Leukemia Research 26 (2002) 909–918
Bone marrow biopsy in adult acute lymphoblastic leukemia:
morphological characteristics and contribution to the study
of prognostic factors
Xavier Thomas a,∗, Quoc-Hung Le a, Catalin Danaı¨la a,
Véronique Lhéritier a, Martine Ffrench b
a Service d’Hématologie Clinique, Hoˆpital Edouard Herriot, 69437 Lyon Cedex 03, France
b Laboratoire d’Histologie et de Cytologie Analytique, Hoˆpital Edouard Herriot, 69437 Lyon Cedex 03, France
Received 22 October 2001; accepted 4 February 2002
Abstract
Bone marrow (BM) sections were examined in 128 untreated adult patients with newly diagnosed acute lymphoblastic leukemia (ALL),
seen in our institution over a 19-year period. BM biopsy was performed in order to assess the incidence, degree and prognostic significance
of histological data of the disease. BM features studied were reticular fibrosis, total cellularity, residual hematopoiesis, mitotic activity,
and blastic infiltration. T-cell lineage ALL were diagnosed in 23% of the cases, while B-cell lineage ALL represented 70% of the cases.
There were 7% of non-T-non-B-cell lineage ALL. The percentage of BM leukemic cells was related to cellularity (P = 0.02), while it
was related to the disappearance of normal cell lines (P < 0.0001). BM cellularity was related to the percentage of circulating leukemic
cells at diagnosis (P = 0.006). Residual hematopoiesis was related to a higher initial granulocyte count (P = 0.04) and lower percentage
of circulating blasts (P = 0.04). The degree of fibrosis was inversely related to that of BM cellularity (P = 0.04). All patients, but four,
received standard ALL induction chemotherapy according to different successive protocols. In this whole cohort of patients, complete
remission (CR) rate was 78%. Median disease-free survival (DFS) and median overall survival (OS) were 13.7 months and 20.2 months,
respectively. In univariate analysis, CR rate was positively affected by mitotic activity (P = 0.01) and residual hematopoiesis (P = 0.008).
OS was positively influenced by a higher leukemic cell mitotic activity (P = 0.03) and the persistence of more than two residual normal
cell lines in BM (P = 0.04). Patients presenting with both of those characteristics had better outcome than patients who did not, as well as,
in terms of CR (P = 0.03), or DFS (P = 0.002), or OS (P = 0.003). T-cell lineage ALL and L3 ALL did not significantly influence those
results. Our findings did not confirm that among marrow features, reticular fibrosis has any prognostic value. A multivariate analysis of both
clinical and histological data was performed to test their prognostic relevance. In a model including age, immunophenotype, Philadelphia
chromosome status, mitotic index, and level of normal residual hematopoiesis, the only significant predictor of CR achievement were the
persistence of normal residual hematopoietic cell lines (P = 0.01) and the mitotic activity of leukemic cells (P = 0.002). Philadelphia
chromosome status (P = 0.03) and age (P < 0.0001) were of prognostic value, respectively for DFS and OS.
We conclude that some characteristics of BM biopsy afford not only descriptive but also prognostic information for predicting the
outcome. The persistence of normal residual hematopoiesis and intense leukemic cells mitotic activity were both factors of favorable
outcome, while BM fibrosis did not display any prognostic value.
© 2002 Elsevier Science Ltd. All rights reserved.
Keywords: Bone marrow biopsy; Fibrosis; Acute lymphoblastic leukemia; Prognosis
1. Introduction
The prognosis of acute lymphoblastic leukemia (ALL)
has improved over the last 20 years, mainly because of more
intensive chemotherapy, improvements in supportive ther-
apy and the introduction of new drugs, and also because
of risk-stratified approach for treatment according to prog-
∗ Corresponding author. Tel.: +33-4-72117395; fax: +33-4-72117404.
E-mail address: xavier.thomas@chu-lyon.fr (X. Thomas).
nostic factors. Parameters measured at diagnosis such as
age, immunophenotype, laboratory values or clinical man-
ifestations can be predictive for either complete remission
(CR) achievement or for remission duration [1,2]. These
features have been utilized to identify patients with low- or
high-risk of relapse, in order to allocate them to regimen with
risk-adapted treatment intensity. Several prognostic models
have been proposed [1,3,4]. High-risk-groups represent the
majority of adult ALL cases. Hoelzer et al. identified time
to CR as greater than 4 weeks, age older than 35 years,
0145-2126/02/$ – see front matter © 2002 Elsevier Science Ltd. All rights reserved.
PII: S0 1 4 5 -2126 (02 )00034 -6
910 X. Thomas et al. / Leukemia Research 26 (2002) 909–918
leukocyte count above 30 × 109/l, and null ALL pheno-
type as poor prognostic features [1]. Based on these criteria,
we recently introduced a new risk-oriented post-remission
treatment policy (collaborative LALA-94 trial) identifying
standard- and high-risk ALL, Philadelphia-positive (Ph+)
ALL, central nervous system-positive (CNS+) ALL.
The diagnosis and classification of ALL are typically
based on morphological and immunophenotypic studies
of peripheral blood (PB) and aspired bone marrow (BM)
leukemic cells. The histological examination of a BM core
biopsy is usually performed only in cases with a dry tap
aspiration or in those cases with inadequate cell numbers
on the smears. However, BM biopsy has also proven to be
of value in predicting the outcome of patients with several
hematological disorders [5–9]. The prognostic value of
BM histologic pattern in ALL has been until now poorly
investigated. However, a progressive increase in marrow
reticulin has been associated with impending relapse [10].
In this setting, ALL presenting with fibrosis features were
until now systematically entered in high-risk-group, and
thus received intensified therapy.
In order to really determine the incidence and the prog-
nostic and therapeutic significance of BM histopathological
parameters seen in adult ALL, we retrospectively reviewed
BM biopsies performed on 128 newly diagnosed patients
seen in our institution over a 18-year period.
2. Materials and methods
2.1. Patients
Between February 1980 and October 1998, a total of 330
consecutive adult patients aged 15 years or more with newly
diagnosed ALL were referred to our institution. During that
time period, BM biopsy was routinely planned at the time
Fig. 1. Percentage of patients with de novo ALL undergoing BM biopsy according to the year of admission. BM biopsy was systematically planned at
diagnosis beginning in February 1980. However, BM biopsy was effectively performed in only about 50% of cases between 1980 and 1991. After 1991,
BM biopsy was performed less and less. Its use was definitively stopped by October 1998 because of its difficult systematic performance.
of diagnosis. Actually, BM biopsy was not systematically
performed (Fig. 1). Overall, BM biopsy was available for
evaluation in 128 patients (39%). Diagnosis of ALL was
based on May-Grünwald-Giemsa (MGG) smears of the BM
aspirates. Leukemic cells were classified according to the
French–American–British (FAB) morphological and cyto-
chemical criteria [11]. Immunophenotyping of leukemic
cells was attempted in patients to define early-pre-B, pre-B,
B-cell, and T-cell subtypes of ALL. The percentage of
positive cells was calculated by counting the number of
positive cells comparatively to the controls. The criteria for
antigen positivity was an expression of surface antigens
by at least 20%, and of intracytoplasmic antigens by at
least 10% of the leukemic cells, as previously described
[12]. Based on the pattern of reactivity, lymphoblasts were
classified as T (CD7+, CD5+ or CD2+), early-pre-B
(CD19+, HLA-DR+, CD10±, CD22±, CD7-, cIg-, sIg-),
pre-B (cIg+), or B (sIg+) [2]. The level of differentiation
into early pre-B stage was evaluated by the presence of
the CD10 and CD20 antigens. Myeloid-antigen-positive
(My+) ALL was defined as co-expression of lymphoid
markers and at least two myeloid-lineage-associated antigen
(CD13/CD14/CD15/CD33) on ≥ 20% of the lymphoblasts.
Cytogenetic analysis was performed on BM and/or on PB
cells before initiation of therapy, using short unstimulated
cultures and RHG banding. Karyotypes were analyzed
according to the International System for Human Cytoge-
netic Nomenclature [13]. Two different classifications were
used. In the first classification, patients were grouped ac-
cording to the presence or absence of normal metaphases
(NN/AN/AA). Twenty normal metaphases were required to
classify a karyotype in the normal group. In the second chro-
mosome grouping, karyotypes were classified according to
ploidy groups or structural (recurrent) abnormalities. Mole-
cular characterization of BCR/ABL, E2A-PBX1, HRX-AF4
rearrangements at diagnosis were routinely performed since
X. Thomas et al. / Leukemia Research 26 (2002) 909–918 911
1992 using a reverse transcription-polymerase chain reac-
tion (RT-PCR) technique. Chromosomal fluorescence in situ
hybridization (FISH) was occasionally used since 1994 with
labeled probes for mBCR/ or MBCR/ABL translocation.
Central nervous system (CNS) leukemia was diagnosed
on the basis of cranial nerve palsies with or without
leukemic blasts in the cerebrospinal fluid, or when at least
five mononuclear cells per microlitre of cerebrospinal fluid
with blasts were present in a cytospin preparation. All
non-T-cell lineage ALL patients were classified as being
at “standard- or higher-risk”of relapse, based on Hoelzer’s
criteria [14] and defined as having at least one of the fol-
lowing factors: Age > 35 years, WBC counts >30 × 109/l,
and Ph+ or 11q23 chromosomal abnormality.
2.2. Treatment
Four of the 128 patients who underwent BM biopsy did
not receive standard ALL chemotherapy. One patient died
early before any chemotherapy could be given. One patient
was treated only with corticosteroids and vincristine because
of advanced age and poor medical condition. Two patients
were misdiagnosed initially with acute non-lymphoblastic
leukemia (ANLL) and treated on an ANLL protocol.
In all 124 other patients, induction chemotherapy con-
sisted of a corticosteroid, vincristine, ±cyclophosphamide,
±anthracycline, ±bleomycin combination. One hundred
and four of those 124 patients were treated according to
seven successive treatments based on therapeutic schedules
of LALA protocols [15–21]. Four patients received L10
schedule [22], eight were treated according to EORTC pro-
tocol 58741 [23], seven with ACVBP schedule [24], and
one according to L3 specific trial [25]. Marrow response
status was determined by BM aspirates at about day 28 of
induction chemotherapy. Patients who did not achieve CR in
one course of chemotherapy, as evaluated by the persistence
of blast in day 28 BM aspirate, received salvage therapy ac-
cording to the protocol design in which they were included.
Patients with CR after induction or salvage were given con-
solidation chemotherapy according to the protocol design
in which they were included. As post-induction therapy,
71 cases followed a chemotherapy program. Adjustments
were made to the dose or timing of consolidation therapy
only under exceptional circumstances. Doses and timing
of maintenance cycles were adjusted frequently to prevent
unacceptable neutropenia or thrombopenia. Twenty-nine
patients, aged less than 50 years, received early BM or PB
stem cell transplantation (SCT). Twenty-four underwent
allogeneic bone marrow transplant (BMT) from a HLA
sibling donor, while five patients received autologous SCT.
2.3. Supportive care
During induction therapy, transfusional support was gen-
erally provided to maintain the hemoglobin level at >80 g/l
and the platelet count at >20 × 109/l. Febrile neutropenic
patients received broad-spectrum antibiotic coverage until
culture and sensitivity results were available. Prophylactic
antibiotics were not administered.
2.4. Evaluation of therapy
CR was defined according to the CALGB criteria as
less than 5% blasts in BM aspirates with evidence of mat-
uration of cell lines and restoration of PB counts [26].
Patients failing induction therapy were categorized as fail-
ures and divided into: (i) death during induction (death
occurred while the patient was receiving induction ther-
apy), and (ii) resistant (the patient survived induction but
resistant leukemia redeveloped). Overall survival (OS) was
defined as the time from diagnosis to death or the last
follow-up. Disease-free survival (DFS) was calculated in
patients who achieved CR from CR to the time of relapse,
death from any cause, or last follow-up. Hematological
relapse was considered when more than 5% blasts were
seen in two BM aspirates obtained at a 15-day interval
[26].
2.5. Bone marrow biopsy
Biopsies were obtained from posterior iliac crest with
a Jamshidi or an Islam needle. Samples were then pro-
cessed into plastic according to the methods previously
described [27]. Briefly, undecalcified bone specimens were
fixed in PBS-buffered formalin, dehydrated in acetone, em-
bedded in a Kushida’s low-viscosity glycol-methacrylate
Immuno-BedTM resin (Polysciences Europe GmbH, Eppel-
heim, Germany) [28]. Sections were cut at a thickness of
2�m using a glass knife in a Sorval JB4 microtome. Sec-
tions were stained with usual histological procedures using
Giemsa and eosin panoptic staining for cytological detail,
and by Gomori’s silver impregnation to show reticulin
fibres. Characteristics recorded from each biopsy sample in-
cluded: total cellularity, mitotic activity, blastic infiltration,
residual hematopoiesis, and reticular fibrosis. The crite-
ria for evaluation and grading of these characteristics are
shown in Table 1. The total cellularity of each BM section
was assessed visually under low-power microscopy and
expressed as a percentage of the total marrow volume. The
leukemic infiltrate was expressed as a percentage observed
among total cells. The stromal fibre content was also graded.
Persisting normal hematopoiesis was evaluated as the num-
ber of persisting normal cell line activity. Persistence of
megakaryocytic, erythroid or granulocytic lineage from
precursors to differentiated cells defined persistent normal
hemopoiesis regarding this specific cell line. Mitotic activ-
ity was expressed as a percentage of mitotic cells observed
among blastic infiltration. All samples were examined at
the time of initial diagnosis at least on five different fields
(50× magnification), and reviewed by the same observer
at time of study. Discordant data were excluded from later
analysis.
912 X. Thomas et al. / Leukemia Research 26 (2002) 909–918
Table 1
Histological data analyzed on bone marrow biopsy
Histological data Grade Characteristic
Reticular fibrosis 0 No reticulin increase
1 Focal, minimal fibrosis
2 Multifocal or diffuse
non-confluent fibrosis
3 Marked, diffuse fibrosis
Total cellularity 1 ≤50%
2 Between 50 and 90%
3 ≥90%
Residual hematopoiesis 0 No persistent cell line
1 Persistence of one-cell line
2 Persistence of two-cell lines
3 Persistence of three-cell lines
Mitotic activity 1 ≤1/400
2 Between 1/400 and 1/100
3 ≥1/100
Blastic infiltration 1 ≤50%
2 Between 50 and 90%
3 ≥90%
2.6. Statistical methods
Univariate analysis of associations between categorical
factors was performed using the χ2 statistics. For contin-
uously distributed variables other than event times, differ-
ences between groups were also tested using the Wilcoxon
rank-sum test. The end points used to assess the prognostic
importance of the different variables were the outcome of
induction chemotherapy, the DFS for patients who achieved
a CR, and the OS duration. The 95% confidence intervals
(CIs) on proportions of CR patients were calculated using
the exact binomial formula. Survival and DFS curves were
estimated by the Kaplan–Meier method, and their 95%
symmetrical CI was calculated according to Greenwood’s
method. For analysis of OS and DFS, patients receiving al-
logeneic or autologous BM transplantation in first CR were
censored at the time of transplantation. Survival curves were
compared using the log-rank test [29]. Only probability val-
ues less than 0.05 were considered statistically significant.
Multivariate analysis was made by a multiple logistic regres-
sion for the CR rate and by the Cox’s proportional hazard
model for DFS and OS [30]. The BM biopsy data selected
for possible inclusion in the regression method were those
for which there was some indication of a significant associ-
ation with survival in univariate analysis. In order to analyze
whether the BM biopsy characteristics have independent
prognostic value, we also included in the multivariate analy-
sis: age, immunophenotype, and karyotype. Those variables
were included because they had proved to be the best combi-
nation of patient’s and disease characteristics for predicting
survival in previous studies [1,17]. Goodness of fit of the
models was tested using the likelihood ratio statistics. Com-
putations were performed using BMDP PC-90 statistical
program (BMDP Statistical Software, Los Angeles, CA,
USA).
In order to assess the prognostic value of BM histological
data, treatment outcome was studied after patients were sep-
arated onto different categories. For the purpose of the sta-
tistical analysis, total cellularity was separated into “poor”
(grade 1 or grade 2) and “rich” (grade 3), mitotic activity
into “moderate” (grade 1 or grade 2) and “intense” (grade 3),
residual hematopoiesis into “null or low” (grade 0 or grade
1) and “persistent” (grade 2 or grade 3), and reticular fi-
brosis into “null or fine” (grade 0 or grade 1) and “coarse”
(grade 2 or grade 3).
3. Results
3.1. Characteristics of patients
Clinical and biological characteristics at diagnosis of the
128 patients (77 males and 51 females) are summarized in
Table 2. A hundred and nine patients were less than 60 years
of age (85%). Six patients (5%) presented with a previous
history of malignant disease. Median leukocyte count of the
entire cohort was 9.5×109/l (range, 0.8–345×109/l) on ad-
mission. About 43% of patients had fever or infection, 20%
had weight loss >5%, 31% had hemorrhage, and 3% pre-
sented with WHO performance status >2. Tumoral syndrome
Table 2
Main clinical and biological characteristics of the 128 adult ALL patients
Characteristics At diagnosis
Clinical characteristics
Age (years) 38 (14–77)a
Sex (males/females) 77/51b
Tumoral syndrome (yes/no/NDc) 81/46/1
CNS involvement (yes/no/ND) 6/120/2
Biological characteristics
Hemoglobin (g/l) 93 (36–176)
WBC count (×109/l) 9.5 (0.8–345)
Platelet count (×109/l) 51 (4–363)
Circulating leukemic cells (%) 47 ( 0–100)
Bone marrow leukemic cells (%) 90 (45–100)
FAB classification (L1/L2/L3/unclassified/ND) 66/34/11/13/4
Immunophenotyped (B/T/non-B-non-T
lineage ALL)
87/28/9
Karyotypee (Ph+/11q23/t(8;14)/
hyperdiploidy ≥50)
14/4/8/8
Risk-groupf (standard-risk/high-risk) 26/74
LDH (UI/l) 775 (176–9980)
Uric acid (�mol/l) 346 (115–905)
Creatininemia (�mol/l) 85 (34–523)
a Median (range).
b Number of patients.
c Not determined.
d Determined in 124 patients.
e Determined in 98 patients.
f According to Hoelzer’s criteria.
X. Thomas et al. / Leukemia Research 26 (2002) 909–918 913
was seen only in 81 out of the 127 cases with available data
(64%), of which 6 cases presented with CNS involvement
(5%). Lymph nodes were palpable in 52%. Liver and spleen
were more than 3 cm from the costal margin in 27 and 37%,
respectively. Mediastinal enlargement was seen only in 12
patients. Among the 124 patients classified in FAB, 66 (53%)
showed L1, 34 (27%) showed L2, and 11 (9%) showed L3
morphology. Thirteen patients (11%) showed undifferenti-
ated leukemia. The immunophenotypic features were deter-
mined in 124 cases. T lymphoid markers were detected in
28 cases (23%), while 87 patients (7