Definitive Radiotherapy +/- Erythropoietin for Squamous Cell Carcinoma of the Head and Neck: Preliminary Report of RTOG 99-03

Reviewer: Neha Vapiwala, MD
Abramson Cancer Center of the University of Pennsylvania
Last Modified: October 5, 2004

Presenter: Mitchell Machtay
Presenter's Affiliation: RTOG
Type of Session: Plenary


The presence of anemia has been associated with poor outcome in cancer patients undergoing radiation therapy.  This has been particularly evident in patients with cervical and head and neck cancer.  A proposed explanatory mechanism for this clinical observation has been the induction of tumor radioresistance in the presence of hypoxia.  RTOG 85-27 (Lee et al. 1998) documented the association of statistically significantly higher locoregional failure rates with lower patient hemoglobin levels.

This RTOG study was designed to investigate the local control impact of erythropoietin (epo) in patients with unresectable locoregionally advanced head and neck cancer. Erythropoietin is a glycoprotein product of recombinant technology that is FDA approved for use in patients with significant chemotherapy-induced anemia. Of note, its use in this trial was off-label, as the baseline hemoglobin levels of patients were higher than those recommended by the FDA or the NCCN guidelines.   


Materials and Methods

  • Prospective randomized trial
  • Eligible pts had non-metatstatic, unresectable locally advanced squamous cell cancers of the head and neck, Karnofsky performance status 0-2, with established anemia of chronic disease
  • Pre-tx hemoglobin (Hb) levels were between 9-13.5 g/dL (9-12.5 g/dL for women)
  • Pts were stratified by stage (I/II vs. III/IV), gender, baseline Hb (9-11.5 g/dL  vs. 11.6-13.5 g/dL), and use of chemotherapy
  • Pts were randomized to one of two arms:
    • Arm 1 = radiation therapy (XRT) to 66-72 Gy 
    • Arm 2 = XRT + epo 40,000 U subcutaneously q week with first dose starting 7-10 days before first XRT dose
  • Epo was discontinued if Hb levels rose to >16 g/dL for men and >14 g/dL for women
  • For stage I/II pts, conventional fractionation was used to total doses of 66-70 Gy
  • For stage III/IV pts, either a) conventional fractionation (70 Gy/35 fraction) with chemotherapy or b) altered fractionation (72 Gy/42 fractions) with no chemotherapy was allowed
  • Primary study endpoint was locoregional control (LRC).
  • Secondary endpoints were overall survival, change in Hb levels, major toxicities and patterns of failure
  • The study was designed with the hypothesis that epo would produce a 33% reduction in relative risk of locoregional failure.


  • These results represent an unplanned interim analysis of 148 pts (initial planned goal was 372 pts).  
  • The study was permanently closed by the Data Monitroing Committee in light of a negative epo German trial whose results were pubished in 10/03 and actually suggested a detrimental effect of epo on pt outcome.    
  • The two arms were well- balanced for pt characteristics.
  • Both the XRT and the epo were given as per protocol in the vast majority of cases.
  • These results were updated on 8/22/04.
  • Of 141 evaluable pts, the results are as follows:
    arm 1 (n = 70)                            
    arm 2 (n=71)
    Mean Change Hb                
    Complete response rate    
    2-yr LR failure                      
    47%   (p=0.75)
    Crude LR Failure                  
    Crude Distant Mets               
    Total Failure Rate                 
    2-yr Overall Surv (OS)
     59%    (p=0.23)
    OS Hazard Ratio = 1.41

  • No statistically significant difference in acute toxicity.

Author's Conclusions

  • This study shows that the use of epo can significantly raise Hb levels in locally advanced head and neck pts.
  • However, there is no significant benefit in locoregional control or survival.
  • Furthermore, one can not rule out an ADVERSE effect of epo on outcome.


Clinical/Scientific Implications

This study adds to the negative results published by Henke et al. in October 2003 that described inferior 3-yr locoregional control with epo compared to placebo. It also highlights the confusing interrelationships between anemia, hypoxia, tumor radiosensitivity, and the use of erythropoietin.

There are multiple radiobiological data to explain the lack of benefit seen in these epo trials. Preclinical studies have repeatedly demonstrated that experimental tumor cells quickly acclimate to acute anemia, and thus there is no significant effect on tumor cell radiosensitivity. Furthermore, chronic anemia conditions do not cause changes in radiosensitivity  (ie: radioresistance) in experimental tumors, either.  Finally, Joiner et al. has shown that correcting anemia with the use of epo does not increase cell radiosensitivity.

Perhaps the most important point to appreciate from this data is that the use of epo may actually promote tumor growth rather than enhancing anti-tumor therapy. Tumor cells are known to express higher levels of epo receptors than normal cells, and this receptor expression is increased in the presence of hypoxia. Epo receptors have been shown to mediate processes contributing to cell growth, and thus activation of these receptors with the ligand epo may actually enhance tumor survival.

Such preclinical data, combined with the negative clinical data outlined here, clearly lead to the conclusion that there is no role for the routine use of epo in this patient group.