Graft Purging in Autologous Bone Marrow Transplantation: A Promise Not Quite Fulfilled

Graft Purging in Autologous Bone Marrow Transplantation: A Promise Not Quite Fulfilled

ABSTRACT: Clonogenic tumor cells contained within hematopoietic stem cell (HPC) grafts may contribute to relapse following autologous transplantation. Graft purging involves either in vivo or ex vivo HPC manipulation in order to reduce the level of tumor cell contamination. Some phase II trials suggest that patients who receive purged products may have a superior transplant outcome. Phase I trials demonstrate the feasibility of purging methods including ex vivo graft incubation with chemotherapeutic drugs, monoclonal antibodies and complement, and CD34+ cell selection. A phase II trial in follicular non-Hodgkin’s lymphoma demonstrates that patients who receive HPC products purged negative for bcl-2 gene rearrangements have a superior outcome, compared with patients who receive polymerase chain reaction (PCR)-positive products. This finding, however, has not been confirmed in a randomized trial. HPC purging has demonstrated no benefit in a phase III trial in myeloma. Phase II trials in acute myelogenous leukemia show comparable outcomes for patients who receive either purged or unpurged HPC grafts. Limitations of purging include possible progenitor cell loss, delayed engraftment, and qualitative immune defects following transplant. Data to justify routine use of HPC graft purging are insufficient. Phase I and II data support development of phase III trials of both in vivo and in vitro purging methods.

Autologous hematopoietic progenitor
cell (HPC) transplantation
is curative and considered
the standard of care for a number
of malignancies. HPC transplantation
permits the use of extremely intensive,
otherwise myeloablative, therapy.
Unfortunately, even where the
maximal feasible dose intensity is successfully
delivered, disease relapse
remains the primary cause of death
following transplantation.

While therapy-resistant systemic
disease is likely a significant source
for relapse, another source may be
infusion of occult, clonogenic tumor
cells contained within the HPC
graft.[1] Bone marrow harvesting and
collection of blood HPC may lead to
inadvertent collection of tumor cells.
The presence of these cells may be
confirmed through the use of flow
cytometry or molecular diagnostic

Gene-marking studies demonstrate
that tumor cells infused with HPC
grafts may contribute directly to relapse
following transplantation. As
such, graft purging has been a key area
of investigation for several groups.
Molecular diagnostic methods demon-
strate that a daunting array of purging
strategies may reduce the level of graft
contamination by tumor cells. In some
instances, the level of contamination
may be reduced below the limits of
detection. Unfortunately, the ultimate
clinical impact of graft purging remains

In this paper, we will review the
prevalence and significance of tumor
cell contamination of HPC grafts and
describe some methods of graft purging.
In addition, we will describe the
clinical experience using purged autologous
HPC grafts for three paradigmatic
hematologic diseases in which
autologous HPC transplantation plays
a significant role: non-Hodgkin's lymphoma
(NHL), multiple myeloma, and
acute myelogenous leukemia (AML).

Graft as the Source of
Clonogenic Tumor Cells
Prevalence of HPC Graft
Contamination by Tumor Cells

Molecular diagnostic studies permit
the detection of tumor cell contamination
of HPC grafts with a sensitivity
that is 3 to 4 logs greater than histologic
methods or flow cytometry.[2]
Assay sensitivity may be as great as
10-6 when disease-specific oligonucleotide
probes are used for a polymerase
chain reaction (PCR) analysis.[3,4]
With these tools, a wide variety of tumor
cell types may be detected in the
HPC grafts of patients undergoing autologous
transplantation, including
those with multiple myeloma,[5]
NHL,[6] AML,[7] neuroblastoma,
Ewing's sarcoma,[8] chronic myelogenous
leukemia,[9] acute lymphoblastic
leukemia,[9] breast cancer,[10] and
germ cell tumors.[11]

Graft contamination by tumor cells
may be easily detected in patients with
multiple myeloma. In an analysis of
blood samples from 152 patients with
multiple myeloma, phenotypically
abnormal CD19+ cells were identified
in a majority. These cells could be detected
following treatment with conventional-
dose therapy without any
correlation between the tumor cell
number and the paraprotein level;
however, they rose in number in the
setting of progressive disease. In eight
patients for whom PCR analysis could
be used to detect disease-specific rearrangement
in the complementary
determining region III (CDRIII) gene,
clonal rearrangements were detected
in the peripheral blood B cells of all
patients.[12] In a separate study, PCR
analysis demonstrated contamination
by CDRIII-rearranged cells in 90% of
leukopheresis products from patients
with multiple myeloma.[13]

PCR studies have gone even further
to permit quantification of tumor cell-
contamination of HPC grafts. In a
study that used semiquantitative PCR
to assess HPC grafts from 14 patients
with multiple myeloma, the median
level of graft contamination by PCRpositive
cells was 10-4 (range: 10-3 to
10-5).[14] There may be differential
contamination between bone marrow
and blood HPC. In 13 patients with
multiple myeloma who underwent
both bone marrow and chemotherapy/
granulocyte colony-stimulating factor
(G-CSF, Neupogen)-mobilized blood
HPC collection, the median percentage
of clonal cells in blood HPC products
was 0.0024% (range: < 0.0002%
to 0.3520%). In bone marrow products,
the median was 0.74% (range:
0.20% to 6.98%).[15]

Similar analysis for rearrangements
in bcl-2, bcl-1, immunoglobulin
heavy-chain, and T-cell receptor genes
commonly detects tumor cell contamination
in HPC products from patients
with NHL. PCR analysis for the presence
of t(14;18) in 52 patients with
NHL demonstrated that 65% of the
patients had evidence of the rearrangement
in either their blood or bone
marrow prior to blood HPC collection.
HPC products from these patients
were PCR-positive in 29 of 52 patients
(including 4 patients whose blood
and bone marrow were PCR-negative).[
13] Rearranged cells may also
be detected in HPC products from patients
with intermediate- and highgrade
NHL. Among 20 such patients,
PCR analysis detected an abnormal
molecular marker in HPC products
from 17 of 20 patients.[6]

Quantitative PCR studies demonstrate
that significant numbers of
clonal cells may be present in the peripheral
blood and HPC products of
patients with NHL. In one study, 15
of 37 patients with diffuse large-cell
NHL had PCR-detectable rearrangements
of CDRIII. At the time of potential
blood HPC collection, the level
of peripheral blood tumor contamination
ranged from 10-2 to < 10-5 (median:
10-2).[16] In a separate quantification
study, 26 of 28 evaluable patients
had evidence of bcl-2/IgH rearrangements
in either blood or bone
marrow at the time of HPC collection.
The number of rearranged cells ranged
from 1 to approximately 105 rear-
ranged cells per million mononuclear

Graft contamination also occurs in
patients with AML.[9] In a representative
study, HPC products from all 15
patients with AML1/ETO-positive
AML, were PCR-positive on the first
day of autologous HPC collection. Of
the 11 patients who underwent a second
collection procedure, all products
were PCR-positive.[7]

Gene-Marking Studies
In the absence of direct evidence
that HPC contaminants might contribute
to systemic relapse, the aforementioned
data would be a mere curiosity.
Gene-marking techniques using transfection
of HPC grafts with the neomycin-
resistance gene, however, provide
direct evidence that contaminating tumor
cells may contribute directly to
systemic relapse.

Brenner and colleagues showed
that the neomycin-resistance marker
could be detected in leukemic blasts
at the time of relapse by PCR analysis
in two children who had undergone
prior autologous transplantation for
AML.[18] Similar studies have demonstrated
that the neomycin-resistance
marker gene may be detected at the
time of systemic relapse in bcr-abl-
positive cells in patients who have
undergone autologous transplantation
for chronic myelogenous leukemia[
19] and in neuroblasts in patients
with neuroblastoma undergoing autologous
HPC transplantation.[20]
The primary shortcoming of these data
is that they do not indicate the frequency
with which graft contamination
constitutes the principle source for
systemic relapse.

Methods of Graft PurgingIn Vivo Purging
Purging technologies attempt to
eliminate tumor cell contamination of
HPC through direct or indirect manipulation
of the graft. Patients may
be treated in vivo with systemic chemotherapy
and/or monoclonal antibody
(MoAb)-based therapies in the
hope of reducing the whole-body and
circulating tumor cell burden. Thus, in
vivo purging methods attempt to both
improve systemic control of the ma-
lignant disease and alter the kinetics
of tumor cell mobilization.

The paradigm for in vivo purging
is derived from data suggesting that
there may be differential hematopoietic
recovery with preferential mobilization
of normal vs malignant
hematopoiesis following intensive
chemotherapy. Carella and colleagues
treated 30 consecutive chronic myelogenous
leukemia patients with either
ICE (idarubicin [Idamycin], 8 mg/m2/d
for 5 days; cytarabine, 800 mg/m2/d
for 5 days; and etoposide, 150 mg/m2/d
for 3 days) or mini-ICE (the same
agents, but with idarubicin and cytarabine
administered for only 3 days)
mobilization chemotherapy. Philadelphia
chromosome-negative HPC
products were collected in 22 patients.[
21] Other groups have confirmed
there results.

  • Rituximab-The recent availability
    of therapeutic MoAbs provides another
    potential mechanism for in vivo
    purging. Rituximab (Rituxan)-a humanized,
    chimeric MoAb directed
    against the B-cell surface antigen
    CD20-has considerable antidisease
    activity in patients with B-cell NHL
    and may produce molecular remissions
    in some patients with bcl-2-
    positive follicular NHL when used in
    combination with the CHOP regimen
    (cyclophosphamide [Cytoxan, Neosar],
    doxorubicin HCl, vincristine [Oncovin],
    prednisone).[22] Alone or in
    combination with systemic chemotherapy,
    this agent may reduce levels
    of circulating bcl-2-positive cells in
    patients with follicular NHL.[23]

    In a group of 23 patients with follicular
    NHL, rituximab was administered
    to 11 patients in standard doses
    1 week and 2 days, respectively, prior
    to autologous HPC collection. By realtime
    PCR analysis of bcl-2 gene rearrangements,
    6 of 10 HPC products
    collected from patients who received
    rituximab were PCR-negative. In contrast,
    only 1 of 9 products from patients
    mobilized without rituximab
    was PCR-negative.[24]

    In a separate study, 28 patients with
    mantle cell lymphoma underwent
    blood HPC collection after each of two
    cycles of mobilization therapy combined
    with rituximab. Real-time PCR
    analysis of products obtained from 17
    patients after the first cycle (cyclophosphamide,
    7 g/m2, and rituximab,
    375 mg/m2) demonstrated that 42%
    of products were PCR-negative for
    bcl-1 gene rearrangements. Analysis
    of 19 patients after the second cycle
    of therapy (cytarabine, 2 g/m2 every
    12 hours for 6 days, and rituximab,
    375 mg/m2) demonstrated that all
    products were PCR-negative.[25]

    When mobilization kinetics from
    patients with NHL who receive
    rituximab as an in vivo purging regimen
    are compared to historical controls
    who did not receive rituximab,
    the median CD34+ cell yield, colonyforming
    (CFU-GM) and burst-forming unit-
    erythrocyte (BFU-E) appear to be
    comparable.[26] Other investigators,
    however, find a trend toward poorer
    mobilization and engraftment kinetics
    in patients who receive rituximab.[27]

  • Imatinib-Novel agents such as
    imatinib mesylate (Gleevec) show
    promise as in vivo purging agents. In
    a phase II trial of autologous blood
    HPC collection in patients with chronic
    myelogenous leukemia, imatinib was
    administered prior to G-CSF-stimulated
    cell collection. All 10 patients
    achieved a complete cytogenetic response
    and bcr-abl rearranged
    metaphases were suppressed below the
    level of detection by fluorescence in
    situ hybridization (FISH). Nine patients
    achieved their collection goal,
    and HPC products from eight patients
    had a normal karyotype and negative
    FISH studies.[28]

Ex Vivo Purging
HPC grafts may also be manipulated
ex vivo through laboratory procedures
performed after collection.
These ex vivo purging methods may
be used to directly remove or destroy
contaminating tumor cells (negative
selection), or conversely, hematopoietic
progenitors cells may be selected
from the graft (positive selection) and
the remainder of the graft (including
tumor cells) discarded.

  • Chemotherapeutic Agents-One
    of the best-studied negative-selection
    methods involves ex vivo incubation
    of the HPC graft with cytotoxic
    chemotherapeutic drugs such as the
    cyclophosphamide cogeners 4-hydroperoxycyclophosphamide
    (4-HC) and
    mafosfamide or with etoposide and
    corticosteroids. The former two drugs
    are analogous agents, with mafosfamide
    having been used more extensively
    in Europe and 4-HC having
    been the subject of considerable study
    in the United States (including use in
    over 700 patients at Johns Hopkins
    Medical Center). While 4-HC largely
    spares primitive hematopoietic stem
    cells, it produces dose-dependent toxicity
    against tumor cells and may reduce
    committed progenitors such as
    CFU-GM by ≥ 99%.[29] 4-HC purging
    may therefore cause significantly
    delayed neutrophil engraftment
    times.[30] Because direct evidence of
    its effectiveness is limited and was
    never confirmed in a prospective randomized
    trial, the US Food and Drug
    Administration (FDA) chose not to
    license 4-HC.[29]
  • Monoclonal Antibodies-Antibodies-MoAbs
    may be used for ex vivo purging. One
    method involves incubation of mononuclear
    cellenriched HPC product
    with one or more disease-appropriate
    MoAbs and exogenous (typically
    rabbit) complement. Use of a single
    MoAb and complement may achieve
    approximately 3 logs of tumor cell
    depletion, and use of multiple antibodies
    may result in up to 6 logs of
    depletion following three cycles of

    Ball and colleagues used a similar
    method for purging HPC in patients
    with AML. HPC are concomitantly
    incubated with the MoAbs PM-81
    (anti-CD15) with or without the addition
    of AML-2-23 (anti-CD14) and
    complement. In 138 patients transplanted
    between 1984 and 1997, median
    engraftment times for neutrophils
    and platelets were comparable to those
    seen using 4-HC-or mafosfamidepurged

    MoAbs directed against specific
    malignancy-associated surface proteins
    may also be conjugated directed
    to a magnetic particle or, following
    incubation with the target cell, may be
    subsequently bound by species-specific
    antibodies that are in turn conjugated
    to a magnetic particle. Bound
    cells may be thereafter removed by
    passage through a magnetic cell separator.
    In 38 patients with B-cell NHL
    whose bone marrow products underwent
    two cycles of immunomagnetic
    separation, median CD34+ cell recovery
    was 57% (range: 38% to 80%),
    whereas CD19+ cells were reduced by
    a median of 1.8 logs (range: 0.1 to 4.7

    In a separate study, bone marrow
    and blood HPC grafts from 40 patients
    with bcr-abl-positive acute lymphoblastic
    leukemia were treated ex vivo
    with immunomagnetic bead separation.
    The median recovery of CD34+
    cells from both HPC sources was comparable
    to that in the prior study. PCR
    analysis of HPC grafts following pur-
    ging demonstrated a median bcr-abl
    depletion of 2.3 logs and 1 log, respectively.
    At the conclusion of purging, 0
    of 19 bone marrow HPC grafts and 4
    of 17 blood HPC grafts previously
    PCR-positive for bcr-abl became

    Alternative forms of MoAb-based
    negative selection include the use of
    MoAbs conjugated to immunotoxins.[
    35] Other novel means of ex
    vivo negative-selection include photodynamic
    therapy,[36] Adenoviral
    vector-based methods of tumor cell
    cytolysis,[37] antisense therapy,[38]
    incubation with lymphokine-activated
    killer cells,[39] and incubation with
    interleukin (IL)-2 (Proleukin).[40]

  • Cell Selection Devices-Positive
    selection alternatively exploits progenitor
    cell expression of the cell surface
    marker CD34. CD34+ cells may be
    bound by MoAbs that are conjugated
    to a magnetic bead and the bound cells
    captured by an immunomagnetic cell
    collection device such as the Baxter
    Isolex 300i or the Miltenyi Clini-
    MACS. CD34+ cells may be similarly
    bound by biotinylated MoAbs that
    may be captured on an avidin column;
    this was the basis for the CellPro
    CEPRATE device. Cells that do not
    express CD34 antigen are discarded.
    Alternatively, CD34+ cells may be
    sorted using ultra-high-speed fluorescence-
    activated cell sorting.[41]

    Among 51 patients or donors undergoing
    blood HPC mobilization for
    either autologous or allogeneic transplantation,
    products processed with
    Isolex 300i achieved a median CD34+
    purity of 88.9% (range: 47.8% to
    98.3%). The median CD34+ cell recovery
    was 45.1% (range: 13.8% to
    76.2%). CD3+ T cells and CD19+ B
    cells were depleted by 3.7 and 2.8 logs,
    respectively.[42] A separate study
    evaluated the processing of 71 autologous
    and allogeneic blood and bone
    marrow HPC grafts with the Clini-
    MACS; the median CD34+ cell purity
    and CD34+ cell recovery was 97.02%
    (range: 68.3% to 99.7%) and 71%
    (range: 24% to 105%), respectively.[
    43] The Isolex 300i is thus far
    the only device in the United States
    that is approved by the FDA for
    CD34+ cell selection.

Clinical Trials Using
Graft Purging
Non-Hodgkin's Lymphoma
Interest in graft purging was dramatically
stimulated by the 1991 publication
of a Dana-Farber Cancer Institute
trial that evaluated the course
of 114 patients with follicular NHL
who underwent assessment for graft
contamination by malignant cells prior
to autologous transplantation.[31]
PCR analysis detected disease-specific
gene rearrangements in bone marrow
samples from all patients prior to bone
marrow purging. Bone marrow HPC
from patients was purged with three
cycles of incubation with MoAbs and
rabbit complement. Bone marrow
HPC from 57 patients remained PCRpositive
for clonal gene rearrangements
following purging, whereas
products from 57 patients were rendered
PCR-negative. Patients underwent
autologous transplantation following
a preparative regimen that
consisted of total-body irradiation and
cyclophosphamide (60 mg/kg IV daily
for 2 days). At a median follow-up of
23 months, 53 of 57 patients who received
PCR-negative products remained
alive and free of disease, compared
with only 21 of 57 patients who
received PCR-positive products. The
difference between groups was statistically
significant. This effect was seen
in patients transplanted in complete
and partial remission. Using a Cox
proportional-hazards regression analysis,
the risk of relapse was 9.9 times
greater for patients who received PCRpositive

These data were updated in a 1999
review of outcomes for 153 patients
with follicular NHL in first relapse or
incomplete initial remission who underwent
transplantation with the aforementioned
regimen. The projected disease-
free survival at 8 years was 47%,
and the overall survival was 66%. A
subgroup of 113 patients was evaluated
by PCR for evidence of bcl-2 rearrangements.
The projected freedom
from relapse was 89% for patients who
received PCR-negative products vs
19% for patients who received PCRpositive
products.[44] This difference
was statistically significant, and in a
univariate analysis, use of PCR-posi-
tive HPCs was associated with a worse
rate of freedom from relapse.

  • GITMO and CUP Trials-Trials-In the
    phase II Gruppo Italiano Trapianto
    Midollo Osseo (GITMO) study, 92
    patients with previously untreated advanced-
    stage follicular NHL underwent
    autologous HPC transplantation
    following intensive, chemotherapybased
    in vivo purging. Patients were
    treated with three cycles of anthracycline-
    based therapy, and those who
    failed to achieve a complete remission
    received two cycles of platinum-based
    therapy. Patients subsequently received
    consecutive cycles of therapy
    with etoposide (2 g/m2), methotrexate
    (8 g/m2), and cyclophosphamide
    (7 g/m2). G-CSF-mobilized blood
    HPCs were collected upon recovery
    from cyclophosphamide. Of 126
    blood HPC products collected from
    patients with a clonal molecular
    marker, 59 were PCR-negative. Although
    18 of 20 patients who received
    PCR-negative products were in continuing
    complete remission at the time
    of the study's publication, only 9 of
    22 patients who received PCRpositive
    products remained in complete
    remission at that time. The
    difference between subgroups was
    statistically significant.[45]

    The European multicenter CUP
    (Chemotherapy, Unpurged, or Purged
    stem cell transplantation) trial is the
    sole phase III trial that examines ex
    vivo graft purging in patients with
    follicular NHL. Eighty-nine eligible
    patients who had achieved at least a
    partial remission with chemotherapy
    were randomized to receive either
    three cycles of conventional-dose chemotherapy
    (n = 24), transplantation
    with unpurged bone marrow/blood
    HPC (n = 33), or transplantation with
    purged HPC products (n = 32). Purging
    was performed using a combination
    of MoAb and complement. The
    transplant preparative regimen consisted
    of total-body irradiation and cyclophosphamide.
    Unfortunately, the
    trial was closed due to poor accrual
    prior to achieving the goal of 100 patients
    per treatment arm. At a median
    of 26 months following randomization,
    progression-free survival in the
    two transplant arms was superior to
    that in the conventional-dose therapy
    arm, but differences between the two
    transplant arms were not statistically

  • Single-Institution Data-There
    are no prospective randomized data
    evaluating HPC purging in patients
    with aggressive NHL. Fouillard and
    colleagues reported their single-institution
    experience using purged HPC
    transplants for patients with NHL.
    Sixty-four of 120 patients had aggressive
    NHL. Patients received either
    unpurged (n = 21) or purged (n = 43)
    bone marrow or blood HPC grafts.
    Grafts were purged with either mafosfamide
    (fixed or individually tailored
    doses), CD34 selection using the
    CEPRATE SC, or immunotoxinbased
    methods. Results of analysis
    suggest that patients whose grafts were
    more intensively purged with mafosfamide
    had a superior outcome.[47]

    In contrast, 20 patients with aggressive
    NHL who received MoAb-based
    immunomagnetic bead purged autologous
    HPC transplants were retrospectively
    compared with 18 similar patients
    who received unpurged grafts.
    The investigators found no difference
    in outcome between patient groups to
    suggest a benefit to purging.[33]

  • Lack of Firm Conclusions-Data
    from the Dana-Farber group and
    GITMO provide perhaps the best evidence
    justifying the use of purging in
    autologous transplantation. Both trials
    suggest that patients with PCR-positive
    vs PCR-negative HPC collections
    may have divergent outcomes. In both
    of these trials, however, it is unclear
    whether this finding reflects an intrinsic
    effect of the purging process or
    differences in the disease biology between
    patients whose HPC may be
    purged below the limits of detection
    and those in whom this is not the

    The absolute significance of HPC
    purging in relapse prevention seems
    to be futher undercut by the pattern of-
    relapse following transplantation. In
    a group of 99 patients who underwent
    purged HPC transplantation for follicular
    NHL, 26 of 33 patients relapsed
    only at sites of prior disease. This finding
    suggests that resistant systemic disease
    rather than reinfused tumor cells
    is the primary source for systemic failure
    in follicular NHL.[48] Ultimately,
    the impact of biologic variation on the
    capacity to purge to the point of PCR
    negativity and the clinical impact of
    HPC purging can only be determined
    in a sufficiently powered phase III trial.
    Unfortunately, the interim results from
    the sole phase III trial do not support
    the use of purging, and the trial is probably
    insufficiently powered to provide
    a definitive answer.

    There are no global consensus recommendations
    regarding the use of
    purged graft transplants in patients
    with follicular NHL. However, when
    an American Society for Blood and
    Marrow Transplantation (ASBMT)
    expert panel reviewed the role of HPC
    transplantation in patients with diffuse
    large B-cell lymphoma, purging was
    judged to be an "inadequately evaluated
    treatment and recommended for
    comparative study."[49]

  • Pages

    Loading comments...
    Please Wait 20 seconds or click here to close