Stanulla M, Schaeffeler E, Flohr T, Cario G, Schrauder A, Zimmermann M, Welte K, Ludwig W, Bartram CR, Zanger UM, Eichelbaum M, Schrappe M, Schwab M. Thiopurine Methyltransferase (TPMT) Genotype and Early Treatment Response to Mercaptopurine in Childhood Acute Lymphoblastic Leukemia. JAMA. 2005;293(12):1485-1489. doi:10.1001/jama.293.12.1485
Author Affiliations: Department of Pediatric
Hematology and Oncology, Hannover Medical School, Hannover, Germany (Drs Stanulla,
Cario, Schrauder, Zimmermann, and Welte); Institute of Human Genetics, Ruprecht-Karls
University, Heidelberg, Germany (Drs Flohr and Bartram); Margarete-Fischer-Bosch
Institute of Clinical Pharmacology, Stuttgart, Germany (Drs Schaeffeler, Zanger,
Eichelbaum, and Schwab); Robert-Rössle Clinic, Department of Hematology,
Oncology and Tumor Immunology, HELIOS Clinic Berlin, Berlin, Germany (Dr Ludwig);
and University Children’s Hospital, Kiel, Germany (Dr Schrappe).
Context Early response to multiagent chemotherapy, including mercaptopurine,
as measured by minimal residual disease is an important prognostic factor
for children with acute lymphoblastic leukemia (ALL). Thiopurine methyltransferase
(TPMT) is involved in the metabolism of mercaptopurine and subject to genetic
polymorphism, with heterozygous individuals having intermediate and homozygous
mutant individuals having very low TPMT activity.
Objective To assess the association of TPMT genotype
with minimal residual disease load before and after treatment with mercaptopurine
in the early treatment course of childhood ALL.
Design, Setting, and Patients TPMT genotyping of childhood ALL patients (n = 814)
in Germany consecutively enrolled in the ALL-BFM (Berlin-Frankfurt-Münster)
2000 study from October 1999 to September 2002. Minimal residual disease was
analyzed on treatment days 33 and 78 for risk-adapted treatment stratification.
A 4-week cycle of mercaptopurine was administered between these 2 minimal
residual disease measurements. Patients (n = 4) homozygous for a
mutant TPMT allele, and consequently deficient in
TPMT activity, were treated with reduced doses of mercaptopurine and, therefore,
not included in the analyses.
Main Outcome Measures Minimal residual disease load before (day 33) and after (day 78) mercaptopurine
treatment. Loads smaller than 10−4 were defined as negative.
Results Patients (n = 55) heterozygous for allelic variants of TPMT conferring lower enzyme activity had a significantly
lower rate of minimal residual disease positivity (9.1%) compared with patients
(n = 755) with homozygous wild-type alleles (22.8%) on day 78 (P = .02). This translated into a 2.9-fold reduction
in risk for patients with wild-type heterozygous alleles (relative risk, 0.34;
95% confidence interval, 0.13-0.86).
Conclusions TPMT genotype has a substantial impact on minimal
residual disease after administration of mercaptopurine in the early course
of childhood ALL, most likely through modulation of mercaptopurine dose intensity.
Our findings support a role for minimal residual disease analyses in the assessment
of genotype-phenotype associations in multiagent chemotherapeutic trials.
Contemporary treatment strategies for childhood acute lymphoblastic
leukemia (ALL) are based on essential therapeutic elements that are consecutively
applied over 2 to 3 years and lead to an overall long-term survival of approximately
80%.1- 3 These therapeutic
elements include the induction of remission (<5% leukemic blasts in the
bone marrow) to restore normal hematopoiesis, extracompartment therapy to
treat leukemic cells in the central nervous system and testes, an induction
consolidation and reinduction phase to further intensify treatment to prevent
emergence of a drug-resistant clone, and a maintenance phase for eradication
of residual leukemic cells.
Another common feature in the current clinical management of children
with ALL is the adjustment of therapy intensity according to the risk of treatment
failure conferred by different prognostic factors.1- 3 These
prognostic factors include clinical and biological characteristics that are
assessable at diagnosis (eg, age at diagnosis, presenting white blood cell
count, cytogenetic aberrations of the leukemic clone) as well as a variety
of estimates of early response to treatment. Measures of early response to
treatment were traditionally based on cytomorphologic evaluation of peripheral
blood or bone marrow smears for leukemic cells at specific points during the
first 2 weeks of ALL treatment.1- 3 In
comparison with cytomorphologic evaluation, minimal residual disease analysis
through polymerase chain reaction (PCR)–based detection of leukemic
clone-specific immunoglobulin and T-cell receptor gene rearrangements or by
flow cytometry provides a more sensitive approach to response evaluation.
Recent studies showed that measuring minimal residual disease at times during
induction and consolidation treatment was highly predictive of disease recurrence.4- 10
Since their introduction to leukemia treatment in the 1950s, the thiopurines
mercaptopurine and thioguanine have played an essential role in treatment
protocols for ALL.11,12 Several
contemporary treatment protocols for childhood ALL apply consecutive cycles
of either mercaptopurine or thioguanine starting as early as during induction
consolidation treatment and continue administration during maintenance therapy
for up to 36 months after diagnosis.1- 3 As
prodrugs, thiopurines require bioactivation by a multistep pathway to form
thioguanine nucleotides, which are thought to be the major cytotoxic compounds
through triggering cell cycle arrest and apoptosis.13,14 This
process is in competition with direct inactivation of thiopurines or their
metabolites by thiopurine S-methyltransferase (TPMT).
TPMT is a cytosolic enzyme ubiquitously expressed in the human body and catalyzes
the S-methylation of thiopurines. The TPMT locus is subject to genetic polymorphism, with heterozygous individuals
(6%-11% of white individuals) having intermediate TPMT activity and homozygous
mutant individuals (0.2%-0.6% of white individuals) having very low TPMT activity.13- 17 To
date 20 variant alleles (TPMT*2-*18) have been identified, which are associated with decreased activity
compared with the TPMT*1 wild-type allele. More than
95% of defective TPMT activity can be explained by the most frequent mutant
alleles TPMT*2 and TPMT*3(A-D). In several independent
studies, TPMT genotype showed excellent concordance
with TPMT phenotype.13,14
With regard to treatment outcome in childhood ALL, Lennard and colleagues18 described in 1990 a higher relapse rate in children
with lower thioguanine nucleotide concentrations measured in erythrocytes
and suggested a substantial role for genetically determined TPMT activity
in the predisposition to the cytotoxic effects of mercaptopurine and, consequently,
ALL outcome. Their hypothesis is supported by the work of Relling and colleagues,19 who demonstrated in a study of 182 children with
ALL that mercaptopurine dose intensity was the strongest predictor of outcome.
In that study a tendency toward better event-free survival was described for
children with intermediate and low TPMT activity compared with that of homozygous
wild-type TPMT phenotypes.
Although the prognostic impact of early response to treatment is well
known and thiopurines are applied as early as during induction consolidation
treatment, the impact of TPMT genotype on mercaptopurine-mediated
antileukemic effects in the early course of childhood ALL therapy has not
yet been determined. To derive a better understanding of a potential prognostic
role for TPMT genotype in childhood ALL in the Berlin-Frankfurt-Münster
(BFM)–based protocols, we analyzed the association of TPMT genotype with minimal residual disease levels before and after
application of a 4-week cycle of mercaptopurine during induction consolidation
The ongoing BFM trial on treatment of childhood ALL (ALL-BFM 2000) enrolls
patients from age 1 to 18 years at diagnosis and uses minimal residual disease
analysis on treatment days 33 and 78 for risk-adapted treatment stratification.
From October 1999 to September 2002, 956 patients enrolled in our ongoing
trial were monitored for minimal residual disease at 2 follow-up points (days
33 and 78) with at least 1 marker having a minimum sensitivity of 10−4 (detection of 1 leukemic cell per 10 000 cells). Of these
956 patients, 814 patients (85.1% of the entire patient population) had additional
DNA available and could be prospectively genotyped at the TPMT locus. The 142 patients not available for TPMT analysis did not differ from the included 814 patients with regard
to characteristics known to be associated with early treatment response (data
Within the ALL-BFM strategy, remission induction is initiated through
7-day monotherapy with orally administered prednisone and 1 dose of intrathecal
methotrexate on treatment day 1. From day 8 onward, treatment is complemented
by intravenous application of 3 additional drugs: vincristine, daunorubicin,
and L-asparaginase. In addition, from day 8 onward, patients
are randomly assigned to corticosteroid treatment with either prednisone or
dexamethasone. This induction strategy leads to cytomorphological remission
(<5% leukemic blasts in the bone marrow) in more than 97% of patients on
treatment day 33. Remission induction is followed by consolidation treatment
with intravenous cyclophosphamide and cytarabine, intrathecal methotrexate,
and oral mercaptopurine. Routine bone marrow aspirates are taken at diagnosis
and after completion of induction (treatment day 33) and consolidation (treatment
day 78). Minimal residual disease for analysis of leukemic cell dynamics is
measured on treatment days 33 and 78.
There were no imbalances with regard to TPMT genotype
and randomization groups or central nervous system involvement at diagnosis
(data not shown). Toxicity data were collected using the National Cancer Institute’s
Common Toxicity Criteria.20 Informed consent
was obtained from patients or their legal guardians and the study was approved
by the local ethics committee.
Minimal residual disease was analyzed using allele-specific oligonucleotide
–PCR protocols for quantitative detection of leukemic clone-specific
immunoglobulin and T-cell receptor gene rearrangements, and TAL1 deletions on a LightCycler instrument (Roche Diagnostics, Mannheim,
Germany).6,21 Genotyping for TPMT (eg, *2 and *3 alleles) was performed using a denaturing high-performance liquid
chromatography method using DNA prepared from either leukemic or remission
bone marrows.14,22 Investigators
performing TPMT genotyping were blinded with regard
to a patient’s minimal residual disease status.
Frequencies of characteristics and common factors known to be associated
with treatment response were obtained in the beginning of the analysis. Proportional
differences between groups were analyzed by χ2 or Fisher exact
tests. The association between TPMT genotype and
minimal residual disease was examined by use of unconditional logistic regression
analysis to calculate relative risks (RRs) and their 95% confidence intervals
(CIs). Minimal residual disease loads smaller than 10−4 were
defined as negative. Statistical significance was set a priori at P<.05. Analyses were computed using SPSS version 12.0 (SPSS Inc,
Genotyping of 814 patients with childhood ALL revealed 755 (92.8%) patients
with TPMT wild-type, 55 (6.8%) with heterozygous,
and 4 (0.5%) with homozygous mutant genotype (*2/*3A, *3A/*3A [n = 2], *3A/*11), respectively,
and genotype frequencies were in Hardy-Weinberg equilibrium. Allele frequencies
were as follows: TPMT*1, 96.12%; TPMT*2, 0.25%; TPMT*3A, 2.95%; TPMT*3C, 0.56%; TPMT*9, 0.06%;
and TPMT*11, 0.06%.
In the Table, patient characteristics
are depicted by TPMT genotype. Except for immunophenotype,
no major differences with regard to characteristics known to be associated
with treatment response were observed between patients homozygous for the
wild-type allele or heterozygous. All patients homozygous for a mutant TPMT allele, and consequently deficient in TPMT activity,
were treated with an approximately 10-fold reduced dose of mercaptopurine
to prevent hematopoietic toxicity (dose adjustments were not performed for
heterozygous patients). Therefore, the 4 patients with deficient TPMT activity
were not included in further analyses.
In heterozygous patients and those homozygous for the wild-type allele,
minimal residual disease levels on treatment day 33 were equally distributed
between the groups (Table). However,
when minimal residual disease levels were assessed on treatment day 78, after
administration of induction consolidation treatment, including a 4-week cycle
of mercaptopurine (60 mg/m2 per day), significant differences with
regard to clearance of minimal residual disease were observed between wild-type
and heterozygous patients (Table). For
heterozygous patients, this distribution translated into a 2.9-fold reduction
in risk of having measurable minimal residual disease after induction consolidation
treatment (RR, 0.34; 95% CI, 0.13-0.86; P = .02).
This point estimate did not significantly change in multivariate analysis
including variables known to be associated with treatment response: sex, age
at diagnosis, presenting white blood cell count, immunophenotype, and prednisone
response (good: <1000 leukemic blood blasts/μL on treatment day 8;
poor: ≥1000/μL) (RR, 0.30; 95% CI, 0.10-0.88; P = .03).
Data on hematopoietic and hepatic toxicity were available for 75% of
heterozygous patients and homozygous wild-type for TPMT and did not differ between the groups (data not shown). Similarly,
there was no difference detectable in the total group of patients when analyzing
time to treatment day 78 (wild-type patients, median of 90 [range,
60-201] days; heterozygous, median of 90 [range, 76-116] days).
Our results indicate that TPMT genotype has
a substantial impact on minimal residual disease after administration of mercaptopurine
during induction consolidation treatment in the early course of childhood
ALL, most likely through modulation of mercaptopurine dose intensity.
Several studies have shown that patients with homozygous mutant TPMT alleles conferring very low enzyme activity are at
high risk of developing severe hematopoietic toxicity after treatment with
standard doses of thiopurines.13,14 However,
whether heterozygous patients need dose reductions as well is less clear;
moreover, the requirement of dose adjustment most likely depends on the thiopurine
dose and concurrently administered chemotherapy.23,24 Based
on the data available for our study, hematopoietic toxicity did not differ
between heterozygous patients and those homozygous wild-type for TPMT. Although these data were available for only 75% of the patients,
the time to treatment day 78 (2 weeks after completion of the 4-week cycle
of mercaptopurine) was available for all patients and as a surrogate marker
for toxicity did not differ between the groups. Therefore, it seems unlikely
that childhood ALL patients with heterozygous mutant TPMT alleles treated in the BFM protocols would benefit from dose reductions
in induction consolidation treatment, as suggested for other protocols.24 Moreover, with the prognostic power of minimal residual
disease, the observed differences in tumor cell clearance depending on TPMT genotype suggest an important role for mercaptopurine
in the induction consolidation phase and could be a relevant determinant of
treatment outcome for childhood ALL treated according to BFM protocols. Finally,
after long-term outcome data become available, our results may provide a rationale
for increasing mercaptopurine dosing according to TPMT genotype
in the early course of childhood ALL. Because this rationale will affect TPMT wild-type individuals, it could have an impact on
the majority of patients and, therefore, substantially influence overall treatment
In addition, our data support a role for combining analysis of genetic
variation in drug-metabolizing enzymes and minimal residual disease in the
assessment of treatment response to specific drugs in multiagent chemotherapeutic
Corresponding Author: Martin Stanulla, MD,
MSc, Department of Pediatric Hematology and Oncology, Hannover Medical School,
Carl-Neuberg-Strasse 1, 30625 Hannover, Germany (Stanulla.Martin@MH-Hannover.de).
Author Contributions: Drs Stanulla and Schwab
had full access to all of the data in the study and take responsibility for
the integrity of the data and the accuracy of the data analysis. Drs Stanulla
and Schaeffeler contributed equally to the study.
Study concept and design: Stanulla, Eichelbaum,
Acquisition of data: Stanulla, Schaeffeler,
Flohr, Schrauder, Welte, Ludwig, Bartram, Zanger, Schwab.
Analysis and interpretation of data: Stanulla,
Schaeffeler, Flohr, Cario, Bartram, Eichelbaum, Schwab, Zimmermann.
Drafting of the manuscript: Stanulla, Schrappe,
Critical revision of the manuscript for important
intellectual content: Stanulla, Schaeffeler, Flohr, Cario, Schrauder,
Welte, Ludwig, Bartram, Zanger, Eichelbaum, Schrappe, Schwab, Zimmermann.
Statistical analysis: Stanulla, Cario, Schwab,
Obtained funding: Bartram, Eichelbaum, Schrappe.
Administrative, technical, or material support:
Schaeffeler, Welte, Ludwig, Bartram, Zanger, Eichelbaum, Schrappe.
Study supervision: Stanulla, Schrappe, Schwab.
Financial Disclosures: None reported.
Funding/Support: This work was supported in
part by the Deutsche Krebshilfe, the Bundesministerium für Bildung und
Forschung, the Robert-Bosch-Stiftung, the Madeleine-Schickedanz-Kinderkrebsstiftung,
and the Verein zur Förderung der Behandlung krebskranker Kinder Hannover
Role of the Sponsors: The sponsors had no role
in the conduct of the study; in the collection, analysis, and interpretation
of the data; or in the data analysis. The sponsors had no role regarding the
decision to publish, did not see the submitted manuscript, and are not aware
of the data obtained as the result of our analysis.
Acknowledgment: We acknowledge all participants
of ALL-BFM 2000.