ARTICLES

Articles

Erythropoietin to treat head and neck cancer patients with

anaemia undergoing radiotherapy: randomised, double-blind,
placebo-controlled trial
Michael Henke, Roland Laszig, Christian Rübe, Ulrich Schäfer, Klaus-Dieter Haase, Burkhard Schilcher, Stephan Mose,
Karl T Beer, Ulrich Burger, Chris Dougherty, Hermann Frommhold

Summary Introduction
The benefits of radiotherapy for patients with cancer
Background Anaemia is associated with poor cancer control, diminish when anaemia is present.1 Correction of anaemia
particularly in patients undergoing radiotherapy. We has been suggested to reverse this haemoglobin effect,2
investigated whether anaemia correction with epoetin ␤ could thereby improving cancer control. Recombinant human
improve outcome of curative radiotherapy among patients erythropoietin can correct anaemia3–5 and improve quality
with head and neck cancer. of life in anaemic patients with cancer.6,7 Furthermore,
preclinical data suggest that erythropoietin increases the
Methods We did a multicentre, double-blind, randomised, radiosensitivity of tumours8,9 and might improve the
placebo-controlled trial in 351 patients (haemoglobin clinical efficacy of radiation10 and chemotherapy.11
<120 g/L in women or <130 g/L in men) with carcinoma of However, the potential of erythropoietin to improve
the oral cavity, oropharynx, hypopharynx, or larynx. Patients cancer outcomes has not been established. Therefore, we
received curative radiotherapy at 60 Gy for completely (R0) investigated whether epoetin ␤ could improve cancer
and histologically incomplete (R1) resected disease, or 70 Gy control and survival of patients irradiated for head and
for macroscopically incompletely resected (R2) advanced neck cancer.
disease (T3, T4, or nodal involvement) or for primary
definitive treatment. All patients were assigned to sub- Patients and methods
cutaneous placebo (n=171) or epoetin ␤ 300 IU/kg (n=180) Patients
three times weekly, from 10–14 days before and continuing We enrolled patients between March, 1997, and April,
throughout radiotherapy. The primary endpoint was 2001. Patients older than 18 years with histologically
locoregional progression-free survival. We assessed also proven squamous-cell carcinoma of the oral cavity,
time to locoregional progression and survival. Analysis was oropharynx, hypopharynx, or larynx who were scheduled
by intention to treat. to undergo definitive treatment with radiotherapy or
postoperative radiotherapy for advanced disease (T3, T4,
Findings 148 (82%) patients given epoetin ␤ achieved or nodal involvement) qualified for the study. Further
haemoglobin concentrations higher than 140 g/L (women) or eligibility criteria were haemoglobin concentration lower
150 g/L (men) compared with 26 (15%) given placebo. than 120 g/L for women or lower than 130 g/L for men,
However, locoregional progression-free survival was poorer and a Karnofsky score of 60 or more. Exclusion criteria
with epoetin ␤ than with placebo (adjusted relative risk 1·62 were treatment-refractory hypertension, thrombocytosis
[95% CI 1·22–2·14]; p=0·0008). For locoregional progres- (>750⫻109/L), epilepsy, any other simultaneous
sion the relative risk was 1·69 (1·16–2·47, p=0·007) and for malignant disease, treatment with any cytostatic drug
survival was 1·39 (1·05–1·84, p=0·02). within 3 months before the study, hypersensitivity to the
preservative in the study medication, pregnancy or
Interpretation Epoetin ␤ corrects anaemia but does not inadequate contraception, or participation in any other
improve cancer control or survival. Disease control might experimental protocol. The trial was approved by the
even be impaired. Patients receiving curative cancer ethics committees of the participating centres and done in
treatment and given erythropoietin should be studied in accordance with the revised Declaration of Helsinki and
carefully controlled trials. good clinical practice guidelines. All patients provided
written informed consent.
Lancet 2003; 362: 1255–60
Methods
We did a double-blind, randomised, placebo-controlled
Abteilung Strahlenheilkunde der Radiologischen Universitätsklinik, trial. We stratified patients according to tumour resection
Hugstetter Strasse 55, D-79106 Freiburg, Germany (M Henke MD, status: stratum 1, postoperative radiation of complete
Prof H Frommhold MD); Universitätsklinik für Hals-, Nasen- und (R0) resection; stratum 2, postoperative radiation of
Ohrenheilkunde (Prof R Laszig MD); Klinik für Strahlentherapie und incompletely resected disease (R1 or R2); and stratum 3,
Radioonkologie, Universitäts-Klinikum, Münster, Germany primary definitive radiotherapy. Centres were supplied for
(Prof C Rübe MD, U Schäfer MD); Klinik für Strahlentherapie und each stratum with individually numbered but otherwise
Radio-Onkologie, Universität Witten-Herdecke, Wuppertal, Germany identical packages containing at random either placebo or
(K-D Haase MD, Prof B Schilcher MD); Klinik für Strahlentherapie, active drug. For allocation to treatment groups, each new
Klinikum der J W-Goethe-Universität, Frankfurt, Germany patient was assigned the next available medication
(S Mose MD); Klinik für Radioonkologie, Inselspital, Universität, package of the appropriate stratum (figure 1). The study
Bern, Switzerland (K T Beer MD); and F Hoffmann-La Roche Ltd, code was kept sealed at the biometric department of the
Basel, Switzerland (U Burger PhD, C Dougherty MD) sponsor. Sealed envelopes with the code for individual
Correspondence to: Dr Michael Henke patients were provided to the treating physicians and all
(e-mail: henke@uni-freiburg.de) were recollected unopened.

THE LANCET • Vol 362 • October 18, 2003 • www.thelancet.com 1255

For personal use. Only reproduce with permission from The Lancet.
 ARTICLES

Placebo or epoetin ␤ (300 IU/kg) were administered 351 patients enrolled

subcutaneously three times per week, with a minimum of
24 h between treatments. Additional 200 mg iron-III-
saccharate (Haussmann, Switzerland) or 187·5 mg iron-
III-gluconate (Nattermann, Köln, Germany) was 171 assigned placebo 180 assigned epoetin ␤
administered intravenously once weekly if transferrin
saturation was lower than 25%. Alternatively, oral iron
could be given. Placebo or epoetin ␤ were started 54 radiotherapy 60 radiotherapy
10–14 days before and continued throughout radiotherapy. violations violations
Treatment was discontinued when target haemoglobin 14 locoregional 14 locoregional
concentrations were achieved (⭓140 g/L in women or control not control not
⭓150 g/L in men) or when haemoglobin increased by more assessed assessed
than 20 g/L within 1 week; it was resumed if the
haemoglobin concentrations fell to lower than the target
121 available for 121 available for
concentration. radiotherapy-correct radiotherapy-correct
Standard or three-dimensional planning techniques were analysis analysis
allowed for radiotherapy. Whichever planning technique a
centre used for the first study patient had to be used for all
subsequent study patients. The radiation volume included 8 medication 20 medication
the tumour (or tumour bed) with a 2–3 cm safety margin violations violations
and the regional lymph-node areas. 6 mega electron volt
linear accelerators were used and standard dose and
fractionation protocols (five fractions of 2·0 Gy per week or 113 available for 101 available for
five fractions of 1·8 Gy per week) followed. We prescribed per-protocol analysis per-protocol analysis
60 Gy (allowable range 56–64 Gy) to regions for R0 or R1
resected disease, and 70 Gy (allowable range 66–74 Gy) for Figure 1: Trial profile
macroscopically incompletely resected tumour (R2) or
primary definitive treatment. The spinal cord was shielded patients, except those who received less than 80% of their
after 30–36 Gy. scheduled study medication administrations.
Patients were seen for first follow-up 6 weeks after We assessed locoregional progression-free survival in the
completion of radiotherapy, and thereafter every 3 months intention-to-treat population with Cox’s proportional
to assess locoregional tumour control and survival. hazards model,12 for which stratum and American Joint
Radiotherapy quality was ascertained by an independent Cancer Committee stage were cofactors. Differences were
radiation oncologist, who was not involved with the study tested with the two-sided Wald ␹2 test. In addition, we
and who was unaware of treatment group. Resection status, calculated Kaplan-Meier estimates,13 hazard ratios or
tumour stage, treatment volume, safety margins, total dose relative risk with 95% CI, two-sided logrank statistics, and
applied, and overall treatment time were verified from did an unadjusted Cox’s regression analysis. In all defined
source documents. Patients were classified as protocol populations we assessed locoregional progression-free
correct when treated according to protocol, and as minor survival, locoregional progression, and survival.
violation when the treatment volume did not include the Haematological changes were assessed descriptively. We
safety margin or when the total dose was 2·0 Gy higher or summarised the frequency of most severe adverse events by
lower than the allowed dose range. All other protocol body system, excluding pre-existing adverse events at
deviations were classified as major violations. baseline. Cancer-related, non-cancer-related, and
The primary endpoint of the study was locoregional potentially drug-related adverse events were summarised
progression-free survival, defined as the time to separately.
locoregional tumour progression or death, whichever came Multivariate Cox’s regression analyses adjusted for
first. We also assessed time to locoregional tumour different baseline characteristics were done for epoetin-
progression and survival. Tumour progression was assumed ␤ effect. Furthermore, we analysed the primary endpoint in
when tumour size increased by more than 25%. We different subgroups, such as radiotherapy stratum, tumour
measured changes in haematological values (haemoglobin location, and baseline haemoglobin concentration.
concentration, platelet and leucocyte counts, serum iron,
transferrin, and ferritin) weekly during the treatment phase. Role of the funding source
In addition, we recorded adverse events and serious adverse The study sponsor was actively involved in the study
events. design, data collection, and organisation of study conduct,
did the statistical analysis, and participated in interpretation
Statistical analysis of results and drafting and final approval of the report.
The study followed a sequential design and two interim
analyses were planned, preserving a nominal p=0·048 for Results
the final analysis. Study power was set at 80% to detect a We enrolled 351 patients in 23 centres in Austria, France,
32% risk reduction for locoregional progression-free Germany, and Switzerland. The last patient entered the
survival at 220 events. The primary analysis was done by study in April, 2001. At that time the sponsor decided to
intention to treat. We also analysed a radiotherapy-correct omit the scheduled second interim analysis because the
population, including patients who received radiation statistical penalty was deemed to be too high and changes
(dose, fraction, and treatment time) according to protocol in overall conduct of the study were not expected. The
and presented on at least one follow-up visit (this data were unmasked at the end of November, 2002,
population differed from the protocol-correct population analyses finished in April, 2003, and results were
analysed in the radiotherapy quality assessment presented to the investigators in July, 2003. Authors
programme) and a per-protocol population. The per- consented to the final draft of the manuscript in August,
protocol population included all radiotherapy-correct 2003. All randomised patients were included in the

1256 THE LANCET • Vol 362 • October 18, 2003 • www.thelancet.com

For personal use. Only reproduce with permission from The Lancet.
 ARTICLES

Placebo Epoetin ␤ occurred in 66 placebo and 72 epoetin-␤ patients. The

(n=171) (n=180) adjusted relative risk for locoregional progression-free
Characteristics survival was 1·42 (95% CI 1·01– 2·01, p=0·04; table 2)
Male sex 145 (85%) 158 (88%) and the Kaplan-Meier estimate for time to progression
White patients 171 (100%) 179 (99%) slightly, but not significantly, favoured placebo (795 vs
Median (range) age (years) 57 (36–87) 58 (35–81) 551 days, p=0·41). In the per-protocol population, 63
Median (range) weight (kg) 65·5 (40–113) 67 (42–115)
placebo and 58 epoetin-␤ patients experienced
Current smoker 91 (53%) 118 (66%)
Median (range) haemoglobin 118 (6·9–14·6) 117 (8·5–14·4) locoregional tumour progression or death, with an
concentration (g/L) adjusted relative risk for locoregional progression-free
Median (range) serum erythropoietin 11 (3·3–168·1) 11 (11–446·2) survival of 1·35 (95% CI 0·94–1·95, p=0·11; table 2).
concentration (U/L) The Kaplan-Meier estimate was 748 days for placebo
Relapse at entry 13 (8%) 18 (10%) compared with 605 days for epoetin ␤ (p=0·8).
Tumour location
Oral cavity 36 (21%) 43 (24%) In the intention-to-treat population, locoregional
Oropharynx 74 (43%) 72 (40%) tumour progression occurred in 49 placebo and
Hypopharynx 43 (25%) 40 (22%) 65 epoetin-␤ patients, with 122 and 115 censored
Larynx 39 (23%) 41 (23%) observations, respectively. The adjusted relative risk for
AJCC stage locoregional progression was 1·69 (95% CI 1·16–2·47,
I 2 (1%) 2 (1%)
II 0 6 (3%)
p=0·007; table 2). The univariate Kaplan-Meier estimate
III 46 (27%) 37 (21%) showed a difference in time to progression favouring
IV 123 (72%) 135 (75%) placebo (median not reached vs 280 days, p=0·09). In the
Treatment stratum same population, 89 placebo and 109 epoetin-␤ patients
1 94 (55%) 102 (57%) died (82 and 71 censored). The adjusted relative risk of
2 38 (22%) 39 (22%)
3 39 (23%) 39 (22%)
death was 1·39 for epoetin-␤ patients (95% CI 1·05–1·84,
Resection status p=0·02; table 2). In the actuarial analysis, patients treated
R0 94 (71%) 102 (72%) with placebo survived a median of 928 days compared
R1 33 (25%) 33 (23%) with 605 days in the epoetin-␤ group (p=0·09).
R2 5 (4%) 6 (4%) According to stratum, locoregional tumour progression
AJCC=American Joint Cancer Committee. Values n (%) unless marked otherwise. or death occurred in 41 placebo and 47 epoetin-␤ patients
Table 1: Baseline characterstics in radiotherapy stratum 1, and the Kaplan-Meier estimate
for locoregional progression-free survival was 1152 and
1049 days, respectively (p=0·9, figure 2). By contrast,
intention-to-treat population. 18 patients withdrew or 16 placebo and 30 epoetin-␤ patients in stratum
were excluded. 242 patients were analysed in the 2 experienced locoregional progression or died, and
radiotherapy-correct and 214 in the per-protocol median locoregional progression-free survival was 1791
populations (figure 1). The characteristics of the patients and 377 days, respectively (p=0·001, figure 2). Similarly,
in the intention-to-treat population were similar in the in stratum 3, 35 placebo and 39 epoetin-␤ patients had
two treatment groups at baseline, with the exception of a locoregional tumour progression or died, and the Kaplan-
higher proportion in the epoetin-␤ group of smokers and Meier analysis showed a favourable outcome for placebo
of patients with relapsed cancer (table 1). (207 vs 141 days, p=0·006; figure 2).
A mean of 63·1 Gy (SD 9·7, range 0–74) was Multivariate analysis, including treatment stratum,
administered to placebo patients in 43·3 days (9·1, 0–57) tumour stage, baseline smoking, and relapse status,
and 62 Gy (10·8, 0–74) to epoetin-␤ patients in tumour site, haemoglobin concentration, transferrin
42·5 days (9·6, 0–56). Radiotherapy quality assessment saturation, and days between start of drug administration
was done for 333 patients. Of these, 243 (73%) patients and radiotherapy supported the finding that epoetin-
had correctly classified resection status and tumour-node- ␤ treatment is associated with unfavourable outcome
metastases status, and were treated according to protocol. (relative risk 1·26 [95% CI 0·93–1·7], p=0·13).
There were 23 (14%) minor and 13 (8%) major protocol In univariate analysis, haemoglobin concentration at
treatment violations in the placebo group and 25 (15%) baseline correlated with locoregional progression-free
and 17 (10%), respectively, in the epoetin-␤ group. survival (0·77 [0·69–0·87], p<0·0001). Time-adjusted
Mean haemoglobin concentrations increased with haemoglobin concentrations (area under curve) during
epoetin-␤ treatment for up to 6 weeks and stayed stable radiotherapy was also significant (0·85 [0·78–0·93],
thereafter. Mean values after 4 weeks of treatment were p=0·0002).
124 g/L (SD 13) for placebo and 148 g/L (18) for epoetin-
␤ patients. After 9 weeks, the values were 129 g/L (19) or
Relative risk (95% CI) p
154 g/L (17). 148 (82%) epoetin-␤ patients achieved
haemoglobin target values during radiotherapy compared Population and outcome
with 26 (15%) placebo patients. Intention to treat
Locoregional progression-free survival 1·62 (1·22–2·14) 0·0008
208 (59%) patients of the 351 intention-to-treat Locoregional progression 1·69 (1·16–2·47) 0·007
population experienced locoregional tumour progression Survival 1·39 (1·05–1·84) 0·02
or died during follow-up—92 in the placebo and 116 in Radiotherapy correct
the epoetin-␤ group. 79 and 64 patients, respectively, Locoregional progression-free survival 1·42 (1·01–2·01) 0·04
were censored. The stage-adjusted and stratum-adjusted Locoregional progression 1·38 (0·88–2·14) 0·15
Survival 1·22 (0·86–1·73) 0·26
relative risk for locoregional progression-free survival was
Per protocol
1·62 for epoetin ␤ (95% CI 1·22–2·14, p=0·0008; Locoregional progression-free survival 1·35 (0·94–1·95) 0·11
table 2), and the corresponding Kaplan-Meier estimate Locoregional progression 1·41 (0·87–2·27) 0·16
showed a median locoregional progression-free survival of Survival 1·13 (0·78–1·64) 0·52
745 days for placebo compared with 406 days for Cox’s proportional hazards analyses adjusted for stratum and American Joint
epoetin ␤ (p=0·04, figure 2). In the radiotherapy-correct Committee on Cancer stage.
population, locoregional tumour progression or death Table 2: Effect of epoetin-␤ treatment on study endpoints

THE LANCET • Vol 362 • October 18, 2003 • www.thelancet.com 1257

For personal use. Only reproduce with permission from The Lancet.
 ARTICLES

All patients Stratum 1

100

80 p=0·9

p=0·04 Placebo
60
Probability of locoregional progression-free survival (%)

Placebo
40 Epoetin ␤

20 Epoetin ␤

0
Numbers at risk
Placebo 171 104 69 32 18 4 94 68 46 18 12 2
Epoetin ␤ 180 94 54 24 15 6 102 73 45 20 14 6

Stratum 2 Stratum 3
100

80 p=0·001
Placebo
60
p=0·006
40
Placebo
20
Epoetin ␤
Epoetin ␤
0
0 12 24 36 48 60 72 0 12 24 36 48 60 72
Time (months)
Numbers at risk
Placebo 38 27 18 11 4 1 39 9 5 3 2 1
Epoetin ␤ 39 19 9 4 1 0 39 2 0 0 0 0

Figure 2: Locoregional progression-free survival

Ticks represent censored patients.

In subgroup analyses, epoetin ␤ was related to a from cardiac disorders, and one placebo and nine epoetin-
significant poor outcome only among patients younger ␤ patients from general disorders.
than 60 years, in patients in whom haemoglobin at Adverse events were judged to be study-drug related in
baseline was higher than 110 g/L, and among patients who ten (6%) placebo and 15 (8%) epoetin-␤ treated patients.
had advanced disease or cancer of the hypopharynx. Six (4%) or 12 (7%) of the respective patients had
Notably, among patients with cancer of the hypopharynx, disorders of the blood and the lymphatic system thought to
lower proportions of placebo-treated patients than epoetin- be drug related. On review, brain-stem infarction, calf-vein
␤ patients had certain unfavourable baseline characteristcs thrombosis, and acute larynx oedema, in one patient each,
(men 86 vs 90%; smoker 40 vs 55%; relapse at baseline were deemed potentially related to the study treatment.
7 vs 15%; stage IV disease 70 vs 85%). The larynx oedema was thought to be associated with
Cancer-related adverse events occurred in 78 (46%) intravenous iron. Further side-effects of iron treatment
placebo patients and in 92 (51%) epoetin-␤ patients, were not reported.
and local tumour progression was reported in 50 (29%)
and 65 (36%), respectively. The rate of distant Discussion
metastases was similar in the two groups (23% vs 25%). Despite a reliable rise in haemoglobin concentrations, we
Non-cancer-related adverse events were documented in saw no benefit for locoregional progression-free survival,
111 (65%) placebo patients and 123 (68%) epoetin-␤ locoregional progression, or survival. On the contrary,
patients, and comprised general disorders (25% vs patients given placebo fared significantly better than those
30%), skin disorders (22% vs 24%), infections (20% vs given epoetin ␤. A contribution of study design or conduct
21%), disorders of the blood and lymphatic system to this unexpected finding is unlikely. Centre performance,
(8% vs 13%), respiratory, thoracic, and mediastinal- data collection, validation, and processing followed good
system disorders (11% vs 6%), and vascular disorders clinical practice guidelines, and adherence to study-drug
(5% vs 11%). Vascular disorders were hypertension, administration and to radiotherapy were ascertained.
haemorrhage, venous thrombosis and pulmonary Furthermore, results of the intention-to-treat analysis were
embolism, and cerebrovascular disorders. partly confirmed in the radiotherapy-correct population
Overall, 89 (52%) patients in the placebo and and in a separate analysis from the largest recruiting centre
109 (61%) in the epoetin-␤ group died. 119 (34%) (data not shown).
patients in the two treatment groups died from cancer. Overall, patients’ baseline characteristics were balanced
Mortality differed between groups for cardiac and general and demographic data and tumour and treatment features
disorders: five placebo and ten epoetin-␤ patients died of our patients compared well with most published reports,

1258 THE LANCET • Vol 362 • October 18, 2003 • www.thelancet.com

For personal use. Only reproduce with permission from The Lancet.
 ARTICLES

as did prognostic factors.14–17 Haemoglobin concentration at Investigators at participating centres

baseline and during radiotherapy significantly correlated A Hackl, Strahlentherapie, Landeskrankenhaus, Graz, Austria; G Stüben,
Strahlenklinik, Universitätsklinikum, Essen, Germany; A Monnier,
with outcome. Thus, adequate patients were selected for Département d’Oncologie, Centre Hospitalier André Boulloche,
this study, although we could not confirm a larger Montbéliard, France; H M Seegenschmiedt, Klinik für Radioonkologie,
prognostic significance of haemoglobin concentrations Strahlentherapie und Nuklearmedizin, Alfried-Krupp-Krankenhaus,
towards the end of radiotherapy.18 Essen Germany; I A Adamietz, Klinik für Strahlentherapie,
Epoetin ␤ affected various populations differently. This Marienhospital, Ruhr-Universität-Bochum, Herne, Germany; O Pradier,
Klinik und Poliklinik für Strahlentherapie und Radioonkologie, Georg-
heterogeneity raises issues in the interpretation of the study August-Universität, Göttingen, Germany; C Rübe, Strahlentherapie,
results, and it is unclear whether the overall result is Radiologische Universitäts-Klinik, Homburg, Germany; Th Wendt,
applicable to all patients undergoing radiotherapy for head Strahlentherapie, Klinik für Radiologie Friedrich-Schiller-Universität,
Jena, Germany; A C Voss, Strahlenklinik, Zentralklinikum, Augsburg,
and neck cancer. In particular, the subgroup of patients Germany; R Souchon, Strahlenklinik, Allgemeines Krankenhaus, Hagen,
irradiated for manifest cancer (strata 2 and 3) fared worse Germany; J Karstens, Strahlentherapie, Medizinische Hochschule
when given epoetin ␤ than when given placebo. Subgroup- Hannover, Germany; M Alfonsi, Clinique Sainte Catherine, Avignon,
specific differences of the haemoglobin effect19 might explain France; J Dunst, Klinik für Strahlentherapie, Martin-Luther-Universität
this observation. Furthermore, epoetin ␤ had a particular Halle, Wittenberg, Germany; G Calais, Clinique d’Oncologie et de
Radiothérapie, Tours, France; G Grabenbauer, Klinik und Poliklinik für
negative impact on outcome of patients with cancer of the Strahlentherapie, Universitätsklinik Erlangen, Germany; R Fietkau, Klinik
hypopharynx, but imbalances of baseline characteristics of und Poliklinik für Strahlentherapie, Zentrum für Radiologie, Universität,
these patients might be the underlying cause. Rostock, Germany; W Haase, Klinik für Strahlentherapie und
Radiologische Onkologie, St Vincentius-Kliniken, Karlsruhe, Germany;
Although baseline imbalances of particular subgroups L Ahlemann, Strahlentherapie und Radioonkologie, Kreiskrankenhaus,
might contribute to the negative impact of epoetin ␤ on Lüdenscheid, Germany; R Greiner, Klinik für Radioonkologie, Inselspital,
the outcomes in our patients, underlying biological Universität, Bern, Switzerland; H-D Böttcher, Klinik für Strahlentherapie,
phenomena are also a possibility. Originally, the Klinikum der J W-Goethe-Universität, Frankfurt, Germany; B Schilcher,
Klinik für Strahlentherapie und Radio-Onkologie, Universität Witten-
haemoglobin effect was thought to directly alter cancer Herdecke, Wuppertal, Germany; B Müller, Klinik für Strahlentherapie
treatment, particularly radiotherapy. Low haemoglobin und Radioonkologie, Universitäts-Klinikum, Münster, Germany;
concentrations reduce tumour oxygenation,20 amplify S Bartelt, K Fischer, F Momm, Abteilung Strahlenheilkunde,
tumour hypoxia,21 and might decrease, via the oxygen Radiologische Universitätsklinik, Freiburg, Germany.
Study design—W Franke, R Guttenberger, D Messinger.
effect,22 radiosensitivity. Statistical analysis—F Gilberg.
Conversely, erythropoietin activates potent antiapop- External interim analysis—M Baumann, Th Hermann, L Edler.
totic pathways that promote erythropoiesis23,24 and protect External radiotherapy quality assurance—F Kamprad.
from damage in non-haemopoietic cells.25,26 Furthermore,
breast-cancer cells express erythropoietin receptors27 that Conflict of interest statement
are functional,28 and there is increasing evidence that M Henke, K D Haase, K T Beer, and H Frommhold have received
consulting fees. B Schilcher has received travel expenses. U Burger
tumour cells use the erythropoietin system for growth and and C Dougherty are employees at F Hoffmann-La Roche.
angiogenesis.29,30 Thus, antiapoptotic mechanisms
activated by endogenous, anaemia-released erythropoietin Acknowledgments
could also explain the haemoglobin effect. This scenario This study was supported financially, in management, and scientifically by
clearly is not restricted to radiotherapy and may account F Hoffmann-La Roche. We thank the staff who took care of our patients’
for unfavourable clinical results in anaemic patients after needs, and who were involved in gathering, documenting, verifying,
forwarding, and processing the clinical data.
surgery31 or chemotherapy,32 and eventually correspond
with the observations of this study.
Most previous erythropoietin studies focused on quality References
of life in palliative cancer treatment.33,34 Only two studies 1 Grau C, Overgaard J. Significance of hemoglobin concentration for
report improved survival on erythropoietin treatment.10,35 treatment outcome. New York: Springer, 2000.
One, however, is missing important baseline and 2 Henke M, Momm F, Guttenberger R. Erythropoietin for patients
treatment characteristics35 and the other is constrained by undergoing radiotherapy: the Freiburg experience. Stuttgart:
a retrospective control group.10 In another trial, impaired Wissenschaftliche Verlagsgesellschaft, 1999.
3 Oster W, Herrmann F, Gamm H, et al. Erythropoietin for the
survival was reported when erythropoietin was given to treatment of anemia of malignancy associated with neoplastic bone
patients treated for metastatic breast cancer, but marrow infiltration. J Clin Oncol 1990; 8: 956–62.
conclusions are limited by pitfalls in the methods.36 Our 4 Ludwig H, Fritz E, Kotzmann H, Hocker P, Gisslinger H, Barnas U.
data describe decreased tumour control and survival rates Erythropoietin treatment of anemia associated with multiple myeloma.
in erythropoietin-treated radiotherapy patients. The study N Engl J Med 1990; 322: 1693–99.
5 Lavey RS, Dempsey WH. Erythropoietin increases hemoglobin in
is well controlled and was designed to investigate cancer cancer patients during radiation therapy. Int J Radiat Oncol Biol Phys
outcome. Thus, given our, and other, results the 1993; 27: 1147–52.
contribution of erythropoietin to curative cancer treatment 6 Abels RI. Recombinant human erythropoietin in the treatment of the
is at present questionable. anaemia of cancer. Acta Haematol 1992; 87 (suppl 1): 4–11.
Although epoetin ␤ efficiently corrects anaemia among 7 Case DC Jr, Bukowski RM, Carey RW, et al. Recombinant human
erythropoietin therapy for anemic cancer patients on combination
patients undergoing curative radiotherapy, it is not chemotherapy. J Natl Cancer Inst 1993; 85: 801–06.
associated with improved cancer control or survival. On 8 Thews O, Koenig R, Kelleher DK, Kutzner J, Vaupel P. Enhanced
the contrary, erythropoietin might impair disease control radiosensitivity in experimental tumours following erythropoietin
when manifest cancer is irradiated. Future erythropoietin treatment of chemotherapy-induced anaemia. Br J Cancer 1998; 78:
trials should thus carefully analyse cancer control and 752–56.
9 Silver DF, Piver MS. Effects of recombinant human erythropoietin
survival, and investigations on the underlying mechanism on the antitumor effect of cisplatin in SCID mice bearing human
of the clinically relevant haemoglobin effect should be ovarian cancer: a possible oxygen effect. Gynecol Oncol 1999; 73:
reinforced. 280–84.
10 Glaser CM, Millesi W, Kornek GV, et al. Impact of hemoglobin
Contributors level and use of recombinant erythropoietin on efficacy of
All researchers contributed to the study design, implementation of the preoperative chemoradiation therapy for squamous cell carcinoma of
study, and collection of the data. U Burger analysed data and M Henke the oral cavity and oropharynx. Int J Radiat Oncol Biol Phys 2001; 50:
undertook plausibility testing. M Henke drafted the report. All researchers 705–15.
took part in the critical revision of the paper and approved the final version. 11 Littlewood TJ. The impact of hemoglobin levels on treatment

THE LANCET • Vol 362 • October 18, 2003 • www.thelancet.com 1259

For personal use. Only reproduce with permission from The Lancet.
 ARTICLES

outcomes in patients with cancer. Semin Oncol 2001; 28 (suppl 8): 25 Brines ML, Ghezzi P, Keenan S, et al. Erythropoietin crosses the
49–53. blood-brain barrier to protect against experimental brain injury.
12 Cox DR. Regression models and life table. J R Stat Soc B 1972; 34: Proc Natl Acad Sci USA 2000; 97: 10526–31.
187–220. 26 Marti HH, Gassmann M, Wenger RH, et al. Detection of
13 Kaplan EL, Meier P. Nonparametric estimation from incomplete erythropoietin in human liquor: intrinsic erythropoietin production
observation. J Am Stat Assoc 1958; 54: 457–81. in the brain. Kidney Int 1997; 51: 416–18.
14 Yu L, Vikram B, Malamud S, Yudelman I, Nussbaum M, Beattie E. 27 Acs G, Acs P, Beckwith SM, et al. Erythropoietin and erythropoietin
Chemotherapy rapidly alternating with twice-a-day accelerated receptor expression in human cancer. Cancer Res 2001; 61:
radiation therapy in carcinomas involving the hypopharynx or 3561–65.
esophagus: an update. Cancer Invest 1995; 13: 567–72. 28 Arcasoy MO, Amin K, Karayal AF, et al. Functional significance of
15 Overgaard J, Hansen HS, Jorgensen K, Hjelm Hansen M. Primary erythropoietin receptor expression in breast cancer. Lab Invest 2002;
radiotherapy of larynx and pharynx carcinoma—an analysis of some 82: 911–18.
factors influencing local control and survival. Int J Radiat Oncol Biol 29 Miller CB, Platanias LC, Mills SR, et al. Phase I-II trial of
Phys 1986; 12: 515–21. erythropoietin in the treatment of cisplatin-associated anemia.
16 Brizel DM, Albers ME, Fisher SR, et al. Hyperfractionated irradiation J Natl Cancer Inst 1992; 84: 98–103.
with or without concurrent chemotherapy for locally advanced head 30 Yasuda Y, Fujita Y, Matsuo T, et al. Erythropoietin regulates
and neck cancer. N Engl J Med 1998; 338: 1798–804. tumour growth of human malignancies. Carcinogenesis 2003; 24:
1021–29.
17 Frommhold H, Guttenberger R, Henke M. The impact of blood
hemoglobin content on the outcome of radiotherapy: the Freiburg 31 Lutterbach J, Guttenberger R. Anemia is associated with decreased
experience. Strahlenther Onkol 1998; 174 (suppl 4): 31–34. local control of surgically treated squamous cell carcinomas of the
glottic larynx. Int J Radiat Oncol Biol Phys 2000; 48: 1345–50.
18 Vigario G, Kurohara SS, George FW 3rd. Association of hemoglobin
levels before and during radiotherapy with prognosis in uterine cervix 32 Hasenclever D, Diehl V. A prognostic score for advanced Hodgkin’s
cancer. Radiology 1973; 106: 649–52. disease. International Prognostic Factors Project on Advanced
Hodgkin’s Disease. N Engl J Med 1998; 339: 1506–14.
19 Guttenberger R, Lutterbach J, Roth A, et al. Hemoglobin levels
33 Glaspy J, Bukowski R, Steinberg D, Taylor C, Tchekmedyian S,
predict local regional control after postoperative radiotherapy for
Vadhan-Raj S. Impact of therapy with epoetin alfa on clinical
advanced head and neck cancer. Eur J Cancer 1997; 33 (suppl):
outcomes in patients with nonmyeloid malignancies during cancer
102–03.
chemotherapy in community oncology practice: Procrit Study
20 Becker A, Stadler P, Lavey RS, et al. Severe anemia is associated with Group. J Clin Oncol 1997; 15: 1218–34.
poor tumor oxygenation in head and neck squamous cell carcinomas.
34 Demetri GD, Kris M, Wade J, Degos L, Cella D. Quality-of-life
Int J Radiat Oncol Biol Phys 2000; 46: 459–66.
benefit in chemotherapy patients treated with epoetin alfa is
21 Vaupel P, Schlenger K, Knoop C, Hockel M. Oxygenation of human independent of disease response or tumor type: results from a
tumors: evaluation of tissue oxygen distribution in breast cancers by prospective community oncology study: Procrit Study Group.
computerized O2 tension measurements. Cancer Res 1991; 51: J Clin Oncol 1998; 16: 3412–25.
3316–22. 35 Littlewood TJ, Bajetta E, Nortier JW, Vercammen E, Rapoport B.
22 Gray LH, Conger AD, Ebert M. The concentration of oxygen Effects of epoetin alfa on hematologic parameters and quality of life
dissolved in tissues at the time of irradiation as a factor in radiotherapy. in cancer patients receiving nonplatinum chemotherapy: results of a
Br J Radiol 1953; 26: 638–48. randomized, double-blind, placebo-controlled trial. J Clin Oncol
23 Krantz SB. Erythropoietin. Blood 1991; 77: 419–34. 2001; 19: 2865–74.
24 Koury MJ, Bondurant MC. The molecular mechanism of 36 Leyland-Jones B. Breast cancer trial with erythropoietin terminated
erythropoietin action. Eur J Biochem 1992; 210: 649–63. unexpectedly. Lancet Oncol 2003; 4: 459–60.

1260 THE LANCET • Vol 362 • October 18, 2003 • www.thelancet.com

For personal use. Only reproduce with permission from The Lancet.