sponsored meta-analyses on the safety of erythropoietins were presented, focused on the difference between the results of studies performed according to the current American Society of Hematology and American Society of Clinical Oncology guidelines and those of studies aimed at increasing hemoglobin to levels that exceed correction of anemia. Clinical studies aimed at maintaining a high hemoglobin level ( 10 , 11 ) seem to report that erythropoietin treatment was associated with a higher risk of thrombovascular events. This higher risk is further supported by the fact that three additional studies targeting hemoglobin levels between 14 and 16 g/dL were closed prematurely because of increased thromboembolic complications in the erythropoietin arm ( 57 ) . In addition, studies evaluating erythropoietin to maintain different hematocrit levels of end-stage renal failure patients with pronounced cardiovascular risk factors showed that patients with high hematocrit levels had an increased mortality due to thrombovascular events ( 57 , 58 ) . Taken together, the major difference between the Henke et al. ( 10 ) study and our systematic review is the substantially higher mean final hemoglobin levels in the former trial at the end of study (15.4 g/dL, SD = 1.7), which may have contributed to the higher number of thrombovascular events. Among the studies included in our analysis, the final hemoglobin levels ranged from 9.82 g/dL (SD = 2.10) ( 31 ) to 13.9 g/dL (SD = 1.85) ( 32 ) . Although the relative risk for thromboembolic events was increased by 58%, with 43 (4%) of 1019 thromboembolic events occurring in the erythropoietin group and 14 (2%) of 719 occurring in the control group, this difference was not statistically significant. Overall, the data evaluated in this systematic review are insufficient to conclude that erythropoietin increases the risk for thromboembolic complications in the clinical settings analyzed in our review.

The reduced tumor control associated with erythropoietin treatment reported in the studies published by Henke et al. ( 10 ) and Leyland-Jones et al. ( 11 ) raises additional pathophysiological considerations. Tumor tissue is often hypoxic, and hypoxia may be more prevalent in anemic patients than in patients with normal hemoglobin levels ( 59 , 60 ) . Tumor hypoxia may impair the effectiveness of chemotherapy and radiotherapy ( 46 , 8 ) . Studies using animal models have provided evidence suggesting that increasing the hemoglobin level with erythropoietin might improve tumor oxygenation ( 61 ) , and results of clinical studies indicate that erythropoietin may improve response to radiotherapy ( 9 ) . However, preclinical studies ( 60 , 61 ) have reported high levels of erythropoietin and erythropoietin receptors in breast cancer cells and other malignancies. Both endogenously produced or exogenously administered erythropoietin could theoretically promote the proliferation and survival of erythropoietin receptor–expressing cancer cells ( 6265 ) . High hemoglobin concentrations may reduce the oxygen supply in the tumor, due to viscous resistance ( 66 ) or thromboembolic events within the tumor bed ( 67 ) and thus diminish tumor response to therapy. In addition, imbalances in baseline prognostic factors between the erythropoietin and control groups in the studies published by Henke et al. ( 10 ) and Leyland-Jones et al. ( 11 ) may also explain the poor tumor control in the erythropoietin-treated patients.

Our literature search ended in December 2001 because that was when the project protocol was finalized and when requests for unpublished data were sent to the investigators. As a consequence, studies published later were not included in this meta-analysis. In addition to the studies of Henke et al. and Leyland-Jones et al., another six studies with approximately 751 patients have been published as full-text articles since. Three of these studies reported survival or mortality data for 90 ( 68 ) , 144 ( 69 ) , and 330 ( 70 ) patients. There were no statistically significant survival differences observed between the erythropoietin group and the untreated patients ( 6870 ) . An update of the Cochrane Review is being planned.

Other adverse events reportedly related to erythropoietin include hypertension and the production of anti-erythropoietin–specific antibodies. In patients with chronic renal failure, hypertension is the most common adverse effect of erythropoietin ( 71 ) . This analysis shows that erythropoietin is associated with a 19% increase (not statistically significant) in the relative risk of hypertension for cancer patients. There has been some discussion about whether neutralizing anti-erythropoietin–specific antibodies cause pure red cell aplasia in patients with chronic renal failure following treatment with erythropoietin ( 7274 ) . The studies included in this review did not report any cases of erythropoietin antibody production, but only six of the included studies have addressed this subject ( 30 , 32 , 36 , 42 , 47 , 51 ) .

In view of the inconclusive evidence presently available, our results suggest that erythropoietin should not be used to increase overall survival outside clinical trials. Erythropoietin may be used routinely outside of clinical trials to increase hemoglobin levels and to reduce the need for transfusion in patients with falling hemoglobin levels approaching 10 g/dL. Adverse events such as thromboembolic complications and hypertension should be monitored.

We thank all authors and coworkers of primary studies, who provided us personally or through pharmaceutical companies with additional study data: K. Aziz, U. Blohmer, S. Cascinu, M. Cazzola, B. Campos, C. Coiffier, D. Antonadou, L. Del Mastro, Italian Cooperative Study Group, C. Kurz, C. Oberhoff, A. Österborg, P. Sevelda, W. Ten Bokkel-Huinink, N. Thatcher, J.A. Thompson, N.A. Throuvalas, and A. Varan. We are indebted for the data contribution of pharmaceutical companies that provided data for various studies (Roche) and admitted additionally insight into their clinical study reports (OrthoBiotech). Special thanks to Ben Djulbegovic and the editors of the Cochrane Haematological Malignancies Group (CHMG), affiliated consumers, and members of the CHMG editorial base for critical advice and strong support.
The CHMG is funded, as part of the Competence Network Malignant Lymphomas, by the German Ministry of Education and Research.
Julia Bohlius was invited by Amgen to participate at a symposium. Jerome Seidenfeld and Margaret Piper are employed by a membership association of health plans. Andreas Engert received research grants for other projects from Amgen, OrthoBiotech, and Roche, and has served as a consultant for Ortho Biotech. Charles Bennett received research grants in the past from Amgen (Thousand Oaks, CA) and Ortho-Biotech (Raritan, NJ) and has done consulting work for Amgen and Johnson & Johnson. Simon Langensiepen and Guido Schwarzer have no potential conflicts of interest.

References

(1)
Knight K, Wade S, Balducci L. Prevalence and outcomes of anemia in cancer: a systematic review of the literature.
Am J Med
 
2004
;
116
(Suppl. 7A):
11S
–26S.
(2)
Engert A. Recombinant human erythropoietin as an alternative to blood transfusion in cancer-related anaemia.
Disease Management and Health Outcomes
 
2000
;
8
:
259
–72.
(3)
Cella D, Kallich J, McDermott A, Xu X. The longitudinal relationship of hemoglobin, fatigue and quality of life in anemic cancer patients: results from five randomized clinical trials.
Ann Oncol
 
2004
;
15
:
979
–86.
(4)
Frommhold H, Guttenberger R, Henke M. The impact of blood hemoglobin content on the outcome of radiotherapy.
Strahlentherapie und Onkologie
 
1998
;
174
(Suppl IV):
31
–4.
(5)
Hockel M, Knoop C, Schlenger K, Vorndran B, Baussmann E, Mitze M, et al. Intratumoral pO2 predicts survival in advanced cancer of the uterine cervix.
Radiother Oncol
 
1993
;
26
:
45
–50.
(6)
Knocke TH, Weitmann HD, Feldmann HJ, Selzer E, Potter R. Intratumoral pO2 measurements as predictive assay in the treatment of carcinoma of the uterine cervix.
Radiother Oncol
 
1999
;
53
:
99
–104.
(7)
Glaser CM, Millesi W, Kornek GV, Lang S, Schull B, Watzinger F, et al. Impact of hemoglobin level and use of recombinant erythropoietin on efficacy of preoperative chemoradiation therapy for squamous cell carcinoma of the oral cavity and oropharynx.
Int J Radiat Oncol Biol Phys
 
2001
;
50
:
705
–15.
(8)
Silver DF, Piver MS. Effects of recombinant human erythropoietin on the antitumor effect of cisplatin in SCID mice bearing human ovarian cancer: a possible oxygen effect.
Gynecol Oncol
 
1999
;
73
:
280
–4.
(9)
Antonadou D, Cardamakis E, Puglisi M, Malamos N, Throuvalas N. Erythropoietin enhances radiation treatment efficacy in patients with pelvic malignancies. Final results of a randomized phase III study.
Eur J Cancer
 
2001
;
37
(Suppl. 6):
S144
.
(10)
Henke M, Laszig R, Ruebe C, Schaefer U, Haase KD, Schilcher B, et al. Erythropoietin to treat head and neck cancer patients with anaemia undergoing radiotherapy: randomised, double-blind, placebo-controlled trial.
Lancet
 
2003
;
362
:
1255
–60.
(11)
Leyland-Jones B; BEST Investigators and Study Group. Breast cancer trial with erythropoietin terminated unexpectedly.
Lancet Oncol
 
2003
;
4
:
459
–60.
(12)
Goodnough LT. Risks of blood transfusion.
Crit Care Med
 
2003
;
31
(12 Suppl):
S678
–86.
(13)
Halstenson CE, Macres M, Katz SA, Schnieders JR, Watanabe M, Sobota JT, et al. Comparative pharmacokinetics and pharmacodynamics of epoetin alfa and epoetin beta.
Clin Pharmacol Ther
 
1991
;
50
:
702
–12.
(14)
Storring PL, Tiplady RJ, Gaines Das RE, Stenning BE, Lamikanra A, Rafferty B. Epoetin alfa and beta differ in their erythropoietin isoform compositions and biological properties.
Br J Haematol
 
1998
;
100
:
79
–89.
(15)
Joy MS. Darbepoetin alfa: a novel erythropoiesis-stimulating protein.
Ann Pharmacother
 
2002
;
36
:
1183
–92.
(16)
Heatherington AC, Schuller J, Mercer AJ. Pharmacokinetics of novel erythropoiesis stimulating protein (NESP) in cancer patients: preliminary report.
Br J Cancer
 
2001
;
84
(Suppl 1):
11
–6.
(17)
Hellstrom Lindberg E. Efficacy of erythropoietin in the myelodysplastic syndromes: a meta-analysis of 205 patients from 17 studies.
Br J Haematol
 
1995
;
89
:
67
–71.
(18)
Quirt I, Micucci S, Moran LA, Pater J, Browman G. The role of erythropoietin in the management of cancer patients with non-hematologic malignancies receiving chemotherapy. Cancer Care Ontario Practice Guidelines Initiative
2003
.
(19)
Bottomley A, Thomas R, van Steen K, Flechtner H, Djulbegovic B. Human recombinant erythropoietin and quality of life: a wonder drug or something to wonder about?
Lancet Oncol
 
2002
;
3
:
145
–53.
(20)
Clark O, Adams JR, Bennett CL, Djulbegovic B. Erythropoietin, uncertainty principle and cancer related anaemia.
BMC Cancer
 
2002
;
2
:
23
.
(21)
Aronson N, Seidenfeld J, Piper MA, Flamm C, Hasselblad V, Ziegler KM. Uses of erythropoietin for anemia in oncology. Evidence report/technology assessment No. 30 Prepared by the Blue Cross and Blue Shield Association Evidence-based Practice Center 2001; Contract No. 290-97-0015. AHRQ Publ No. 01-E009. Rockville (MD): Agency for Healthcare Research and Quality.
(22)
Seidenfeld J, Piper M, Flamm C, Hasselblad V, Armitage JO, Bennett CL, et al. Epoetin treatment of anemia associated with cancer therapy: a systematic review and meta-analysis of controlled clinical trials.
J Natl Cancer Inst
 
2001
;
93
:
1204
–14.
(23)
Bohlius J, Langensiepen S, Schwarzer G, Seidenfeld J, Piper M, Bennett CL, et al. Erythropoietin for patients with malignant disease. The Cochrane Database of Systematic Reviews 2004, Issue 3. Art No.: CD003407. DOI: 10.1002/14651858.CD003407.pub2.
(24)
Ihaka R, Gentleman R. R: a language for data analysis and graphics.
J Computat Graph Stat
 
1996
;
5
:
299
–314.
(25)
Parmar MK, Torri V, Stewart L. Extracting summary statistics to perform meta-analyses of the published literature for survival endpoints.
Stat Med
 
1998
;
17
:
2815
–34.
(26)
Altman DG, Andersen PK. Calculating the number needed to treat for trials where the outcome is time to an event.
BMJ
 
1999
;
319
:
1492
–5.
(27)
Egger M, Smith GD, Schneider M, Minder C. Bias in meta-analysis detected by a simple, graphical test.
BMJ
 
1997
;
315
:
629
–34.
(28)
Thompson SG, Sharp SJ. Explaining heterogeneity in meta-analysis: a comparison of methods.
Stat Med
 
1999
;
18
:
2693
–708.
(29)
Akaike H. Fitting autoregressive models for prediction.
Ann Inst Stat Math
 
1969
;
21
:
243
–7.
(30)
Osterborg A, Brandberg Y, Molostova V, Iosava G, Abdulkadyrov K, Hedenus M, et al. Randomized, double-blind, placebo-controlled trial of recombinant human erythropoietin, epoetin beta, in hematologic malignancies.
J Clin Oncol
 
2002
;
20
:
2486
–94.
(31)
Osterborg A, Boogaerts MA, Cimino R, Essers U, Holowiecki J, Juliusson G, et al. Recombinant human erythropoietin in transfusion-dependent anemic patients with multiple myeloma and non-Hodgkin's lymphoma—a randomized multicenter study.
Blood
 
1996
;
87
:
2675
–82.
(32)
Ten Bokkel Huinink WW, De Swart CA, Van Toorn DW, Morack G, Breed WP, Hillen HF, et al. Controlled multicentre study of the influence of subcutaneous recombinant human erythropoietin on anaemia and transfusion dependency in patients with ovarian carcinoma treated with platinum-based chemotherapy.
Med Oncol
 
1998
;
15
:
174
–82.
(33)
Cazzola M, Messinger D, Battistel V, Bron D, Cimino R, Enller Ziegler L, et al. Recombinant human erythropoietin in the anemia associated with multiple myeloma or non-Hodgkin's lymphoma: dose finding and identification of predictors of response.
Blood
 
1995
;
86
:
4446
–53.
(34)
Henke M, Guttenberger R, Barke A, Pajonk F, Potter R, Frommhold H. Erythropoietin for patients undergoing radiotherapy: a pilot study.
Radiother Oncol
 
1999
;
50
:
185
–90.
(35)
Kunikane H, Watanabe K, Fukuoka M, Saijo N, Furuse K, Ikegami H, et al. Double-blind randomized control trial of the effect of recombinant human erythropoietin on chemotherapy-induced anemia in patients with non-small cell lung cancer.
Int J Clin Oncol
 
2001
;
6
:
296
–301.
(36)
Thatcher N, De Campos ES, Bell DR, Steward WP, Varghese G, Morant R, et al. Epoetin alpha prevents anaemia and reduces transfusion requirements in patients undergoing primarily platinum-based chemotherapy for small cell lung cancer.
Br J Cancer
 
1999
;
80
:
396
–402.
(37)
Carabantes FJ, Benavides M, Trujillo R, Cobo M, Hebrero ML, Garcia S, et al. Epoetin alfa in the Prevention of Anemia in Cancer Patients Undergoing Platinum-Based Chemotherapy (CT). A prospective randomized study [meeting abstract]. Proc ASCO 1999; abstract 2303.
(38)
Coiffier B, Boogaerts MC. Impact of epoetin beta versus standard care on quality of life in patients with malignant disease. 6th Congress of the European Haematology Association 2001; abstract 194.
(39)
Quirt I, Micucci S, Moran LA, Pater J, Browman G. The role of recombinant human erythropoietin (EPO) in reducing red blood cell transfusions and maintaining quality of life (QOL) in patients with lymphoma and solid tumors receiving cytotoxic chemotherapy. Results of a randomized, double-blind, placebo-controlled clinical trial.
Blood
 
1996
;
88
(Suppl 1):
347a
.
(40)
Rose E, Rai K, Revicki D, Brown R, Reblando J and the EPO in Anemia of CLL Study Group. Clinical and health status assessments in anemic chronic lymphocytic leukemia (CLL) patients treated with epoetin alfa (EPO).
Blood
 
1994
;
84
(Suppl 1):
526a
.
(41)
Throuvalas NA, Antonadou D, Boufi M, Lavey R, Malamos N. Erythropoietin decreases transfusion requirements during radiochemotherapy. Proc ASCO 2000; Abstract 1558.
(42)
Abels R. Erythropoietin for anemia in cancer patients.
Eur J Cancer
 
1993
;
29a
(Suppl 2):
2
–8.
(43)
Cascinu S, Fedeli A, Del Ferro E, Fedeli SL, Catalano G. Recombinant Human Erythropoietin Treatment in Cisplatin-Associated Anemia: a randomized, double-blind trial with placebo.
J Clin Oncol
 
1994
;
12
:
1058
–62.
(44)
Case DC, Bukowski RM, Carey RW, Fishkin EH, Henry DH, Jacobson RJ, et al. Recombinant human erythropoietin therapy for anemic cancer patients on combination chemotherapy.
J Natl Cancer Inst
 
1993
;
85
:
801
–6.
(45)
Del Mastro L, Venturini M, Lionetto R, Garrone O, Melioli G, Pasquetti W, et al. Randomized phase III trial evaluating the role of erythropoietin in the prevention of chemotherapy-induced anemia.
J Clin Oncol
 
1997
;
15
:
2715
–21.
(46)
Dunphy FR, Harrison BR, Dunleavy TL, Rodriguez JJ, Hilton JG, Boyd JH. Erythropoietin reduces anemia and transfusions.
Cancer
 
1999
;
86
:
1362
–7.
(47)
Henry DH, Brooks BJ Jr, Case DC Jr, Fishkin E, Jacobson R, Keller AM, et al. Recombinant human erythropoietin therapy for anemic cancer patients receiving cisplatin chemotherapy.
Cancer J Sci Am
 
1995
;
1
:
252
–60.
(48)
Italian Cooperative Study Group. A randomized double-blind placebo-controlled study with subcutaneous recombinant human erythropoietin in patients with low-risk myelodysplastic syndromes.
Br J Haematol
 
1998
;
103
:
1070
–4.
(49)
Kurz C, Marth C, Windbichler G, Lahousen M, Medl M, Vavra N, et al. Erythropoietin treatment under polychemotherapy in patients with gynecologic malignancies: a prospective, randomized, double-blind placebo-controlled multicenter study.
Gynecol Oncol
 
1997
;
65
:
461
–6.
(50)
Littlewood TJ, Bajetta E, Nortier JW, Vercammen E, Rapoport B. Effects of epoetin alfa on hematologic parameters and quality of life in cancer patients receiving nonplatinum chemotherapy: results of a randomized, double-blind, placebo-controlled trial.
J Clin Oncol
 
2001
;
19
:
2865
–74.
(51)
Oberhoff C, Neri B, Amadori D, Petry KU, Gamucci T, Rebmann U, et al. Recombinant human erythropoietin in the treatment of chemotherapy-induced anemia and prevention of transfusion requirement associated with solid tumors: a randomized, controlled study.
Ann Oncol
 
1998
;
9
:
255
–60.
(52)
Thompson JA, Gilliland DG, Prchal JT, Bennett JM, Larholt K, Nelson RA, et al. Effect of recombinant human erythropoietin combined with granulocyte/macrophage colony-stimulating factor in the treatment of patients with myelodysplastic syndrome. GM/EPO MDS Study Group.
Blood
 
2000
;
95
:
1175
–9.
(53)
Welch RS, James RD, Wilkinson PM. Recombinant human erythropoietin and platinum-based chemotherapy in advanced ovarian cancer.
Cancer J Sci Am
 
1995
;
1
:
261
.
(54)
Wurnig C, Windhager R, Schwameis E, Kotz R, Zoubek A, Stockenhuber F, et al. Prevention of chemotherapy-induced anemia by the use of erythropoietin in patients with primary malignant bone tumors (a double-blind, randomized, phase III study).
Transfusion
 
1996
;
36
:
155
–9.
(55)
Dammacco F, Castoldi G, Rodjer S. Efficacy of epoetin alfa in the treatment of anaemia of multiple myeloma.
Br J Haematol
 
2001
;
113
:
172
–9.
(56)
Silvestris F, Romito A, Fanelli P, Vacca A, Dammacco F. Long-term therapy with recombinant human erythropoietin (rHu-EPO) in progressing multiple myeloma.
Ann Hematol
 
1995
;
70
:
313
–8.
(57)
Luksenburg H, Weir A, Wager R. Safety Concerns Associated with Aranesp (darbepoetin alfa) Amgen, Inc. and Procrit (epoetin alfa) Ortho Biotech, L.P., for the Treatment of Anemia Associated with Cancer Chemotherapy [monograph on the internet], FDA Briefing Document, May 4, 2004, Oncologic Drugs Advisory Committee [cited 2005 Jan 3]. Available from http://www.fda.gov/ohrms/dockets/ac/04/briefing/4037B2_04_FDA-Aranesp-Procrit.htm
(58)
Besarab A, Bolton WK, Browne JK, Egrie JC, Nissenson AR, Okamoto DM, et al. The effects of normal as compared with low hematocrit values in patients with cardiac disease who are receiving hemodialysis and epoetin.
N Engl J Med
 
1998
;
339
:
584
–90.
(59)
Becker A, Stadler P, Lavey RS, Hansgen G, Kuhnt T, Lautenschlager C, et al. Severe anemia is associated with poor tumor oxygenation in head and neck squamous cell carcinomas.
Int J Radiat Oncol Biol Phys
 
2000
;
46
:
459
–66.
(60)
Henke M, Guttenberger R. Erythropoietin in radiation oncology—a review.
Oncology
 
2000
;
58
:
175
–82.
(61)
Kelleher DK, Thews O, Vaupel P. Can erythropoietin improve tumor oxygenation?
Strahlentherapie und Onkologie
 
1998
;
174
(Suppl IV):
20
–3.
(62)
Acs G, Acs P, Beckwith SM, Pitts RL, Clements E, Wong K, et al. Erythropoietin and erythropoietin receptor expression in human cancer.
Cancer Res
 
2001
;
61
:
3561
–5.
(63)
Acs G, Zhang PJ, Rebbeck TR, Acs P, Verma A. Immunohistochemical expression of erythropoietin and erythropoietin receptor in breast carcinoma.
Cancer
 
2002
;
95
:
969
–81.
(64)
Arcasoy MO, Amin K, Karayal AF, Chou SC, Raleigh JA, Varia MA, et al. Functional significance of erythropoietin receptor expression in breast cancer.
Lab Invest
 
2002
;
82
:
911
–8.
(65)
Yasuda Y, Fujita Y, Matsuo T, Koinuma S, Hara S, Tazaki A, et al. Erythropoietin regulates tumour growth of human malignancies.
Carcinogenesis
 
2003
;
24
:
1021
–9.
(66)
Vaupel P, Mayer A. Letter.
Lancet
 
2004
;
363
:
992
.
(67)
Janecka IP. Letter.
Lancet
 
2004
;
363
:
993
–4.
(68)
Rosen FR, Haraf DJ, Kies MS, Stenson K, Portugal L, List MA, et al. Multicenter randomized phase II study of paclitaxel (1-hour infusion), fluorouracil, hydroxyurea, and concomitant twice daily radiation with or without erythropoietin for advanced head and neck cancer.
Clin Cancer Res
 
2003
;
9
:
1689
–97.
(69)
Bamias A, Aravantinos G, Kalofonos C, Timotheadou N, Siafaka V, Vlahou I, et al. Prevention of anemia in patients with solid tumors receiving platinum-based chemotherapy by recombinant human erythropoietin (rHuEpo): a prospective, open label, randomized trial by the Hellenic Cooperative Oncology Group.
Oncology
 
2003
;
64
:
102
–10.
(70)
Witzig TE, Silberstein PT, Loprinzi CL, Sloan JA, Novotny PJ, Mailliard JA, et al. Phase III, randomized, double-blind study of epoetin alfa versus placebo in anemic patients with cancer undergoing chemotherapy.
J Clin Oncol
 
2004
;Sep 27 [Epub ahead of print].
(71)
Cirillo M, Capasso G, DeSanto NG. Relationship between hematocrit and blood pressure: implications for primary hypertension.
Nephron
 
1993
;
65
:
505
–10.
(72)
Prabhakar SS, Muhlfelder T. Antibodies to recombinant human erythropoietin causing pure red cell aplasia.
Clin Nephrol
 
1997
;
47
:
331
–5.
(73)
Casadevall N, Nataf J, Viron B, Kolta A, Kiladjian JJ, Martin Dupont P, et al. Pure red-cell aplasia and antierythropoietin antibodies in patients treated with recombinant erythropoietin.
N Engl J Med
 
2002
;
346
:
469
–75.
(74)
Bennett CL, Luminari S, Nissenson AR, Tallman MS, Klinge SA, Mc Williams N, et al. Pure red-cell aplasia and epoetin therapy.
N Engl J Med
 
2004
;
351
:
1403
–8.