Journal of the National Cancer Institute, Vol. 98, No. 3, February 1, 2006 BRIEF COMMUNICATION 215
Acute Myeloid Leukemia
Lymphoma: A Population-
Based Study of 35 511 Patients
Sara J. Schonfeld , Ethel S. Gilbert ,
Graça M. Dores , Charles F. Lynch ,
David C. Hodgson , Per Hall ,
Hans Storm , Aage Andersen , Eero
Pukkala , Eric Holowaty , Magnus
Kaijser , Michael Andersson , Heikki
Joensuu , Sophie D. Fosså , James M.
Allan , Lois B. Travis
Treatments for Hodgkin lymphoma
are associated with large relative risks
of acute myeloid leukemia (AML), but
there are few estimates of the excess
absolute risk (EAR), a useful measure
of disease burden. One-year Hodgkin
lymphoma survivors (N = 35 511) were
identifi ed within 14 population-based
cancer registries in Nordic countries
and North America from January 1,
1970, through December 31, 2001.
We used Poisson regression analysis
to model the EAR of AML, per 10 000
person-years. A total of 217 Hodgkin
lymphoma survivors were diagnosed
with AML (10.8 expected; unadjusted
EAR = 6.2; 95% confi dence inter val =
5.4 to 7.1). Excess absolute risk for
AML was highest during the fi rst 10
years after Hodgkin lymphoma diag-
nosis but remained elevated thereaf-
ter. In subsequent analyses, adjusted
for time since Hodgkin lymphoma
diagnosis and presented for the 5 – 9
year interval, the EAR was statisti-
cally sig ni fi cantly ( P <.001) larger in
patients diagnosed with Hodgkin lym-
phoma at age 35 years and older than
in those diagnosed before 35 years of
age. The EAR of AML declined sta-
tistically signifi cantly after 1984 (7.0
to 4.2 and 16.4 to 9.9 in the <35 and
≥ 35 age groups, respectively), which
may be associated with modifi cations
in chemotherapy. [J Natl Cancer Inst
2006;98:215 – 8]
Treatments for Hodgkin lymphoma
have resulted in excellent survival rates
but are associated with large, increased
relative risks of acute myeloid leukemia
(AML) ( 1 – 8 ) . However, few estimates
of the excess absolute risk (EAR), a use-
ful measure of disease burden in a popu-
lation ( 9 ) , exist; most ( 1 , 4 , 5 , 7 ) , but not
all studies ( 10 ) , have been based on
relatively small numbers of AML. No
large, population-based investigation has
estimated the EAR of AML following
Hodgkin lymphoma, simultaneously
evaluating the effects of age, calendar
year of Hodgkin lymphoma diagnosis,
time since Hodgkin lymphoma diagno-
sis, and initial treatment. It is of par-
ticular interest to determine whether
applications of newer treatments are
refl ected in a decreased absolute risk of
AML among Hodgkin lymphoma survi-
vors in the general population.
One-year survivors (N = 35 511) of
Hodgkin lymphoma who were diagnosed
between January 1, 1970, and December
31, 2001, were identifi ed within 14
population-based cancer registries in
Denmark, Finland, Norway, Ontario,
Sweden, and the National Cancer Insti-
tute’s Surveillance, Epidemiology and
End Results (SEER) Program. Study
dates varied slightly according to regis-
try: Ontario, 1970 – 2000; Denmark,
1970 – 1998; Finland and Sweden, 1970 –
2001; Norway, 1970 – 1999; and the
SEER Program, 1973 – 1999 (Connecticut
[1973+], Hawaii [1973+], Iowa [1973+],
New Mexico [1973+], and Utah [1973+],
and the metropolitan areas of San
Francisco – Oakland [1973+], Detroit
[1973+], Seattle – Puget Sound [1974+],
and Atlanta [1975+]). Analyses were
restricted to 1-year survivors to permit a
suffi cient latent period for the develop-
ment of therapy-related leukemias ( 11 ) .
All registries collect demographic infor-
mation and vital status at last follow-up.
Except Sweden and Ontario, all regis-
tries also record the type of initial cancer
treatment, which we categorized as
radiotherapy alone (RT), chemotherapy
alone (CT), radiotherapy and chemo-
therapy (RT + CT) or other/unknown.
Detailed data with regard to initial treat-
ment regimens (e.g., cytotoxic drug
names and doses) were not available
from the registry fi les, nor was informa-
tion on subsequent therapy. Patients
diagnosed with AML at least 1 year after
Hodgkin lymphoma were identifi ed
through a search of cancer registry inci-
dence fi les. Population-based incidence
data for acute leukemia, not otherwise
specifi ed (AL – NOS) show age-specifi c
trends that are strikingly similar to those
of AML ( 12 ) . Thus, we grouped the 15
patients with AL – NOS with the 202
patients with AML ( 13 ) , and for simplic-
ity, later refer to the category as AML.
Small numbers (n = 13) of acute lym-
phoblastic leukemia (ALL) precluded
analysis of this disease category.
Person-years (PY) and AML patients
were categorized by calendar year of
Hodgkin lymphoma diagnosis (1970 –
1984, 1985 – 2001), initial treatment, reg-
istry, and by 5-year intervals of attained
age, calendar year, time since Hodgkin
lymphoma diagnosis, and age at Hodg-
kin lymphoma diagnosis. Accumulated
person-years at risk for each category
defi ned by registry, sex, and 5-year age
and calendar year intervals were multi-
plied by the corresponding AML inci-
dence rates to calculate the number of
expected cases in each category. O i , E i ,
and PY i , denote observed cases, expected
cases, and person-years in a specifi c cat-
egory i , respectively.
The general Poisson regression meth-
ods used in this study have been described
previously ( 14 ) and were implemented
Affi liations of authors: Division of Cancer Epi-
demiology and Genetics (SJS, ESG, LBT), Divi-
sion of Cancer Prevention (GMD), National Cancer
Institute, National Institutes of Health, Department
of Health and Human Services, Bethesda, MD; The
University of Iowa, Iowa City (CFL); Department
of Radiation Oncology, University of Toronto,
Princess Margaret Hospital, Toronto, Ontario,
Canada (DCH); Karolinska Institutet, Stockholm,
Sweden (PH, MK); Danish Cancer Society,
Copenhagen, Denmark (HS, MA); Cancer Registry
of Norway, Oslo (AA); Finnish Cancer Registry,
Helsinki, Finland (EP); Cancer Care Ontario,
Toronto, Ontario, Canada (EH); Helsinki Univer-
sity Central Hospital, Helsinki, Finland (HJ);
Radiumhospitalet Trust, Oslo, Norway (SDF); The
University of York, United Kingdom (JMA).
Correspondence to: Lois B. Travis, MD, ScD,
Division of Cancer Epidemiology and Genet-
ics, National Cancer Institute, NIH, DHHS, 6120
Executive Blvd., EPS # 7086, MSC 7238, Bethesda,
MD 20892-7238 (e-mail: email@example.com ).
See “ Notes ” following “ References. ”
© The Author 2006. Published by Oxford University
Press. All rights reserved. For Permissions, please
by guest on February 19, 2013
216 BRIEF COMMUNICATION Journal of the National Cancer Institute, Vol. 98, No. 3, February 1, 2006
with the AMFIT module of the software
package EPICURE ( 15 ) . An objective
was to obtain estimates of the EAR and
confi dence intervals (CIs) for categories
defi ned by age at Hodgkin lymphoma
diagnosis (designated by a; a = 1 for
age < 35 years; 0 otherwise) and calendar
year of diagnosis (designated by c; c = 1
for 1985+; 0 otherwise) with adjustment
for time since diagnosis (designated by
t and expressed in years). The adjust-
ment for t was important for evaluating
calendar year effects given the shorter
follow-up for Hodgkin lymphoma patients
diagnosed after 1984. Another objective
was to describe the pattern of risk over
time. Results shown in Table 1 and
Fig. 1 were based on a model in which
the statistical expectation of O a,c,t was
assumed to be
E a,c,t + PY a,c,t EAR (a, c, t)
with the logarithm of EAR ( a, c, t )
expressed as a linear function of a, e, t,
and t 2 . The EAR is expressed per 10 000
person-years. Analyses using continuous
variables were based on midpoints of
5-year intervals. This model was chosen
because it was the simplest model pro-
viding an adequate fi t to the data. Spe-
cifi cally, expression of the EAR using
age at Hodgkin lymphoma diagnosis as a
categorical variable (with two catego-
ries) fi tted the data better than expression
of the EAR as a continuous function
based on fi ner categories of age at Hodg-
kin lymphoma diagnosis, and a linear –
quadratic function for time since Hodgkin
lymphoma diagnosis fi tted the data sta-
tistically signifi cantly better than a linear
model ( P = .003). Because the EAR
depended strongly on t , it was necessary
to present results in Table 1 for a specifi c
period, chosen to be the 5 – 9 year interval
( t = 7.5), although all data contribute to
the analyses. Analyses that allowed
parameters to depend on initial treatment
(RT only or with any chemotherapy [CT
or RT + CT]) were also conducted. Cal-
endar year of Hodgkin lymphoma diag-
nosis served as a surrogate for treatment
era: mechlorethamine, vincristine, pro-
carbazine, and prednisone (MOPP) ( 16 )
was the primary chemotherapy regimen
used between 1970 and 1984, whereas
other regimens gained more widespread
use in later years ( 17 – 21 ) . Sex was not
included in the model because there was
little evidence that any of the parameters
depended on sex. All statistical tests
were two-sided, and P <.05 was consid-
ered statistically signifi cant.
Compared with 10.8 expected, 217
cases of AML were diagnosed (overall,
unadjusted EAR = 6.2, 95% confi dence
interval [CI] = 5.4 to 7.1; observed/
expected = 20.2, 95% CI = 17.6 to 23.0).
During the 1 – 4, 5 – 9, 10 – 14, and 15+ year
periods following Hodgkin lymphoma
diagnosis, 92, 87, 29, and nine cases
occurred (median = 5.9 years; range = 1
to 33 years). The EAR of AML was high-
est during the fi rst 10 years after Hodgkin
lymphoma diagnosis (EAR = 7.9, 95%
CI = 6.8 to 9.3); risks were comparable
during the 1 – 4 and 5 – 9 year periods.
During the 10 – 14 year period, the EAR
was lower but remained statistically
Fig. 1. Excess absolute risk of acute myeloid leukemia per 10 000 person-years (including 15 patients with
acute leukemia, not otherwise specifi ed, as described in the text) by time since diagnosis of Hodgkin lym-
phoma. Excess absolute risk after 14 years is not available for patients diagnosed with Hodgkin lymphoma
between 1985 and 2001. Age (years) and calendar year of Hodgkin lymphoma diagnosis ( closed circles ,
≥ 35 years, 1970 – 1984; open circles , ≥ 35 years, 1985 – 2001; closed triangles , <35 years, 1970 – 1984;
open triangles , <35 years, 1985 – 2001).
1-4 y5-9 y 10-14 y 15+ y
Time since diagnosis of Hodgkin lymphoma
EAR per 10,000 person-years
Table 1. Excess absolute risk (EAR) * of acute myeloid leukemia (AML) † following Hodgkin lymphoma
Calendar year of
All patients (N = 35 511)
Any chemotherapy (n = 12 196) ‡ §
Radiotherapy only (n = 8241) §
No. AML EAR (95% CI) No. AML EAR (95% CI)No. AML EAR (95% CI)
All patients ||
Age at Hodgkin
lymphoma diagnosis, y
<35 (n = 19 742)
N/A 2178.3 (6.8 to 10.1) 8311.3 (8.5 to 14.7) 40 5.4 (3.7 to 7.7)
1970 – 1984
1985 – 2001
1970 – 1984
1985 – 2001
7.0 (5.2 to 9.1)
4.2 (3.0 to 5.8)
16.4 (12.5 to 21.0)
9.9 (7.2 to 13.4)
10.7 (6.9 to 15.9)
4.8 (2.9 to 7.6)
27.6 (18.7 to 39.5)
12.4 (7.9 to 18.8)
3.2 (1.6 to 5.7)
1.6 (0.6 to 3.6)
14.3 (8.7 to 22.1)
7.0 (2.9 to 13.8)
≥ 35 (n = 15 769)
* Based on the model shown in the text. EAR = excess number of cases per 10 000 person-years. Risk estimates presented for 5- to 9-year period after diagnosis
of Hodgkin lymphoma. CI = confi dence interval; N/A = not applicable.
† The category of AML includes 15 patients with acute leukemia, not otherwise specifi ed, as described in the text.
‡ Any chemotherapy = chemotherapy alone or chemotherapy given with radiotherapy.
§ Numbers include only those patients reported to registries that collect data on initial course of cancer treatment (the National Cancer Institute’s Surveillance,
Epidemiology, and End Results Program and cancer registries in Denmark, Finland, and Norway).
|| Estimates not adjusted for age or calendar year of Hodgkin lymphoma diagnosis.
by guest on February 19, 2013
Journal of the National Cancer Institute, Vol. 98, No. 3, February 1, 2006 BRIEF COMMUNICATION 217
signifi cantly elevated compared with the
general population (EAR = 4.6, 95%
CI = 3.0 to 6.5; P <.001). For patients
diagnosed with Hodgkin lymphoma
before 1984, the EAR was slightly ele-
vated after 15 years (EAR = 1.3, 95%
CI = 0.4 to 2.6). Subsequent results are
adjusted for time since Hodgkin lym-
phoma diagnosis and presented for the
5 – 9 year period.
Although risks were elevated for all
groups, patients whose initial treatment
included any chemotherapy, compared
with RT alone, had a statistically signifi -
cantly larger EAR of AML (EAR = 11.3,
95% CI = 8.5 to 14.7 and EAR = 5.4,
95% CI = 3.7 to 7.7, respectively;
P <.001; Table 1 ). Excess absolute risks
were higher for older patients ( ≥ 35 years
versus <35 years) and for the earlier
period (1970 – 1984 versus 1985 – 2001).
The decline over calendar year, as mea-
sured by the difference in the EARs, was
particularly apparent among patients
who initially received any chemotherapy.
Patterns of risk by age and calendar year
of Hodgkin lymphoma diagnosis for the
entire period of follow-up are summa-
rized in Fig. 1 .
To our knowledge, this is the fi rst
large, international population – based
study of Hodgkin lymphoma patients to
demonstrate an overall
although not elimination, in the burden
of AML over calendar year time. This
decline likely refl ects in part changes in
chemotherapy (given initially or at
relapse) that were implemented over the
last few decades. Our results are consis-
tent with reports of lower risks of
AML among patients treated with doxo-
rubicin, bleomycin, vinblastine, and
dacarbazine (ABVD) compared with
MOPP ( 21 – 23 ) . The positive association
between EAR and age at Hodgkin lym-
phoma diagnosis may refl ect in part the
increase in population baseline risks for
AML with increasing age. Because most
of the excess risk is concentrated in
the 10-year period following Hodgkin
lymphoma diagnosis, age at Hodgkin
lymphoma diagnosis and age at risk of
AML are similar; it is therefore diffi cult
to separate the individual effects of these
Reports based on detailed treatment
data suggest that AML risk is no longer
elevated 10 or more years after the ter-
mination of chemotherapy for Hodgkin
lymphoma ( 2 , 24 , 25 ) , although several
reports may not have had adequate
statistical power to evaluate risk in long-
term survivors ( 24 , 25 ) . The persistence
of increased EAR for over one decade in
our series may refl ect the effect of subse-
quent chemotherapy (e.g., at relapse).
Potential limitations of our study
include those common to registry-based
studies, such as incomplete treatment
data and lack of detailed therapy infor-
mation. Further, our estimates of EAR
likely represent a minimal gauge of risk,
because underreporting of secondary
leukemia ( 26 , 27 ) has been observed in
population-based cancer registries. It
seems unlikely, however, that underre-
porting would be differential by initial
treatment for Hodgkin lymphoma.
Our study also has several strengths,
which include the large cohort size,
inclusion of patients diagnosed with
Hodgkin lymphoma over a 30-year
period, and adjustment for time since
Hodgkin lymphoma diagnosis. Also, our
investigation is population-based.
In conclusion, the risk of AML
following Hodgkin lymphoma has
decreased statistically signifi cantly over
calendar-year time, likely due to modifi -
cations in chemotherapy. Analytic stud-
ies with detailed treatment data are
required to correlate these decreases with
changes in therapy and to better under-
stand the long-term risk of AML after
(1) Tucker MA, Coleman CN, Cox RS, Varghese A,
Rosenberg SA. Risk of second cancers after
treatment for Hodgkin’s disease. N Engl J
Med 1988 ; 318 : 76 – 81.
(2) Kaldor JM, Day NE, Clarke EA, van Leeuwen
FE, Henry-Amar M, Fiorentino MV, et al.
Leukemia following Hodgkin’s disease. N
Engl J Med 1990 ; 322 : 7 – 13.
(3) van Leeuwen FE, Chorus AM, van den
Belt-Dusebout AW, Hagenbeek A, Noyon R,
van Kerkhoff EH, et al. Leukemia risk follow-
ing Hodgkin’s disease: relation to cumulative
dose of alkylating agents, treatment with
teniposide combinations, number of episodes
of chemotherapy, and bone marrow damage.
J Clin Oncol 1994 ; 12 : 1063 – 73.
(4) van Leeuwen FE, Klokman WJ, Hagenbeek A,
Noyon R, van den Belt-Dusebout AW,
van Kerkhoff EH, et al. Second cancer risk
following Hodgkin’s disease: a 20-year follow-
up study. J Clin Oncol 1994 ; 12 : 312 – 25.
(5) Hoppe RT. Hodgkin’s disease: complications
of therapy and excess mortality. Ann Oncol
1997 ; 8 Suppl 1: 115 – 8.
(6) Metayer C, Lynch CF, Clarke EA, Glimelius B,
Storm H, Pukkala E, et al. Second cancers
among long-term survivors of Hodgkin’s dis-
ease diagnosed in childhood and adolescence.
J Clin Oncol 2000 ; 18 : 2435 – 43.
(7) van Leeuwen FE, Klokman WJ, Veer MB,
Hagenbeek A, Krol AD, Vetter UA, et al.
Long-term risk of second malignancy in
survivors of Hodgkin’s disease treated during
adolescence or young adulthood. J Clin Oncol
2000 ; 18 : 487 – 97.
(8) Abrahamsen AF, Andersen A, Nome O,
Jacobsen AB, Holte H, Abrahamsen JF, et al.
Long-term risk of second malignancy after
treatment of Hodgkin’s disease: the infl uence
of treatment, age and follow-up time. Ann
Oncol 2002 ; 13 : 1786 – 91.
(9) Rothman KJ. Modern epidemiology. Boston
(MA): Little, Brown and Company; 1986 .
(10) Dores GM, Metayer C, Curtis RE, Lynch CF,
Clarke EA, Glimelius B, et al. Second
malignant neoplasms among long-term sur-
vivors of Hodgkin’s disease: a population-
based evaluation over 25 years. J Clin Oncol
2002 ; 20 : 3484 – 94.
(11) van Leeuwen FE, Travis LB. Second can-
cers. In: VT Devita, et al., editors. Cancer:
principles and practice of oncology, 7th ed.
Philadelphia (PA): Lippincott, Williams, and
Wilkins; 2005 . p. 2575 – 602.
(12) Surveillance, Epidemiology, and End Results
(SEER) Program. SEER*Stat Database;
Incidence — SEER 9 Regs Public Use, Nov
2004 (1973-2002), National Cancer Institute,
DCCPS, Surveillance Research Program,
Cancer Statistics Branch, released April 2005,
based on the November 2004 submission.
Available at: http://seer.cancer.gov .
(13) Ries LA, Eisner MP, Kosary CL, Hankey
BF, Miller BA, Clegg L, et al. SEER cancer
statistics review. 1975-2000. Bethesda (MD):
National Cancer Institute; 2003 . Available at:
(14) Travis LB, Fossa SD, Schonfeld SJ, McMaster
ML, Lynch CL, Storm H, et al. Second can-
cers among 40,576 testicular cancer patients:
focus on long-term survivors. J Natl Cancer
Inst 2005 ; 97 : 1354 – 65.
(15) Preston DL, Lubin JH, Pierce DA. EPICURE
user’s guide. Seattle (WA): HiroSoft Interna-
tional Corporation; 1991 .
(16) Devita VT Jr, Serpick AA, Carbone PP.
Combination chemotherapy in the treatment
of advanced Hodgkin’s disease. Ann Intern
Med 1970 ; 73 : 881 – 95.
(17) Bonadonna G, Santoro A. ABVD chemo-
therapy in the treatment of Hodgkin’s disease.
Cancer Treat Rev 1982 ; 9 : 21 – 35.
(18) Canellos GP, Anderson JR, Propert KJ, Nissen
N, Cooper MR, Henderson ES, et al. Chemo-
therapy of advanced Hodgkin’s disease with
MOPP, ABVD, or MOPP alternating with
ABVD. N Engl J Med 1992 ; 327 : 1478 – 84.
(19) Connors JM, Noordijk EM, Horning SJ.
Hodgkin’s lymphoma: basing the treatment on
the evidence. Hematology (Am Soc Hematol
Educ Program) 2001 : 178 – 93.
(20) Diehl V, Franklin J, Pfreundschuh M, Lathan B,
Paulus U, Hasenclever D, et al. Standard and
increased-dose BEACOPP chemotherapy
compared with COPP-ABVD for advanced
Hodgkin’s disease. N Engl J Med 2003 ;
348 : 2386 – 95.
by guest on February 19, 2013
218 BRIEF COMMUNICATION Journal of the National Cancer Institute, Vol. 98, No. 3, February 1, 2006 Download full-text
(21) Duggan DB, Petroni GR, Johnson JL, Glick JH,
Fisher RI, Connors JM, et al. Randomized
comparison of ABVD and MOPP/ABV hybrid
for the treatment of advanced Hodgkin’s
disease: report of an intergroup trial. J Clin
Oncol 2003 ; 21 : 607 – 14.
(22) Delwail V, Jais JP, Colonna P, Andrieu JM.
Fifteen-year secondary leukaemia risk ob -
served in 761 patients with Hodgkin’s disease
prospectively treated by MOPP or ABVD
chemotherapy plus high-dose irradiation.
Br J Haematol 2002 ; 118 : 189 – 94.
(23) Bonadonna G, Viviani S, Bonfante V, Gianni
AM, Valagussa P. Survival in Hodgkin’s
disease patients — report of 25 years of expe-
rience at the Milan Cancer Institute. Eur J
Cancer 2005 ; 41 : 998 – 1006.
(24) Blayney DW, Longo DL, Young RC,
Greene MH, Hubbard SM, Postal MG, et al.
Decreasing risk of leukemia with prolonged
follow-up after chemotherapy and radio-
therapy for Hodgkin’s disease. N Engl J Med
1987 ; 316 : 710 – 14.
(25) Pedersen-Bjergaard J, Specht L, Larsen SO,
Ersboll J, Struck J, Hansen MM, et al. Risk
of therapy-related leukaemia and preleu-
kaemia after Hodgkin’s disease. Relation to
age, cumulative dose of alkylating agents,
and time from chemotherapy. Lancet 1987 ; 2 :
83 – 8.
(26) Curtis RE, Hankey BF, Myers MH, Young
JL Jr. Risk of leukemia associated with the
fi rst course of cancer treatment: an analysis
of the Surveillance, Epidemiology, and End
Results Program experience. J Natl Cancer
Inst 1984 ; 72 : 531 – 44.
(27) Astrom M, Bodin L, Tidefelt U. Adjustment
of incidence rates after an estimate of com-
pleteness and accuracy in registration of acute
leukemias in a Swedish population. Leuk
Lymphoma 2001 ; 41 : 559 – 70.
Supported by the Intramural Research Program
of the National Institutes of Health, National Can-
cer Institute, Division of Cancer Epidemiology and
We are indebted to Jeremy Miller, Information
Management Services, Rockville, MD, for expert
computer support and data management and to
Denise Duong, National Cancer Institute, Bethesda,
MD, for typing assistance.
Manuscript received June 29, 2005 ; revised
November 14, 2005 ; accepted November 15, 2005.
by guest on February 19, 2013