Data from shorter time periods than those described in this manuscript were presented at the 2008 HOPA/ISOPP conference, June 18–21, 2008, Anaheim, CA; and in abstract form at the 2008 Annual American Society of Clinical Oncology Meeting: JClinOncol 2008,26(Suppl):Abstract 20589.
Conflict of interest: Amgen funded the work described in this manuscript.
Over one million new cases of cancer are diagnosed in the United States each year . Chemotherapy treatment is associated with the development of anemia in approximately 50–60% of these patients , a risk that is increased in patients over 65 years of age . Fatigue resulting from anemia can have a profound impact on physical functioning, mental health, and overall quality of life . Erythropoiesis-stimulating agents are used to treat chemotherapy-induced anemia and to reduce the risk of transfusions [5–10].
Systematic reviews of randomized clinical studies using meta-analysis have shown that erythropoiesis-stimulating agents significantly reduce transfusion risk by 36–59% [11–13]. However, a retrospective 2009 study  based on 1991–2002 SEER-Medicare data concluded that transfusion rates were unchanged after the introduction of erythropoiesis-stimulating agents, whereas a recent registry study [15, 16] and a practice-based Medicare study  found that transfusion rates did increase as erythropoiesis-stimulating agent use declined.
Before the introduction of erythropoiesis-stimulating agent therapy, red blood cell transfusion was the only option for correction of anemia. Red blood cell transfusion continues to be indicated in situations of life-threatening anemia or when rapid reversal is required; however, transfusions carry their own inherent risks. Packed red blood cells have the potential to mediate graft-versus-host disease and cause volume overload, transfusion-associated lung injury, and febrile, nonhemolytic reactions . Risks of infection and alloimmunization have also been documented extensively . These risks are compounded in patients who require multiple transfusions to treat their anemia as hemoglobin levels decline.
Before 2007, practice guidelines for erythropoiesis-stimulating agent treatment of chemotherapy-induced anemia recommended that corrective therapy be considered for symptomatic patients with hemoglobin levels 10–11 g/dL, with the goal of raising the level to the 10–12 g/dL range [20–22]. Since then, results from eight clinical trials in patients receiving erythropoiesis-stimulating agents for anemia associated with chemotherapy [8, 23–25], or the unapproved indications of radiotherapy [26, 27] and anemia of cancer [28, 29], have raised concerns regarding increased mortality and potential risks such as venous thromboembolism and tumor progression. All eight studies targeted erythropoiesis-stimulating agent use to hemoglobin levels ≥12 g/dL. These concerns also prompted three Oncology Drugs Advisory Committee meetings convened by the US Food and Drug Administration to provide guidance in minimizing this putative risk, the most recent occurring in March 2008 [30–32].
Responding in part to these concerns, the Centers for Medicare and Medicaid Services opened a National Coverage Decision, which was finalized in July 2007. For chemotherapy-induced anemia patients, this decision restricted reimbursement for erythropoiesis-stimulating agents to those with hemoglobin levels <10 g/dL at initiation of therapy. Erythropoiesis-stimulating agents would no longer be covered if the pretreatment hemoglobin level equaled or exceeded 10 g/dL, if the hemoglobin level rose to exceed this level after 4 weeks of therapy, or if the observed hemoglobin levels increased by more than 1 g/dL in a 2-week period. The maximum duration of covered therapy was set to 8 weeks after completion of a chemotherapy regimen. Reimbursements for the nonlabeled use of erythropoiesis-stimulating agents to treat anemia of cancer and anemia associated only with radiotherapy were also discontinued.
After the National Coverage Decision, the American Society of Clinical Oncology/American Society of Hematology and National Comprehensive Cancer Network updated their evidence-based practice guidelines to state that erythropoiesis-stimulating agent therapy should be considered as the hemoglobin level is falling and approaches 10 g/dL and that the dose should be adjusted to maintain hemoglobin levels within the 10–12 g/dL range while avoiding transfusions.
The U.S. product labeling for erythropoiesis-stimulating agents, revised most recently in August 2008 , more fully describes the potential risks of tumor progression and mortality and includes information on each of the eight studies generating concerns. Current prescribing information for darbepoetin alfa contains a boxed warning stating that erythropoiesis-stimulating agents shortened overall survival and/or time to tumor progression in clinical studies in patients with breast, non-small-cell lung, head and neck, lymphoid, and cervical cancers when dosed to target a hemoglobin level of ≥12 g/dL . The warning also states that the risks of shortened survival and tumor progression have not been excluded when erythropoiesis-stimulating agents are dosed to target a hemoglobin of <12 g/dL, within the range recommended by current evidence-based guidelines. Also, erythropoiesis-stimulating agents are not indicated in any patient who is being treated with the expected outcome of cure, regardless of treatment with myelosuppressive chemotherapy (a recommendation made in the apparent absence of a documented evidence base). The current recommendations are to initiate erythropoiesis-stimulating agents at hemoglobin level <10 g/dL and to use the lowest dose required to avoid transfusions; the goal of erythropoiesis-stimulating agent treatment is defined as transfusion avoidance rather than a specific target hemoglobin range.
Because patients with hemoglobin levels <10 g/dL are already at increased risk for transfusions , the goal of transfusion avoidance would seem to be underserved by requiring patients' hemoglobin levels to fall to this level before initiating erythropoiesis-stimulating agent treatment. This has prompted concerns that chemotherapy-induced anemia patients receiving chemotherapy may be required to undergo more transfusions as a result of the July 2007 National Coverage Decision. In this study, we examined erythropoiesis-stimulating agent utilization and transfusion rates from electronic medical records in an equivalent timeframe before and after the National Coverage Decision to determine the impact, if any, of this change on red blood cell transfusions.
At the time of the study, the database comprised Varian electronic medical records (Varian Medical Systems, Palo Alto, CA) from 52 clinics and 350 oncologists, representing more than 150,000 patients. Among these sites, not all reported integrated transfusion data. For this study, the transfusion analysis set included seven practices that did record transfusion administrations in structured data fields within the electronic medical record. These practices were located in seven states with 39 sites of care, representing 194 physicians and 34,450 patients.
The Oncology Drugs Advisory Committee and National Coverage Decision meetings concerning the use of erythropoiesis-stimulating agents in cancer were largely conducted from March 2007 through June 2007. To compare changes in transfusion practice pre- and post-National Coverage Decision, two equivalent observational periods were selected. The calendar months were identical to account for any potential seasonal influences. The pre-National Coverage Decision study period was defined as June 2006 through March 2007. The post-National Coverage Decision comparator period was defined as June 2007 through March 2008. Within each period, episodes of chemotherapy care were defined as a group of chemotherapy administrations, with a 90-day gap delimiting the beginning and end of an episode. Therefore, the data set includes observations for transfusions delivered through June 2007 in the pre-National Coverage Decision group and June 2008 in the post-National Coverage Decision group because it uses the 90-day look-forward period for episodes of chemotherapy care ending in March. High or moderate myelosuppressive chemotherapy was defined as any of the following: any regimen containing topotecan, docetaxel, or paclitaxel; or cyclophosphamide or doxorubicin in combination with either paclitaxel or vincristine with average days between cycles >17.5 and <28. Highly myelosuppressive chemotherapy was defined as a regimen of cyclophosphamide, docetaxel, and doxorubicin; cyclophosphamide, doxorubicin, and ifosfamide; or cyclophosphamide or doxorubicin, with either paclitaxel or vincristine with ≤17.5 average days between cycles. The percent of episodes of chemotherapy care with myelosuppressive chemotherapy, erythropoiesis-stimulating agent administration, blood transfusions, and the patient's minimum hemoglobin measure during the episodes of chemotherapy care were collected. Statistical significance was determined for differences pre- and post-National Coverage Decision using the χ2 test of independence for differences in proportions, the t-test for differences in means, and the two-sample linear rank test for differences in medians. No adjustments were made for the multiple comparisons that were conducted.
Characteristics of patients in the study are summarized in Table I. The pre-National Coverage Decision sample (n = 4,784) included fewer patients than the post-National Coverage Decision sample (n = 5,605). The mean age in both groups was approximately 62 years, with slightly less than half of patients ≥65 years of age. There was a significantly greater percentage of female patients in the pre-National Coverage Decision sample (P = 0.025). As a result, there were more gynecologic cancer diagnoses in the pre-National Coverage Decision patients (P = 0.004). Lung cancer was also more prevalent among pre-National Coverage Decision patients (P = 0.038). No other cancer type showed significant differences in distribution between groups. The mean duration of episodes of chemotherapy care was lower in the pre-National Coverage Decision period (P = 0.012), but the median duration was the same across periods (P = 0.175).
Post-National Coverage Decision period was June 2007 to March 2008, and the pre-National Coverage Decision period was June 2006 to March 2007.
SD is standard deviation.
P-values for χ2 test on proportions (reported as percentages in the table), t-test on means, and two-sample linear rank test on medians, comparing pre-National Coverage Decision with post-National Coverage Decision.
The number of episodes of chemotherapy care was 4,784, and the number of patients was 4,430.
The number of episodes of chemotherapy care was 5,605, and the number of patients was 5,193.
Episodes of chemotherapy care duration (days), mean
Episodes of chemotherapy care duration (days), median
Age categories (%)
Cancer type (%)
Table II summarizes outcomes by treatment received pre- and post-National Coverage Decision. Over the two test periods, the numbers of episodes of chemotherapy care with chemotherapy treatment were comparable. Similar proportions of patients received myelosuppressive chemotherapy in the pre-National Coverage Decision period (29.3%; n = 1,403 episodes of chemotherapy care) and the post-National Coverage Decision period (28.5%; n = 1,598 episodes of chemotherapy care) (P = 0.360). In contrast, erythropoiesis-stimulating agent administrations significantly decreased by 10.9 percentage points (a relative 26% reduction) from the pre-National Coverage Decision period (41.3%; n = 1,975 episodes of chemotherapy care) to the post-National Coverage Decision period (30.4%; n = 1,704 episodes of chemotherapy care) (P < 0.001). Concurrently, there was a significant increase of 1.2 percentage points in the rate of blood transfusions, a 17% increase from the pre-National Coverage Decision period (7.6%; n = 364) to the post-National Coverage Decision period (8.9%; n = 501 episodes of chemotherapy care) (P = 0.015). The mean minimum hemoglobin measured during the episodes of chemotherapy care decreased from the pre-National Coverage Decision (10.9 g/dL) to the post-National Coverage Decision period (10.7 g/dL; P < 0.001).
Table 2. Outcomes in the Pre-National Coverage Decision and Post-National Coverage Decision Periods
Patients aged ≥65
Pre-decision (n = 4,784)
Post-decision (n = 5,605)
Relative change (%)
Pre-decision (n = 2,188)
Post-decision (n = 2,592)
Relative change (%)
Post-National Coverage Decision period was June 2007 to March 2008 and the pre-National Coverage Decision period was June 2006 to March 2007.
P-values are for χ2 tests on proportions, comparing pre-National Coverage Decision with post-National Coverage Decision.
Episodes of chemotherapy care with:
Erythropoietin-stimulating agent administration
Mean minimum hemoglobin during episode (g/dL)
The changes in the frequency of erythropoiesis-stimulating agent administrations and red blood cell transfusions were most pronounced in the patient group aged ≥65 years old. Erythropoiesis-stimulating agent administrations decreased significantly by 13.4 percentage points among these patients, a 29% decrease from the pre-National Coverage Decision period (46.0%; n = 1,007) to the post-National Coverage Decision period (32.6%; n = 845) (P < 0.001). The rate of blood transfusions in older patients significantly increased by 2.1 percentage points, a 31% increase from the pre-National Coverage Decision period (6.7%; n = 146) to the post-National Coverage Decision period (8.8%; n = 227) (P = 0.007). The mean minimum hemoglobin measured during the episodes of chemotherapy care decreased from the pre-National Coverage Decision period (10.9 g/dL) to the post-National Coverage Decision period (10.8 g/dL; P = 0.008). In contrast, among patients aged <65 years, the percent of episodes of chemotherapy care with erythropoiesis-stimulating agent administrations decreased by 24% (P < 0.001); however, the 8% increase in blood transfusions was not significant (P = 0.358; Fig. 1).
In this comparative study of the frequency of erythropoiesis-stimulating agent administrations and red blood cell transfusions based on electronic medical records, erythropoiesis-stimulating agent use decreased and transfusions increased significantly after the March 2007 Oncology Drugs Advisory Committee recommendations and the July 2007 Erythropoiesis-Stimulating Agent National Coverage Decision. One consequence of diminished erythropoiesis-stimulating agent use in patients undergoing chemotherapy may be an increase in rescue transfusions for patients with moderate anemia that was previously managed successfully through erythropoiesis-stimulating agent therapy. This influx to the overall pool of chemotherapy-induced anemia patients at risk for transfusions adds to the population previously identified to be at risk, including those with prior transfusions  and low baseline hemoglobin [6, 36].
The pattern of results in this study differs from findings of a 2009 investigation of use of erythropoiesis-stimulating agents in patients treated with chemotherapy from 1991 through 2002 in the Surveillance, Epidemiology, and End Results-Medicare database . In that study, although the proportion of patients receiving an erythropoiesis-stimulating agent increased, the rate of blood transfusions per year remained stable. Importantly, the study was based on data collected from 9 to 19 years ago, before the July 2007 Erythropoiesis-Stimulating Agent National Coverage Decision. In addition, the study covered a broad analytical window, e.g., transfusions in the year before patients received an erythropoiesis-stimulating agent were included. The results of the study reported herein constitute an important complement to the Hershman et al. study , and help put those data into current context along with other recent publications and similar findings.
Chemotherapy-induced anemia can pose serious risks to patients with cancer, including fatigue, tachycardia, shortness of breath, and dizziness. In older patients, particularly, cognitive impairment is a potential concern, as well as augmentation of existing comorbidities . Patients may be reluctant to discuss their anemia symptoms with caregivers, and physicians may not recognize the impact these symptoms have on daily living. The burdens created by anemia symptoms, resulting in decreased work productivity and increased caregiver time, can be significant . The risks and benefits of anemia treatments for chemotherapy-induced anemia patients should be considered carefully by oncologists and their patients.
Over the last 2 years, the risk-to-benefit profile of erythropoiesis-stimulating agents in chemotherapy-induced anemia has been revisited. Before 2007, erythropoiesis-stimulating agents were indicated for use in chemotherapy-induced anemia up to a hemoglobin level of approximately 12 g/dL and were to be withheld if hemoglobin exceeded 13 g/dL. More recently, several studies have shown that chemotherapy-induced anemia patients who received erythropoiesis-stimulating agents to target hemoglobin greater than 12 g/dL have an increased risk of serious cardiovascular and thromboembolic events, mortality, and tumor progression. Through 2008, Food and Drug Administration-mandated label changes have responded to these signals by restricting initiation of erythropoiesis-stimulating agents to patients with hemoglobin <10 g/dL and to the use of the lowest erythropoiesis-stimulating agent dose required to avoid a red blood cell transfusion.
Transfusions, although commonplace, are not without quantifiable risk. Although the most common adverse reaction to a blood transfusion is a febrile nonhemolytic transfusion reaction, which generally resolves without lasting effects, mortality can occur related to transfusions. Worldwide, transfusion-related acute lung injury is the leading cause of transfusion-related mortality. The estimated frequency of occurrence is 1 in 5,000 red cell transfusions, but may be greater because of underreporting . In the United States, errors resulting in infusion of incompatible blood occur every 1 in 14,000 transfusions , and bacterial contamination and sepsis together account for an estimated 10% of transfusion-related deaths . Acute hemolytic transfusion reactions and graft-versus-host disease complications are relatively rare, but the associated sequelae are serious and potentially fatal . Patients requiring chronic transfusions, such as those with myelodysplastic syndromes, are at risk of developing transfusion-associated iron overload . Transfusions also are a proven transmission route for serious infections. As a result of increased surveillance, the current risk for transfusion-transmitted viral and bacterial infections is low, but emerging pathogens continue to pose a potential risk, particularly in cases where adequate detection methods have yet to be developed [19, 43]. In patients undergoing tumor resection with curative intent, transfusion-related immunomodulation may increase the incidence of postoperative infection or promotion of tumor recurrence. Recently, a meta-analysis published by the Cochrane Collaboration reported an increased risk for cancer recurrence in colorectal cancer patients receiving perioperative transfusions .
Like other healthcare resources, the blood supply in the United States is limited. The 2005 National Blood Collection and Utilization Survey Report estimates that the US blood supply carries approximately 650,000 surplus units of blood to accommodate a spike in demand and shifting regional needs . Currently, the majority of red blood cell units are used in critical care and surgical/trauma settings where transfusions serve as a life-saving therapy. Hospitals have postponed elective surgery because of blood inventory shortages . Although historically accounting for a small proportion of blood consumption, an additional influx of cancer patients requiring red blood cell transfusions could add to this episodic shortage and result in additional challenges for the health care system. These considerations assume additional significance in the context of the limited shelf-life of stored red blood cells. In a recent study of patients undergoing cardiac surgery, transfusion of red blood cells that had been stored for more than 2 weeks was associated with increased risk of postoperative complications and reduced short- and long-term survival .
The number of new cancer patients is expected to double from 1.36 million to 2.0 million persons between 2000 and 2050 . During this time, the ≥65 year age group is expected to increase to 88 million . Currently, 50% of all malignancies occur in persons aged 65 years and older. By 2030, this percentage is expected to rise to 70%, during which time the absolute number of cancers in patients in this age group is also expected to double . Because elderly patients with cancer often experience clinical symptoms of anemia at higher hemoglobin levels than do anemic patients without cancer , it is important that treatment guidelines are based on clinical symptoms as well as laboratory values.
The results of this study are consistent with those of a modeling simulation that compared the number of red blood cell units transfused in chemotherapy-induced anemia patients treated with erythropoiesis-stimulating agents with the number of red blood cell units that would be transfused if erythropoiesis-stimulating agents were limited or discontinued . The model predicted that up to 18% of the 2004 marginal US blood supply and up to 15% of the 2008 marginal US blood supply would be required to cover the incremental demand for blood that would arise from a 25% decrease in erythropoiesis-stimulating agent use. The results suggest that limiting erythropoiesis-stimulating agent use in chemotherapy-induced anemia patients could negatively impact the blood supply margin. The authors caution that the public health consequences of such an outcome should be borne in mind when revisions to erythropoiesis-stimulating agent use are being considered.
In this study, patients older than 65 had a substantially greater increase in transfusions compared with patients under age 65, who also experienced a significant increase. Therefore, in addition to the overall increased transfusion risk for cancer patients with chemotherapy-induced anemia, this increase suggests that the subgroup of patients in the Medicare population bear a greater share of the burden of the Centers for Medicare and Medicaid Services decision to restrict erythropoiesis-stimulating agent use.
In conclusion, changes in practice patterns for chemotherapy-induced anemia treatment that followed the Centers for Medicare and Medicaid Services reimbursement decision for erythropoiesis-stimulating agents seem to be impacting patient care. In the year after the July 2007 National Coverage Decision, marked decreases in erythropoiesis-stimulating agent utilization and increases in red blood cell transfusions were observed, with patients in the ≥65-year-old age group experiencing greater changes than were observed in younger patients. Although the study findings are consistent with other recent publications [15–17], further research will be required to determine whether this represents a persistent change in physicians' practice patterns. It is important that practicing oncologists strive to maintain a clear perspective of the relative risks that these treatment options present to patients.
The authors thank Geoffrey Smith, PhD, at Amgen, Inc., for assistance with the writing of the manuscript and Jane Saiers, PhD, at The WriteMedicine, Inc., for assistance with editing.