|
|
Previous Article | Table of Contents | Next Article 
Blood, Vol. 95 No. 11 (June 1), 2000:
pp. 3310-3322
CLINICAL OBSERVATIONS, INTERVENTIONS, AND THERAPEUTIC TRIALS
Improved outcome in childhood acute lymphoblastic leukemia despite
reduced use of anthracyclines and cranial radiotherapy: results of
trial ALL-BFM 90
Martin Schrappe,
Alfred Reiter,
Wolf-Dieter Ludwig,
Jochen Harbott,
Martin Zimmermann,
Wolfgang Hiddemann,
Charlotte Niemeyer,
Günter Henze,
Andreas Feldges,
Felix Zintl,
Bernhard Kornhuber,
Jörg Ritter,
Karl Welte,
Helmut Gadner, and
Hansjörg Riehm for the German-Austrian-Swiss ALL-BFM Study Group
From the Department of Pediatric Hematology and Oncology,
Medizinische Hochschule Hannover, Federal Republic of Germany (FRG);
Department of Hematology, Oncology and Tumor Immunology, Charité,
Campus Berlin-Buch, Humboldt University, Berlin, FRG; Oncogenetic
Laboratory, University Children's Hospital, Gießen, FRG; Department
of Hematology and Oncology, University Göttingen; Germany;
Department of Pediatrics, University Freiburg, FRG; Ostschweizerisches
Kinderspital, St Gallen, Switzerland; Department of Pediatric
Hematology and Oncology, Charité, Humboldt University, Berlin,
FRG; Department of Pediatric Hematology and Oncology, University Jena,
FRG; Department of Pediatric Hematology and Oncology, University
Frankfurt, FRG; Department of Pediatric Hematology and Oncology,
University Münster, FRG; and St Anna Kinderspital, Wien, Austria.
A complete list of the members of the German-Austrian-Swiss ALL-BFM
Study Group appears at the end of this article.
 |
Abstract |
Trial ALL-BFM 90 was designed to improve outcome in patients with
childhood acute lymphoblastic leukemia (ALL) by using a reduced
treatment regimen. Patients were stratified into a standard-risk group (SRG), a medium-risk group (MRG), both defined by adequate early
treatment response; and a high-risk group (HRG), defined by inadequate
response to the cytoreductive prednisone prephase, induction failure,
or Philadelphia-chromosome-positive ALL. Four treatment
modifications were evaluated: dose intensification in induction by a more rapid drug sequence; administration
of L-asparaginase during consolidation therapy in the
MRG (randomized); enforced consolidation by rotational elements in
the HRG; and reduction in the dose of anthracyclines and use of only
12-Gy preventive cranial radiotherapy in the MRG and HRG, with the aim
of avoiding toxicity. Among all 2178 patients ( 18 years of age),
the 6-year event-free survival (EFS) rate (± SE) was
78% ± 1%, with a median observation time of 4.8 years. EFS was
85% ± 2% in the SRG (n = 636) and 82% ± 1% in the
MRG (n = 1299). L-asparaginase did not improve outcome
in the MRG: the event-free interval was 83% ± 2% with
L-asparaginase (n = 528) and 81% ± 2% without it
(n = 557). Because there were more systemic relapses in the HRG
(n = 243), EFS was 34% ± 3%, an outcome inferior to that in
the HRG in a previous trial, ALL-BFM 86, in which EFS was
47% ± 5% (P = .04). The rates of isolated central
nervous system relapse in the MRG and HRG were 0.8% and 1.6%,
respectively; thus, the 12-Gy preventive cranial radiotherapy regimen
apparently provided sufficient central nervous system prophylaxis.
The overall improvement over the results in ALL-BFM 86 (6-year EFS,
72%; P = .001) was based on fewer recurrences among
patients in the MRG with B-cell-precursor ALL, indicating an advantage
of more condensed induction therapy. In multivariate analysis,
inadequate in vivo response emerged as the strongest adverse prognostic variable.
(Blood. 2000;95:3310-3322)
© 2000 by The American Society of Hematology.
 |
Introduction |
Contemporary research on childhood acute lymphoblastic
leukemia (ALL) has focused on the identification of biologic and
clinical prognostic markers to generate better risk-adapted treatment
strategies.1-3 The identification of several chromosomal
aberrations and improved molecular detection techniques allow
definition of patient subsets with distinct prognostic
features.4-6 Nevertheless, treatment itself remains one of
the strongest prognostic factors, as has been shown in several
well-designed large clinical trials.7-15 Intrinsic drug
resistance is the major cause of treatment failure. So far, this
phenomenon has not been linked to any specific clonal abnormality.
However, knowledge gained from systematic measurements of in vitro and
in vivo drug resistance could be used for better identification of
patients at increased risk of treatment failure.3,16-23
With treatments for ALL having achieved long-term cure rates above 70%
in unselected patient populations, the acute and long-term toxicity of
such treatments identified by the end of the 1980s had to be taken into
account by researchers introducing new therapies.24-29 Thus, one major focus of the design of trial ALL-BFM 90 was a reduction
in the use of treatment elements with long-term toxicity; thus, during
induction, the cumulative anthracycline dose was reduced by 25%, from
160 mg/m2 of body-surface area (the dose used in trial
ALL-BFM 86) to 120 mg/m2. The elimination of preventive
cranial radiotherapy (CRT) in patients with low-risk ALL and the
successful stepwise reduction in CRT to 12 Gy in those with
intermediate-risk ALL (begun in previous ALL-BFM studies1)
was the basis for introducing the use of 12-Gy CRT in patients at
medium and high risk. To strengthen the extracompartmental treatment,
the ALL-BFM study group introduced high-dose methotrexate (HD-MTX) in
trial ALL-BFM 86.3
The experience from a series of large multicenter trials conducted by
the ALL-BFM study group (particularly, trial ALL-BFM 863)
made it evident that the BFM treatment concept could provide a cure
rate of more than 75% for approximately 90% of all patients. This
large patient subset was characterized by an adequate early response to
cytoreductive therapy with 7 days of prednisone (and 1 intrathecal dose
of methotrexate [MTX] on day 1). An adequate response was easily
measured as a leukemic blast-cell count of under 1000/µL in
peripheral blood (PB) on day 8 ("prednisone: good response"
[PGR]). Demonstration of in vivo prednisone resistance defined
high-risk patients, who comprised about 10% of the study population.16 Prognosis in this subset was unfavorable,
with an event-free survival (EFS) rate below 50%.3
Therefore, a new strategy for this well-defined patient subgroup was
developed, based on early intensification with treatment elements
derived from the strategy for treating ALL relapse.
Most relapses, however, occurred in medium-risk patients who did not
have any specific prognostic biologic or clinical
characteristics.3 Improvement in the largest patient subset
was therefore sought by using a more condensed induction regimen and
enforcement of the consolidation phase. This strategy had the following
features: to increase dose intensity, induction element protocol I was
shortened by 1 week by starting L-asparaginase on day 12 instead of day 19 (as in trial ALL-BFM 86); and to intensify
consolidation in medium-risk patients, 4 high-dose pulses of
L-asparaginase were added in a randomized manner to the
existing consolidation phase using HD-MTX and 6-mercaptopurine
(protocol M in trial ALL-BFM 86).
Trial ALL-BFM 90, which had 2178 unselected patients, was the largest
cooperative trial performed so far by the BFM study group and can thus
offer treatment results with regard to biologic and clinical variables
evaluated for their prognostic importance. To better assess the
importance of the advances and pitfalls in this trial, direct
comparisons with updated results of previous ALL-BFM trials were
performed. This was possible because patient populations in all ALL-BFM
trials are unselected and because treatment modifications were limited
and well defined.
 |
Patients and methods |
Patients
From April 1, 1990, until March 31, 1995, 2300 patients up to 18 years of age were enrolled in the 96 participating centers in Germany,
Austria, and Switzerland. A total of 122 patients (5.3%) were not
eligible for ALL-BFM 90 according to the protocol criteria. Thirty-five
of these patients had undergone major pretreatment (steroids or
cytostatic drugs given within 4 weeks before diagnosis), 27 were pilot
patients for the subsequent trial ALL-BFM 95, and 20 were treated with
a different protocol (non-BFM), had the wrong treatment assignment, or
received inadequate treatment because of nonmedical reasons. In 11 of
the 122 patients, diagnosis or treatment was done outside the
participating countries; in 10, the diagnosis of ALL could not be
established; in 7, a major additional medical condition prevented
protocol therapy (patients with Down syndrome were excluded only if
they also had a severe congenital heart defect); in 4, the BFM risk
factor (BFM-RF) could not be calculated because of missing data; in 4, treatment was stopped early because of nonmedical reasons; and in 4, ALL was a second malignancy or relapse of previously unrecognized ALL.
Thus, 2178 patients were evaluable for this study. Informed consent was
obtained from the guardians of all patients. Patients with
Philadelphia-chromosome-positive (Ph+) ALL were included
in a previous report.6
Diagnosis
The diagnosis was established when at least 25% lymphoblasts were
present in the bone marrow (BM) or when blasts were present in the PB
or cerebrospinal fluid (CSF). BM and blood smears and CSF cytospin
preparations were stained by using a modified Wright staining technique
and cytochemistry reactions (periodic acid-Schiff reaction, acid
phosphatase, -naphthyl acetate esterase, and myeloperoxidase reaction) and were reviewed in the central laboratory of the study center by using the French-American-British criteria.30
Central nervous system (CNS) involvement was diagnosed if more than 5 cells per µL were counted in the CSF and if lymphoblasts were
identified unequivocally or if intracerebral infiltrates were detected
on cranial computed tomography.31
Immunophenotyping
Immunophenotyping was done as described elsewhere.32,33
Surface antigens were considered positive if at least 20% of the leukemic cells expressed the antigen with more than 98% fluorescence intensity compared with negative control cells. Positivity for terminal
deoxynucleotide transferase (TdT) and cytoplasmic (cy) antigens was
defined as more than 10% of the cells exhibiting nuclear or
intracytoplasmic fluorescence (TdT, cyIgM, and cyCD3). In 1994, 2-color
flow cytometric analysis was introduced; the procedure uses appropriate
monoclonal antibodies directly conjugated to fluorescein isothiocyanate
or phycoerythrin. Immunophenotypic subgroups were defined according to
the definition provided by the European Group for the Immunological
Characterization of Leukemias, as follows: pro-B ALL, TdT+,
CD19+, CD10 , cyIgM ,
surface immunoglobulin (sIg) ; common ALL,
TdT+, CD19+, CD10+,
cyIgM , sIg ; pre-B ALL,
TdT+, CD19+, CD10+/ ,
cyIgM+, sIg ; and T-cell ALL (T-ALL),
TdT+, cyCD3+, CD7+.34
Coexpression of myeloid antigen was defined as simultaneous expression
of one or more of the myeloid-lineage associated molecules tested
(CD13, CD33, CD65s) on at least 20% of the lymphoblasts.
Cytogenetic and molecular genetic analysis
Cytogenetic studies were carried out by using standard techniques as
described elsewhere.35 In November 1992, screening for
BCR/ABL based on reverse transcriptase-polymerase chain
reaction was initiated.36
DNA index
Cellular DNA content was determined by using flow cytometry as
previously described.37 The DNA index of the leukemic
blasts was defined as the ratio of DNA content in leukemic
G0/G1 cells to that in normal diploid
lymphocytes. A cut-off DNA-index value of 1.16 was used to distinguish
prognostic categories.
Estimation of the leukemic cell mass at diagnosis
The leukemic cell mass estimate (the BFM-RF) was calculated with the
following equation: BFM-RF = 0.2 × log (blood
blasts/µL + 1) + 0.06 × liver size in centimeters below
the costal margin + 0.04 × spleen size in centimeters below
the costal margin.38
Definition of prednisone response
In all patients, treatment started with 7 days of monotherapy with
prednisone and 1 intrathecal dose of MTX on day 1. The first day of
treatment was the day of the first administration of prednisone. The
dosage of prednisone was increased steadily to 60 mg/m2
daily in accordance with leukemic cell mass, renal, and metabolic variables to circumvent complications of acute cell lysis. The number
of leukemic blasts in the blood on day 8 was calculated from the
absolute leukocyte count and the percentage of blasts in PB smears
determined by central review in the study center. The presence of at
least 1000/µL blasts in PB on day 8 was defined as a "prednisone:
poor response" (PPR); fewer than 1000/µL leukemic cells was
required for a classification of PGR.16
Patient stratification and treatment
Patients were assigned to 1 of 3 branches: a standard-risk group
(SRG), a medium-risk group (MRG), and a high-risk group (HRG). The main
criteria for stratification were the leukemic cell mass estimate
(BFM-RF) and the treatment response.3,38 Additional criteria included the presence of the T-cell immunophenotype, rearrangement BCR/ABL or translocation t(9;22), and CNS involvement.
Therefore, patients in the SRG had fewer than 1000/µL blasts in PB on
day 8 (PGR), a BFM-RF below 0.8, no CNS disease, and no T-ALL or
mediastinal mass. Those in the MRG had fewer than 1000/µL blasts in
PB on day 8 (PGR) and a BFM-RF of 0.8 or higher, or a BFM-RF below 0.8 and CNS disease or T-ALL, or a mediastinal mass. Patients in the HRG
had more than 1000/µL blasts in PB on day 8 (PPR), or fewer than
1000/µL blasts in PB but 5% or greater marrow blasts on day 33 (M2/M3), or Ph+ ALL.
An outline of the treatment strategy is shown in Figure
1, and the details of each treatment
element are provided in Table 1. All
patients who did not qualify for HRG therapy received induction
protocol I, consolidation/extracompartmental protocol M, reinduction
(delayed intensification) protocol II, and maintenance therapy. In
protocol I, daunorubicin was given 4 times at a dose of 30 mg/m2 each time. During consolidation, HD-MTX (5 g/m2 per 24 hours) was combined with a late leucovorin
rescue. The first dose of intravenous leucovorin (30 mg/m2)
was scheduled for hour 42 after the start of MTX administration; the 2 subsequent doses (15 mg/m2 each) were given at hours 48 and
54. Additional leucovorin was given only if the MTX level exceeded 1.0 µmol/L at hour 42 or 0.4 µmol/L at hour 48. If toxicity was
acceptable after the first exposure to systemic MTX, the leucovorin dose at hour 42 was decreased to 15 mg/m2 in subsequent
courses. SRG and MRG patients who did not initially have CNS disease
received a total of 11 injections of intrathecal MTX during the
intensive-treatment phase but no intrathecal MTX during maintenance
therapy.

View larger version (25K):
[in this window]
[in a new window]
| Fig 1.
Treatment in trial ALL-BFM 90.
SRG indicates standard-risk group; MRG, medium-risk group; and HRG,
high-risk group. Details on treatment elements I, M, M-A, II, HR-1,
HR-2, and HR-3 are given in Table 1. Asterisk indicates no preventive
radiotherapy if patient was under 1 year of age; patients with central
nervous system involvement received no radiation if they were under 1 year of age, 18 Gy if they were older than 1 year of age but under 2 years of age, and 24 Gy if they were 2 years of age or older.
|
|
HRG patients were treated with a shorter induction (Table 1) and
continued on a more intensive rotational consolidation schedule consisting of 3 different 6-day-long pulses of high-dose chemotherapy (HR-1, HR-2, and HR-3) that were repeated 3 times. These elements were
derived directly from the ALL-BFM REZ relapse strategy39 using HD-MTX (5 g/m2 per 24 hours) and high-dose cytarabine
in various combinations (Table 1). CNS-directed preventive therapy for
HRG patients consisted of 3 doses of intrathecal MTX in induction
(protocol I/A), and 9 doses of triple-drug intrathecal therapy (MTX,
cytarabine, and prednisolone) during intensive consolidation.
Reinduction with protocol II was not used in these patients.
In all treatment elements, the use of either Escherichia coli
L-asparaginase (Bayer, Leverkusen, Federal Republic of
Germany [FRG], or Medac, Hamburg, FRG) or Erwinia
L-asparaginase (Speywood, London, UK) at equal dosages was
permitted. The protocol did not require any specific preparation.
However, the Bayer E coli L-asparaginase was
eventually no longer available.
Preventive CNS irradiation at a dose of 12 Gy was given only in MRG
patients at the end of reinduction and in HRG patients after intensive
consolidation (Figure 1). In patients who initially had CNS disease,
CRT was administered in age-adapted dosages; thus, patients under 1 year of age received no CRT, patients older than 1 year of age but
under 2 years of age received 18 Gy, and patients 2 years of age or
older received 24 Gy. These patients also received 2 additional doses
of intrathecal MTX in both protocol I and protocol II and, if in the
HRG, 1 additional administration of intrathecal triple-drug therapy
(MTX, cytarabine, and prednisolone) in each cycle HR-2. In boys with
clinically overt testicular involvement, local irradiation (24 Gy) was
performed. Other forms of local radiotherapy were not scheduled in
protocol ALL-BFM 90.
Maintenance therapy was initiated 2 weeks after the end of reinduction
(protocol II) or the ninth HR element. The scheduled dose of
6-mercaptopurine was 50 mg/m2 a day given orally. MTX was
given orally at a dose of 20 mg/m2 once a week, with
adjustments in dosage made in accordance with the white blood cell
(WBC) count (target range, 2-3 × 109/L). For all
patients, the total duration of therapy was 24 months.
MRG patients were randomly assigned at the end of protocol I and
received either standard consolidation with 6-mercaptopurine and HD-MTX
or additional L-asparaginase during consolidation (Figure 1). In branch MRG-2, protocol M-A was used, providing 4 cycles of
25 000-IU/m2 L-asparaginase each time, after
infusion of HD-MTX. HRG patients who achieved complete remission (CR)
were randomly assigned to receive or not to receive granulocyte
colony-stimulating factor (G-CSF) prophylactically between the pulses,
during intensive consolidation. An interim analysis of this trial was
reported earlier.40 In this report, patients from the HRG
are analyzed as a common cohort because there was no difference in
outcome between the randomized subgroups.
Allogeneic bone marrow transplantation (BMT) was recommended for a
subset of HRG patients if a matched sibling donor was available. The
following qualifying criteria for BMT were developed on the basis of
the results of trial ALL-BFM 86: either Ph+ ALL defined by
translocation t(9;22) or BCR/ABL rearrangement, or nonresponse
to induction therapy (no CR at day 33 of protocol I), or PPR and at
least one of the following: T-ALL, coexpression of a myeloid marker,
BFM-RF of 1.7 or higher, and t(4;11).
Response criteria
CR was defined as the absence of leukemic blasts in PB and CSF, less
than 5% lymphoblasts in marrow aspiration smears, and no evidence of
localized disease. Relapse was defined as recurrence of lymphoblasts or
localized leukemic infiltrates at any site.
Statistical analysis
For the random assignment to treatment with
L-asparaginase in branch MRG, the following estimate was
made on the basis of results of previous studies: 30% of patients will
be at risk for relapse after induction protocol I. Sample-size
calculations then determined that 230 patients were needed in each of
the randomization branches, MRG-1 and MRG-2, to detect a decrease from
30% to 20% in the relapse rate resulting from intensification of
protocol M with a power of 0.80 ( error = 0.05).
The Kaplan-Meier method41 was used to estimate survival
rates. Differences were compared with the 2-sided log-rank
test.42 EFS was calculated from diagnosis to the time
of analysis or to the first event; SE and 95% confidence intervals
(CI) are provided. Failure to achieve remission (early death or
resistant leukemia), relapse, death during continuous complete
remission (CCR), and second malignancy were evaluated as events;
failure to achieve remission on day 1 was registered as event on day 1. Patients who did not have CR at day 33 of induction (after
protocol I/A) were treated in the HRG. Nonresponse was registered as an
event in patients who did not have CR after the third HR element, even if CR was achieved later. Patients lost to follow-up were censored at
the time of their withdrawal. For patients in the randomized subsets
MRG-1 and MRG-2, the event-free interval (EFI) was calculated from the
time of randomization to the end of the first remission (relapse, death
in CCR, or second malignancy) or to the time of analysis. For the
estimate of probability of disease-free survival, only relapse was
considered to be an event. The results of trial ALL-BFM 90 (and of
previous trials ALL-BFM 83 and ALL-BFM 86) were updated in April 1998.
Differences in the distribution of variables among patient
subsets were analyzed by using the 2 test for
categorized variables and the Wilcoxon rank sum test for continuous
variables. Differences between EFS distributions for patient
subpopulations were evaluated by using 2-sided log-rank tests.42 The prognostic relevance of clinical and biologic
variables in the whole group was examined with use of a stepwise Cox
regression analysis.43 For the continuous variables of age,
WBC count, and BFM-RF, multiple cut-off points were used, each of which
divided all patients into 2 complementary subsets. All available
variables that were considered to have prognostic relevance because of
our own results or previously published data were used as covariables in the Cox regression model.
For patients with CR, cumulative incidence functions were
calculated for each of the following competing causes of failure: isolated BM relapse, isolated CNS relapse, combined BM and CNS relapse,
other relapses, and other events. Comparisons were done by using the
95% CI for the difference between 6-year point estimates for the
incidence functions.44
 |
Results |
Patient characteristics
The median age of all 2178 evaluable patients was 4.6 years (range, 0.01-18.53 years); 2.7% of patients were infants under 1 year of age. The median WBC count at presentation was
11.8 × 109/L (range,
0.3-1496.0 × 109/L). Clinical and biologic
characteristics of the whole study population and the 3 risk groups are
summarized in Table 2. The major
extramedullary disease manifestations were a mediastinal mass in 8% of
patients, nodal involvement (without mediastinal involvement) in 36%,
liver and spleen enlargement (organ palpable more than 4 cm below the
costal margin) in 31% and 27%, respectively, and CNS involvement in
2.5%. Eight boys (0.6%) had testicular involvement. B-cell-precursor
ALL predominated (86.5% of patients); 13.5% patients had T-ALL. Among
patients with B-cell-precursor ALL, pro-B ALL was diagnosed in 6%,
common ALL in nearly 75%, and pre-B ALL in 19%.
Coexpression of myeloid markers was found in 19% of all patients. This
proportion was higher than that in trial ALL-BFM 86 because of the
introduction of a more sensitive, direct immunofluorescence technique
using phycoerythrin-conjugated anti-CD13 and anti-CD33 antibodies.3,45 The distribution of immunosubtypes was
otherwise identical to that in previous results.3 Among
1205 patients successfully analyzed with cytogenetic or molecular
genetic methods, 27 (2.2%) were found to have Ph+ ALL. The
rate of detection of Ph+ ALL improved greatly after
molecular screening for BCR/ABL was introduced in
1992.36 Translocation t(4;11) was found in 2.9% and
t(1;19) in 2.1% of patients who had cytogenetic analysis. The
proportion of patients with inadequate response to the 7-day prednisone
prephase regimen and 1 application of intrathecal MTX on day 1 (ie,
those with PPR) was 9.5% (n = 202).
Treatment results
EFS.
After a median observation time of 4.8 years (range, 0-8.1 years), the
estimate for EFS of all 2178 evaluable patients was 78% ± 1% at
6 years and 77% ± 1% at 8 years; 1.7% of all patients did not
have CR, 17.7% had relapse, 0.5% had a second malignancy diagnosed,
and 1.6% died of complications of therapy (Table
3). Among all 2300 patients enrolled, EFS
was 76% ± 1% at 8 years. The estimate of probability of
disease-free survival for all evaluable patients was 78% ± 1%,
and the estimate of probability of survival was 85% ± 1%. The
78% ± 1% 6-year EFS (CI, 76%-79%) in trial ALL-BFM 90 was
significantly higher than the EFS at 6 years in trials ALL-BFM 86 and
ALL-BFM 83, in which it was 72% ± 1% (CI, 69%-75%) and
64% ± 2% (CI, 60%-68%), respectively (P = .001 for
ALL-BFM 86 compared with ALL-BFM 90; P = .0001 for ALL-BFM 83 compared with ALL-BFM 90; Figure
2).

View larger version (17K):
[in this window]
[in a new window]
| Fig 2.
Kaplan-Meier estimate of event-free survival of all
evaluable patients in trials ALL-BFM 90, ALL-BFM 83, and ALL-BFM 86.
|
|
Figure 3 shows Kaplan-Meier plots for the 3 risk groups in trial ALL-BFM 90. EFS at 6 years was 85% ± 2% in
the SRG and 82% ± 1% in the MRG. Thus, 6-year EFS was above
80% for approximately 90% of all patients. The results in the SRG
were significantly better than those in the MRG (P = 0.03).
In contrast, patients in the HRG had an EFS rate of only
34% ± 3%.

View larger version (17K):
[in this window]
[in a new window]
| Fig 3.
Kaplan-Meier estimate of event-free survival of patients
in trial ALL-BFM 90, according to risk group.
|
|
Remission failures.
A total of 2140 patients (98.3%) achieved first CR. The CR rate was
lowest in the HRG (92.2%) because all patients with induction failure
were, by definition, stratified into that group. Thirty-eight of the
2178 evaluable patients did not have remission because of early death
or resistant disease. Before and during the first 5 weeks of induction,
22 patients died. Ten of these patients died of complications
(hyperleukocytosis, cardiomyopathy, encephalopathy, or bleeding) before
or within the first few days of treatment, and 12 patients died of
complications that were more closely related to the treatment (sepsis
or pneumonia, 10 patients; massive bleeding from an ulcer, 1 patient;
and hepatopathy and cardiomyopathy after 3 doses of vincristine and
daunorubicin, 1 patient). Thus, the early mortality rate was 1.0%
(Table 3). Sixteen patients, of whom 4 were initially classified as MRG
patients with PGR, did not have remission at day 33 and were also not
in CR after the third HR pulse (nonresponse). Six of the 16 nonresponse
patients had CR very late 4 after allogeneic BMT and 1 after extended
chemotherapy but all had relapse. One patient had CR after all cycles
of HRG treatment, underwent BMT 6.5 months after diagnosis, and is
still in first CR at 7.5 years.
Another 40 patients (5 initially defined as SRG patients, 13 as MRG,
and 22 as HRG) who were also not in CR at day 33 had CR during HRG
treatment; however, 32 of them subsequently had relapse, indicating a
poor prognosis for patients with induction failure. EFS at 6 years for
all patients with induction failure (n = 56) was only
11% ± 5%.
Deaths in CR.
Thirty-four patients (1.6%) died in first CR from complications (Table
3). Four of these patients died because of toxicity related to
allogeneic BMT (branch HR), 19 died of infection (sepsis and
pneumonia in 18 patients and cytomegalovirus infection in 1), 5 patients died as a result of massive unexpected bleeding during
induction, and 5 died of organ failure (1 patient each of
cardiomyopathy, hepatopathy, and ileus and 2 of encephalopathy). One
patient was found dead at home and an autopsy was not performed. Twenty-six of the 30 deaths not related to BMT occurred within 12 months of diagnosis, ie, during the intensive-treatment phase or early
in the maintenance-treatment phase. Two patients died in the second
year of therapy, and 2 died at the end of maintenance therapy (1 patient with Down syndrome who died of pneumonia and 1 patient
who died of cytomegalovirus encephalitis).
Relapses.
Relapse occurred in 385 patients (17.7%; Table 3). Thirty-three
percent of all recurrences were in the HRG, 48% were in the MRG, and
19% were in the SRG. Most relapses (85%) in the HRG occurred within
the first 2 years of diagnosis, ie, during therapy. In contrast, most
relapses in the SRG and MRG occurred after the end of treatment (Figure
3). Systemic failures were more frequent among HRG patients; they
developed in 39.1% of patients in that group but in only 6.9% of SRG
patients and 9.2% of MRG patients. There were, however, no differences
in the rates of extramedullary recurrences among the 3 risk groups. The
overall incidence of isolated and combined CNS relapse was 1.0% and
1.9%, respectively. Most of the other extramedullary relapses involved
lymph nodes or mediastinal sites (or both), the thymus, or the testes.
Second malignancy.
In 10 patients, a secondary malignancy developed at a median time of
40.2 months (range, 15-68 months) after diagnosis. Secondary malignancies occurred in all 3 risk groups; there was no evidence of a
higher incidence among more intensely treated patients (Table 3). Five
second malignancies were acute myelogenous leukemia (AML),
2 were brain tumors, 1 was Hodgkin disease, 1 was basal cell carcinoma,
and 1 was malignant histiocytosis.
Impact of new or modified treatment elements: intensification of
consolidation in MRG patients by high-dose L-asparaginase.
The probability of EFI among MRG patients randomly assigned to receive
4 courses of 25 000 IU/m2 L-asparaginase
(branch MRG-2, n = 528) was 83% ± 2%. In patients receiving
the standard consolidation without L-asparaginase
(n = 557), the probability of EFI was 81% ± 2%. The
difference in EFI in the 2 groups was not significant
(P = .67). In the Cox regression including all variables
found to be relevant for prognosis, no significant influence of
additional L-asparaginase treatment was found.
Impact of new or modified treatment elements: intensive early
consolidation with rotational high-dose pulses and reduced preventive
CRT in HRG patients.
Evaluation of the modified approach for high-risk patients had to be
performed by comparing results with those in the matching subset of
patients in trial ALL-BFM 86.3 The only difference at
diagnosis between the high-risk patients in the 2 trials was the
distribution of age subgroups: there were more infants (< 1 year of
age) in trial ALL-BFM 86 and more patients older than 6 years of age in
trial ALL-BFM 90. The EFS for high-risk patients in trial ALL-BFM 86 (group EG) was 47% ± 5%, whereas it was 34% ± 3% in
trial ALL-BFM 90 (P = .04). The difference was due to the
higher number of systemic recurrences in high-risk patients in ALL-BFM
90: the cumulative incidence of isolated BM relapse at 6 years was
42.7% ± 4% in ALL-BFM 90, but 24.7% ± 6% in ALL-BFM 86 (P = .01). Despite the reduction in CRT, the cumulative
incidence of isolated and combined CNS relapse was only 1.8% and
2.7%, respectively, in ALL-BFM 90, whereas it was 7.4% and 5.4%,
respectively, in ALL-BFM 86 (P not significant).
To eliminate the impact of the more frequent detection of t(9;22) in
the evaluation of the HRG in ALL-BFM 90, comparisons were performed in
HRG patients defined only by PPR (which is the largest subset within
HRG) (Figure 4). EFS was still
significantly more favorable among patients with PPR (n = 95) in
trial ALL-BFM 86 (46% ± 5%) than among patients with PPR
(n = 202) in ALL-BFM 90 (34% ± 3%; P = .04). When
the study impact was tested in a Cox model together with BMT as a
time-dependent covariable, the difference remained about the same
(P = .03). Accordingly, when all patients were censored at
the time of transplantation (1 allogenic BMT in ALL-BFM 86 and 35 BMTs
in ALL-BFM 90) the 6-year EFS for high-risk patients in trial ALL-BFM
86 was 47% ± 5%, whereas that in ALL-BFM 90 was
34% ± 4% (P = .04).

View larger version (14K):
[in this window]
[in a new window]
| Fig 4.
Kaplan-Meier estimate of event-free survival of patients
with a poor response to prednisone (PPR) in trials ALL-BFM 90 and
ALL-BFM 86.
|
|
Patients randomly assigned to receive G-CSF or no G-CSF between the
pulses had the same outcome.40 The probability of EFI at 6 years was 34% ± 9% in patients not given G-CSF and
38% ± 9% in patients who received the agent
(P = .98).
Impact of new or modified treatment elements: outcome in
standard-risk and medium-risk patients according to modified induction
in patients with PGR.
EFS among all patients with PGR (n = 1935) at 6 years was
82% ± 1%. As shown in Figure 5,
this was a significant improvement over results in comparable patients
in trials ALL-BFM 83 (n = 467; 6-year EFS, 69% ± 2%;
P = .0001) and ALL-BFM 86 (n = 783; 6-year EFS,
77% ± 2%; P = .0012 [standard-risk patients treated without reinduction in ALL-BFM 83 and ALL-BFM 86 were excluded to
reduce bias1,3]). To identify the subset in which the
greatest improvement occurred, patients from the most comparable
trials, ALL-BFM 86 and ALL-BFM 90, were analyzed according to risk
groups. In SRG patients, no difference was found: EFS was
85 ± 2% in trial ALL-BFM 90 (Figure 3) and 84% ± 3% in
ALL-BFM 86 (n = 175; only patients who received reinduction were
included). In MRG patients, a significant improvement was noted: EFS at
6 years was 82% ± 1% in trial ALL-BFM 90 and 75% ± 2%
in ALL-BFM 86 (P = .001). This difference was due to a
reduced cumulative incidence of isolated BM relapses in trial ALL-BFM
90, which at 6 years, was 10.1% ± 0.9% in the MRG in ALL-BFM 90 and 15.1% ± 1.5% in the respective subset in trial ALL-BFM 86 (P = .006). The reduction in isolated BM recurrences was
achieved exclusively in patients with B-cell-precursor ALL.
Accordingly, the 6-year EFS among medium-risk patients with
B-cell-precursor ALL was 82% ± 1% in trial ALL-BFM 90 and
73% ± 2% in trial ALL-BFM 86 (P = .0001). The 6-year
EFS among patients with T-ALL was 80% ± 3% in the MRG in
ALL-BFM 90 and 84% ± 4% in the corresponding subgroup
in ALL-BFM 86 (P = .45).

View larger version (18K):
[in this window]
[in a new window]
| Fig 5.
Kaplan-Meier estimate of event-free survival of patients
with a good response to prednisone (PGR) in trials ALL-BFM 90, ALL-BFM
83, and ALL-BFM 86.
Asterisk indicates that in trials ALL-BFM 83 and ALL-BFM 86, patients
treated without reintensification had a significantly poorer outcome.
Therefore, these patients were excluded from analysis to reduce the
bias in this comparison.
|
|
Impact of new or modified treatment elements: reduction of
preventive CRT in MRG patients with a BFM-RF of 1.2 or higher.
The impact of reducing CRT to 12 Gy in trial ALL-BFM 90 was analyzed by
comparing the higher risk patients within the MRG (those with a large
cell load [BFM-RF 1.2] but no initial CNS involvement) with the
respective subset of patients in trial ALL-BFM 86, who were treated
with 18 Gy. The 6-year EFS was 77% ± 2% in this subset of
patients in ALL-BFM 90 (n = 570) and 68% ± 3% in the
subset in ALL-BFM 86 (n = 259; P = .003). The cumulative
incidence of relapse with CNS involvement was 4.0% in ALL-BFM 90 and
6.7% in ALL-BFM 86 (P = .19); that of non-CNS recurrences
was 16% ± 1.8% in ALL-BFM 90 and 23.6% ± 2.8% in
ALL-BFM 86 (P = .02). Thus, the reduction in CRT did not
adversely affect the rate of CNS recurrences in MRG patients in trial
ALL-BFM 90 who had no CNS involvement and a BFM-RF of 1.2 or higher
(Figure 6).

View larger version (20K):
[in this window]
[in a new window]
| Fig 6.
Impact of reduced preventive cranial radiotherapy (CRT)
in medium-risk patients with a BFM-RF of 1.2 or higher and no central
nervous system (CNS) involvement initially.
Shown is the cumulative incidence of relapse with CNS involvement
compared with any other kind of relapse in trials ALL-BFM 86 (preventive CRT dose, 18 Gy) and ALL-BFM 90 (preventive CRT dose, 12 Gy).
|
|
Outcome according to prednisone response.
The 6-year EFS in patients with PGR in trial ALL-BFM 90 was
82% ± 1%. Among patients with PPR, the EFS was 34% ± 3%
(Figures 4 and 5). The overall proportion of patients with PPR was
9.5% but varied widely among subsets of patients (Table
4). PPR was infrequent among patients who
were 1 to 9 years of age (7.2%), had a WBC count below
50 × 109/L (4.3%), had pre-B ALL (4%) or common
ALL (5%), or met standard-risk criteria developed by the National
Cancer Institute (NCI) consensus conference (3.5%).46 With
regard to 6-year EFS rates among patients with PPR, a homogeneous
profile emerged: in no such defined subgroup was EFS above 50%. Within
each subset, patients with PPR had an outcome that was significantly
worse than that in the corresponding group with PGR.
Outcome in subgroups defined by NCI consensus risk criteria.
To facilitate comparability of the results, we adopted the risk
criteria of the NCI consensus conference46 and found 2 distinct subgroups (Table 5 and Figure
7). The largest group, the NCI SRG (WBC
count < 50 × 109/L and age 1-10 years), comprised 64% of all patients, and its 6-year EFS was
86% ± 1%. This was significantly better than the EFS in the NCI
HRG (WBC count 50 ×
109/L or age 10 years; 33.2% of
all patients), which was 64% ± 2% (P = .0001). In
contrast to the NCI definition, we combined patients with
B-cell-precursor ALL and those with T-ALL, did not exclude patients
with cytogenetic abnormalities, and included infants (patients less
than 1 year of age) as an extra group. The 6-year EFS for infants was
50% ± 7%. When the 2 NCI risk groups were analyzed according to
prednisone response (Table 4), 2 subsets were identified in each risk
group (infants excluded). The largest poor-risk subset was derived from
the NCI HRG; 20% of these patients were defined by PPR. The PPR group
in the NCI SRG was smaller but contributed 25% of the patients to the
total PPR group.

View larger version (18K):
[in this window]
[in a new window]
| Fig 7.
Kaplan-Meier estimate of event-free survival of evaluable
patients, according to modified National Cancer Institute consensus
risk criteria.
Results are independent of immunophenotype and results in infants.
Standard risk is a white blood cell (WBC) count below
50 × 109L and age 1 year to under 10 years; high risk is a WBC of
50 × 109L or higher or age 10 years
or older.
|
|
Prognostic factors
Table 5 shows treatment results, defined by the 6-year EFS, in
relation to a large number of clinical and biologic variables. The
following variables were adversely associated with EFS: age under 1 year, age over 10 years, WBC count higher than
50 × 109L, hepatomegaly and
splenomegaly, BFM-RF 1.7 or higher, mediastinal mass, CNS involvement,
pro-B-cell ALL, T-ALL, t(9;22), t(4;11), nonresponse at day 33, and
PPR. In contrast, the following variables were associated with a more
favorable EFS: female sex, age 1 to 5 years, WBC count below
10 × 109L, BFM-RF below 0.8, hemoglobin value below 80 g/L, common ALL, DNA index 1.16 or higher, and PGR. Coexpression of a myeloid marker or markers had no
significant impact on EFS.
Variables that were known about all or most patients and were shown to
have significant univariate prognostic significance (Table 5) were
included in the Cox stepwise analysis. This revealed the unfavorable
prognostic impact of several variables, including male sex, high cell
mass (WBC count 50 × 109/L), pre-B and pro-B
immunophenotype, initial CNS disease, age over 6 years or under 1 year,
PPR, and nonresponse at day 33 (Table 6).
When the analysis included only patients who were assessed cytogenetically, Ph+ ALL was an additional unfavorable
factor (risk ratio, 2.96; P = .001). Other independent
adverse prognostic factors in that subset were high WBC count, pro-B
subtype, CNS disease, PPR, and induction failure (nonresponse at day
33). DNA ploidy could not be identified as an independent prognostic
factor when the Cox model was used to analyze patients in whom the DNA
index was determined. When Down syndrome was included in the Cox model,
it was found to have only borderline significance (risk ratio, 1.88;
P = .06).
 |
Discussion |
With 2178 evaluable patients, trial ALL-BFM 90 was the
largest of the 6 trials conducted by the ALL-BFM study group so far. It
was done in 3 countries with nearly 100 participating centers. Because
of the trial's unselected study population, a large panel of clinical
and biologic characteristics could be analyzed with respect to early
response and treatment outcome. The 6-year EFS of 78% ± 1% is
the most favorable treatment result ever achieved in an ALL-BFM
trial.1,3,47 More remarkably, for approximately 90% of the
patients in the trial (ie, for all patients who had a low burden of
leukemic cells in the PB after 1 week of prednisone therapy [the PGR
group]), an EFS rate of 80% or greater was achieved with the more
condensed induction phase. The reduction of the anthracycline dose in
induction and the reduced preventive CRT for patients with a larger
cell mass were appropriate because no adverse effects of these therapy
changes on outcome were found. The intensified consolidation therapy
with high-dose L-asparaginase did not provide an additional
improvement in medium-risk patients.
An inadequate early response to prednisone (and 1 dose of intrathecal
MTX on day 1) again emerged as the strongest adverse prognostic factor
other than nonresponse at day 33 of induction therapy, an indicator of
a very high risk of treatment failure in a small group of patients
(Table 6). In nearly all clinically and biologically defined subgroups,
the response to prednisone separated subsets with a good prognosis from
those with a poor prognosis. The new approach of treating high-risk
patients (mainly characterized by an inadequate early response) with
rotational high-dose pulses could not abrogate the inferior prognosis
of that subgroup (Figure 4). Among patients with PPR, those with common
ALL had a better prognosis than those with T-ALL (P = .007) or pro-B-cell ALL (P = .0001). Compared with the outcome in
high-risk patients in trial ALL-BFM 86, the outcome in all immunologic
subgroups was nevertheless inferior, although this result was more
pronounced in patients with T-ALL.3
The cumulative dosages of the high-risk therapies in trial ALL-BFM 90 and trial ALL-BFM 86 (intensive phase of treatment only) might provide
an explanation for the observed difference in outcome. In ALL-BFM 90, the HRG treatment regimen contained fewer alkylating agents (4 times
less ifosfamide and no cyclophosphamide), less prednisone, and no
mitoxantrone (which was given 4 times at a dose of 10 mg/m2
in ALL-BFM 86). These reductions obviously could not be compensated by
the use, in ALL-BFM 90, of more dexamethasone (900 mg/m2
compared with 236 mg/m2), etoposide (none in ALL-BFM 86),
L-asparaginase (285 000 IU/m2 compared with
120 000 IU/m2), 6-thioguanine (1500 mg/m2
compared with 840 mg/m2), cytarabine (36 000
mg/m2 compared with 17 800 mg/m2), and
intravenous MTX (30 000 mg/m2 compared with 20 000
mg/m2). The other possible explanation for the difference
in outcome is that a more continuous drug exposure without therapy-free
intervals caused by ablative components of therapy has more
antileukemic power.15 The reduction in CRT to 12 Gy in HRG
patients is not likely to have induced this negative effect, since the
incidence of all CNS-related relapses decreased at the same time. For
that effect, the more intensive intrathecal treatment must also be considered.
Therefore, despite improved CNS-directed therapy in the HRG in ALL-BFM
90, the systemic failure rate was disappointing. As a consequence, in
subsequent trial ALL-BFM 95, alkylating agents and protocol II have
been reintroduced. The therapeutic approach that will push the EFS rate
above 50% in this patient subset has still not been found. In the
German COALL 85/89 study, ALL patients defined as high risk on the
basis of age and WBC count were also treated with rotational
chemotherapy. Patients treated with rapidly alternating treatment
elements fared worse than those who had more continuous drug
exposure.48 The prognostic importance of inadequate
reduction of leukemic blasts in PB was confirmed in St Jude Total
Therapy Study XI, which investigated the early response to various
cytostatic drugs.17 In that study, increased leukemic cell
mass was also found to have an adverse prognostic impact, but age under
1 year or over 10 years was the only identified adverse factor in
patients with B-lineage dis |