As discussed above, it was hypothesized that increasing Hb levels using erythropoietin might improve tumor oxygenation and thus improve the effectiveness of radiation and chemotherapy and subsequently improve survival, a theory that was supported by data from the study by Littlewood et al.56 and a subsequent meta-analysis.57 In this meta-analysis, the data from 19 randomized controlled trials published before May 2002, including 2,865 patients, were assessed; results indicated a survival benefit for patients receiving erythropoietin (adjusted data: hazard ratio 0.81, 95% CI 0.67–0.99; unadjusted data: hazard ratio 0.84, 95% CI 0.69–1.02).55, 57 None of the studies included were designed to detect differences in survival. These promising results were contradicted by two large randomized controlled trials published in 2003, which were specifically designed to assess survival.70, 71

In the first study, published by Henke and co-workers, 351 patients undergoing radiotherapy for head and neck cancer were randomized to receive epoetin beta or placebo, given in parallel with radiotherapy.70 The primary endpoint was locoregional progression-free survival. Reported Hb targets were Hb of 14 g/dl or more in women and 15 g/dl or more in men. Patients received a relatively high dose of epoetin beta (300 IU/kg three times a week). Survival in patients receiving epoetin beta was significantly lower than in patients receiving placebo (RR of death 1.39, 95% CI 1.05–1.84, P = 0.02). There were more episodes of hypertension, hemorrhage, thrombosis and pulmonary embolism in patients receiving epoetin beta than in those receiving placebo (11% versus 5%), and more patients died of cardiac disorders in the treatment group (5.5% versus 3%). The rate of locoregional tumor progression was also higher in patients receiving epoetin beta, with an RR of 1.69 (95% CI 1.16–2.47, P = 0.007).

The second study (BEST), reported by Leyland-Jones et al., was a multinational, multicenter study including 939 women with metastatic breast cancer. Women were prospectively randomized to receive epoetin alfa 40,000 IU once weekly or placebo.71, 72 This study was terminated prematurely by an independent data monitoring committee based on the first 4 months of safety data. There was a significant survival difference between patients receiving epoetin alfa (70%) and those in the placebo group (76%) at 12 months (P = 0.01). This difference was because of an increased mortality in the treatment arm in the first 4 months (41 deaths versus 16 deaths). In particular, mortality rates due to fatal thrombovascular events (1.1% versus 0.2%) and disease progression (6% versus 2.8%) were both higher with epoetin alfa than with placebo. At 19 months there was a convergence of survival curves. Both the Leyland-Jones et al. and Henke et al. studies had methodological limitations, such as baseline imbalances, suboptimal cancer treatment70 and lack of prospective documentation of prognostic factors.71 In the study published by Henke et al., patients with stage III and IV head and neck cancer underwent surgery and received radiotherapy, as this was the standard treatment at the time of protocol development for this study.70 Treatment for this patient group, however, has evolved over time and would nowadays include chemotherapy as well. Some causes of death still remain unclear, despite intensive re-evaluation.

The unexpected results of these and three other studies, which had to be closed prematurely because of increased rates of thromboembolic events, prompted an Oncologic Drugs Advisory Committee hearing at the FDA in May 2004.73 At this meeting, hypotheses to explain these results were discussed, including both methodological and biological factors relating to thromboembolic complications and tumor growth stimulation by erythropoietin.

Opposing mechanisms on the effects of erythropoietin on tumor growth have been proposed. As discussed earlier, tumor tissue is often hypoxic, and increasing Hb levels with erythropoietin might improve tumor oxygenation, and subsequently tumor control and survival. By contrast, it has been hypothesized that erythropoietin might directly stimulate tumor cell growth. Preclinical studies have reported high levels of erythropoietin and erythropoietin receptors in breast cancer cells and other malignancies.74, 75, 76, 77 Either endogenously produced or exogenously administered erythropoietin could theoretically promote the proliferation and survival of erythropoietin-receptor-expressing cancer cells. Preclinical studies used non-physiologic pharmacologic doses to trigger tumor proliferation.76 It remains unclear whether the increased rates of deaths due to tumor progress in both the BEST study and the study by Henke et al. can be explained by tumor growth stimulation, since this observation has not been seen consistently across trials.70, 71, 72

Other randomized controlled trials reported either no differences in tumor control78, 79 or improved tumor control with erythropoietin treatment. Antonadou et al.80 investigated tumor response in 385 participants with pelvic malignancies undergoing radiotherapy with or without erythropoietin. They showed a statistically significant improved disease-free survival at 4 years for participants treated with erythropoietin (85.3%) compared with the control group (67.2%, P <0.001). Blohmer et al.81 investigated the impact of epoetin alfa in patients with high-risk carcinoma of the uterine cervix (n = 257) treated with sequential chemoradiotherapy. The first interim report suggests an improvement in relapse-free survival (17% versus 25%) for patients treated with epoetin alfa (10,000 IU three times weekly); however, this observation is not statistically significant (P = 0.058). Both studies have not been fully published to date.

Whether thromboembolic complications are triggered by higher Hb levels in patients receiving erythropoietin has also been considered. As described above, in the Henke et al. study more patients receiving erythropoietin experienced thrombovascular events and died from cardiac disorders. Very high Hb levels were achieved at the end of the study (15.4 g/dl in men, SD 1.7), which might have substantially contributed to the high number of thrombovascular events and cardiac deaths.70 In three prematurely closed randomized controlled trials, 24 of 34 patients with vascular events treated with epoetin alfa had Hb levels greater than 13 g/dl within 28 days prior to the thromboembolic event.73 Thus, it was concluded that studies targeting Hb levels above 12 g/dl, aiming beyond the mere correction of anemia as suggested in the ASCO/ASH guidelines,50 were associated with a higher risk of thrombovascular events. Similarly, in studies evaluating the effects of erythropoietin on hematocrit in patients with end-stage renal failure who had pronounced cardiovascular risk factors, patients with high hematocrit levels had an increased mortality due to thrombovascular events.82 Erythropoietin might also have a thrombogenic potential independent of Hb levels. A retrospective case–control study in 147 consecutive cervical carcinoma patients undergoing concurrent chemotherapy, radiation and recombinant human erythropoietin found a statistically significant association between erythropoietin treatment and thromboembolic complications (odds ratio compared with women not receiving erythropoietin 10.3, 95% CI 2.3–46.2).83 No association was found between the mean or maximum Hb level and risk of thromboembolic complications, suggesting that erythropoietin might have a genuinely thrombogenic potential. This notion is supported by in vitro studies that demonstrate augmented platelet reactivity in erythropoietin-stimulated RBCs84 and endothelial activation in healthy volunteers.85