Download Immune escape from a graft-versus-leukemia effect may play a role

Bone Marrow Transplantation, (1997) 19, 989–999
 1997 Stockton Press All rights reserved 0268–3369/97 $12.00
Immune escape from a graft-versus-leukemia effect may play a role in
the relapse of myeloid leukemias following allogeneic bone marrow
transplantation
S Dermime, D Mavroudis, Y-Z Jiang, N Hensel, J Molldrem and AJ Barrett
BMT Unit, Hematology Branch, NHLBI, National Institutes of Health, Bethesda, MD, USA
Summary:
We studied patients relapsing with myeloid leukemias
following allogeneic bone marrow transplantation
(BMT) for evidence of immune escape by clonal evolution of the leukemia. Relapsed cells from four out of
five patients had a reduced ability to stimulate proliferation of lymphocytes from an HLA-mismatched
responder. There was decreased susceptibility to lysis
by CTL in three and reduced susceptibility to NKmediated lysis in one. Relapsed leukemias had marked
alterations in expression of critical surface molecules
involved in immune responsiveness. Three had
decreased expression of MHC class I and II, with no
change or increase in CD54 (ICAM-1) or CD80 (B7.1).
None of these responded to treatment with donor
lymphocytes. Three patients showed no change, or
increased expression of MHC with no change or
decrease in ICAM-1 or B7.1. Two achieved remission –
one in response to donor lymphocytes and one following
withdrawal of cyclosporine. In one patient transplanted
with myelodysplastic syndrome in transformation,
interferon-gamma upregulated expression of MHC
molecules in relapsed cells and increased their stimulatory capacity and target susceptibility to unmatched
responder lymphocytes. These results suggest that
immune escape through clonal evolution of the leukemia
is a common occurrence in patients who relapse with
myelogenous leukemias after BMT.
Keywords: BMT; GVL; relapse; CML; AML; immune
escape
Although allogeneic BMT from an HLA-identical sibling
donor has a high curative potential in the treatment of
myeloid leukemias, relapse of the disease occurs in about
20% of acute myeloid leukemias (AML) and 15% of
chronic myelogenous leukemias (CML).1,2 The risk of
relapse has been linked to the degree to which the allograft
can exert a graft-versus-leukemia (GVL) effect against
residual leukemia remaining after the myeloablative preparative regimen. Clinical data supports a role for T lymCorrespondence: Dr AJ Barrett, Building 10, Room 7C-103, National
Institutes of Health, 9000 Rockville Pike, Bethesda MD 20892-1652, USA
Received 21 October 1996; accepted 26 January 1997
phocytes in the GVL response: relapse rates are higher after
T cell-depleted BMT3–5 while remissions can be reinduced
with donor lymphocyte transfusions in patients who relapse
after BMT.6 There is also experimental evidence for the
involvement of NK cells in GVL.7,8 It is therefore generally
assumed that leukemic relapse after BMT is related to
qualitative or quantitative defects in donor immune function.9 One reason for the failure of GVL could be immune
escape by clonal evolution in the leukemia which selects
malignant cells that can avoid immune recognition. Malignancies that resist the effects of cellular immunotherapy
have been demonstrated in murine models10,11 and suggested in man.12 As described in other tumors, the failure
to recognize leukemia cells directly may result from lack
of expression of MHC-complex molecules,13–16 lack of a
suitable tumor antigen,17 defective antigen processing,18,19
production of inhibitors that actively suppress antitumor
response,20 lack of the appropriate costimulatory molecules
necessary to induce an immune response,21–23 downregulation of CD54 (ICAM-1)24,25 or lack of expression of CD95
(Fas) antigen,26,27 a cell surface molecule involved in
apoptosis.
In this report, we compared the immunogenicity of
myeloid leukemia cells before BMT and at relapse in
patients with AML, myelodysplastic syndrome (MDS) and
CML. Using mismatched responder lymphocytes we used
proliferative and cytotoxic alloimmune responses to detect
and quantitate defects in lymphocyte stimulation and target
susceptibility of leukemia cells. We also compared the
immunophenotype of the leukemia before and after relapse
using a direct immunofluorescence assay to measure critical
molecules involved in antigen presentation and lymphocyte
responses: MHC class I and II (DR and DQ), ICAM-1, Fas
and B7.1. We present data supporting the possibility that
leukemias relapsing after BMT are clonally selected to
evade donor-derived immune control.
Patients and methods
Patients and donors
Between September 1993 and January 1996, 28 patients
with acute or chronic myeloid leukemia or myelodysplastic
syndrome received a bone marrow transplant from an HLAmatched sibling. Six developed hematological relapse of
leukemia and form the basis of this study (Table 1). After
Mechanisms of escape from GVL by relapsed leukemia cells
S Dermime et al
990
informed consent, cells from patients with leukemia and
from healthy donors were obtained from the peripheral
blood or bone marrow. All patients received a BMT from
serologically HLA-A, B and DR-matched donors. The
BMT dose of T lymphocytes was adjusted to 2 × 105/kg
CD3+ cells by elutriation, but subsequently all patients
received a donor lymphocyte add-back of 2 × 106 and
5 × 107 CD3+ cells/kg on days 30 and 45 post-BMT. All
patients received cyclosporin GVHD prophylaxis from day
−4 to day +100 post-BMT. The cells were separated using
Ficoll–Hypaque gradient-density (Organon Teknika, Durham, NC, USA) and subsequently frozen in RPMI-1640
complete medium (CM, 25 mm HEPES buffer, 2 mm l-glutamine, 20% gentamicin; GIBCO BRL, Gaithersburg, MD,
USA) supplemented with 20% heat inactivated fetal bovine
serum (FBS; Atlanta Biologicals, Norcross, GA, USA), or
20% heat-inactivated pooled human AB serum (HS; PelFreez, Brown Deer, WI, USA) and 10% DMSO according
to standard protocols. Before use, the cells were thawed,
washed and suspended in CM + 10% HS.
Cell lines
Lymphoblastoid cell lines (LCL) were obtained by incubating 107 peripheral blood mononuclear cells (PBMC) for 1 h
at 37°C with 1 ml of supernatant from the Epstein–Barr
virus (EBV)-producing cell line B95-8 (ATCC, Rockville,
MD, USA) and then culturing in CM + 10% FBS. Cyclosporin A (2 mg/ml) was added to the cultures during the
first 4 weeks. LCL were used after 8 weeks of culture. The
K562 NK-sensitive leukemia cell line (ATCC) was maintained in CM + 10% FBS.
Differential cell count of patients’ leukemia samples
The percentage of leukemia cells from Ficoll–Hypaque separated samples were determined by examination of cytospins of pre-BMT and relapsed bone marrow cells stained
with Wright’s Giemsa. Two hundred cells per sample
were counted.
Proliferative assay
The proliferative assay was performed as previously
described. 28 In brief, 105 responder mismatched normal
individual (D1) PBMCs were cultured with 105 5000 cGY
irradiated pre-BMT or relapsed leukemic cells as stimulators. Patient and donor HLA typing data are shown in
Table 2. Before the test, the stimulators were thawed,
washed and cultured for 24 h in the presence or absence
of 103 U/ml recombinant interferon-g (IFN-g; Biosource,
Camarillo, CA, USA). The test was set up in six replicates
in 96-well U-bottom plates in a total volume of 200 ml CM
+ 10% HS in each well. Controls were responder cells with
10 000 cGY irradiated K562 cells, responder cells alone
and stimulators alone. Cultures were incubated at 37°C in
5% CO2 for 5 days and pulsed for the last 18 h with
1 mCi/well of 3H-labeled methyl thymidine (3HTdR; Amersham International, Amersham, UK). Cells were harvested
and 3HTdR incorporation was measured in a beta-counter
(Wallac, Gaithersburg, MD, USA). The results are
expressed as mean counts per minute (c.p.m.) ± s.d.
Cell-mediated cytotoxicity (CMC)
Cytotoxic T lymphocyte (CTL) activity: A total of 107 normal PBMCs from donor D1 (used in the proliferative assay)
were cultured with 107 5000 cGY irradiated stimulating
pre-BMT cells in 25-cm2 flasks containing 10 ml CM +
10% HS. Recombinant interleukin-2 (IL-2; Biosource) was
added to the cultures on day 3 and day 7 at 60 IU/ml. On
day 10, lymphocytes were harvested, resuspended, and serially diluted to be used in the CTL assay as effectors.
Natural killer (NK) cell activity: Fresh PBMCs isolated
from the heparinized peripheral blood of a normal individual donor D2 were separated by gradient-density using Ficoll–Hypaque and used immediately as effectors in the cytotoxicity assay against pre-BMT and relapsed leukemia
cells.
Cytotoxicity assay: A semi-automated mini-cytotoxic
assay was used.29 Effector cells (T lymphocytes or NK
cells) were prepared in doubling dilutions from 6 × 103 to
24 × 103 or 25 × 103 to 105 cell/well and plated in 40 ml,
60-well Terasaki trays (Robbins Scientific, Mountain View,
CA, USA) with six replicates per dilution. Target cells (preBMT or post-BMT leukemia cells, K562 cells) at a concentration of 2 × 106 cell/ml were stained with 10 mg/ml of
Calcein-AM (CAM; Molecular Probes, Eugene, OR, USA)
for 30 min in a 37°C water bath. After washing three times
in CM + 10% HS, target cells were resuspended at 105
cell/ml for K562 and 2 × 105 cells/ml for leukemia cells
(103 or 2 × 103 target cells in 10 ml medium were added to
each well containing effector cells). Wells with target cells
alone and medium alone were used for maximum (Mx) and
minimum (Mn) fluorescence emission, respectively. After
4 h incubation at 37°C in 5% CO2, 5 ml FluoroQuench (EB
Stain-Quench Reagent; One Lambda, Canoga Park, CA,
USA) was added to each well and the trays were centrifuged for 1 min at 60 g before measurement of fluorescence
using an automated Lambda FluoroScan (One Lambda). A
decrease in the fluorescence emission is proportional to the
degree of lysis of target cells, once the released dye is
quenched by hemoglobin in the FluoroQuench reagent. The
percentage of lysis was calculated as follows:
% lysis = (1 − [(mean experimental emission − mean
Mn) / mean Mx − mean Mn]) × 100.
Antibodies and flow cytometry
Phenotypes of pre-BMT and post-BMT leukemia cells were
determined by flow cytometry. Cells (106) were incubated
with labeled monoclonal antibodies on ice, in the dark for
30 min. After washing three times in phosphate-buffered
saline (PBS), cells were analyzed by flow cytometry
(Becton Dickinson, San Jose, CA, USA). The fluorescein
isothiocyanate (FITC)-conjugated monoclonal antibodies
used were: HLA class I (CALTAC Laboratories, San Francisco, CA, USA), HLA class II DQ (Becton Dickinson)
Mechanisms of escape from GVL by relapsed leukemia cells
S Dermime et al
Table 1
Patient characteristics
Patient
(UPN)
Sex/
Diagnosis
Age
(years)
991
Status
pre-BMT
Status at
relapse
Karyotype
pre-BMT
Karyotype
at relapse
Relapse
on/off
CsA
Acute
GVHD
grade
Treatment
of
relapse
t(12;14), del8,
t(4;21) complex
abnormalities
5q−, add1, inv4
inv X, del19
t(9;22), t(1;12),
t(2;20), t(3;8)
trisomy 20,
monosomy 22
t(9;22), t(4;17),
t(7;10), t(2;5)
del16q
t(9;22)
on
II skin
DLT + IL-2
NR, died day
76
off
I skin
DLT + IL-2
off
0
DLT + IFN-a
DLT + IL-2
second BMT
NR, died day
510
CR, alive day
330
off
0
DLT + IFN-a
CR, alive day
450
on
0
DLT + IL-2
NR, died day
110
on
II gut
Stop CsA
CR, alive day
320
P1 (029)
M/25
AML
M0 4th rel
M4
NA
P2 (005)
F/51
AML
CMML/tMDS
M5
5q−
P3 (025)
M/23
CML
CP
AP
t(9;22)
P4 (019)
F/34
CML
CP
CP
t(9;22)
P5 (017)
M/18
CML
BC (myeloid)
P6 (026)
F/29
CML
MF
BC
(myeloid)
MF
t(9;22),
trisomies,
+8, +21
t(9;22) t(9;22)
Outcome
AML = acute myeloid leukemia; CML = chronic myeloid leukemia; CMML = chronic myelomonocytic leukemia; tMDS = myelodysplastic syndrome
in transformation; CP = chronic phase; AP = accelerated phase; BC = blast crisis; MF = myelofibrosis; CsA = cyclosporin A; DLT = donor lymphocyte
transfusion; NA = not available; IL-2 = interleukin-2; IFN-a = interferon-alpha; NR = no response; CR = cytogenetic remission.
Table 2
HLA typing of patients and third-party donors (D1) used in
the mixed lymphocyte cultures and T cell-mediated cytotoxicity and D2
as a source of NK cells
Patients
(UPN)
P1
P2
P3
P4
P5
P6
(029)
(005)
(025)
(019)
(017)
(026)
Donors
D1 (0331)
D2
HLA typing
HLA-A
HLA-B
1, 3
24, 29
1, 2
1, 2
2, 26
2, 26
7,
39,
35,
63,
13,
38,
2, —
2, 23
13, 27
58, 62
8
44
62
—
47
50
HLA-DR HLA-DQ
8, 11
7, 15
10, 14
3, 8
7, —
13, 7
4, 7
2, 6
5, 52
6, 4
2, 9
6, 9
1, 7
4, 7
2, 5
2, 8
Numbers in bold represent the HLA shared between each patient and the
third-party donor D1.
and HLA class II DR, CD54, CD80 and CD95 (all from
ImmunoTech, Marseille, France). Phycoerythrin (PE)-conjugated anti-CD33 (ImmunoTech) was used to select for
CD33+ cells. Controls were CD33+ bone marrow cells,
LCLs from normal donors, and K562 cells. Isotypematched irrelevant monoclonal antibodies were used as
controls.
Results
Patient details and relapse data are shown in Table 1. HLA
typing of patients and responders are shown in Table 2.
Patient 029 (P1) was in a fourth relapse at BMT with an
AML M0. At relapse however, the leukemia had M4 morphology and t(12;14), t(4;21) and del(8) karyotype. Patient
005 (P2) was transplanted with tMDS (.65% blast cells)
and relapsed with 100% blast cells showing M5 morphology and 5q−, invX, inv4, add1 and del19 karyotype.
She had not received additional chemotherapy prior to
BMT or at the time the sample was taken at relapse. Patient
025 (P3) with CML chronic phase t(9;22) relapsed into
accelerated phase with additional chromosomal rearrangements (t9;22, t1;12, t2;20, t3;8, trisomy 20, monosomy 22).
At the time of collection of the pre-BMT and relapsed
samples he was not receiving chemotherapy. Patient 019
(P4) was transplanted in chronic phase CML and relapsed
into chronic phase. The karyotype at relapse showed clonal
progression from a t9;22 karyotype to include t4;17, t7;10,
t2;5 and del16q. Patient 017 (P5) was transplanted in
myeloid blast crisis CML and appeared morphologically
unchanged at relapse. Patient 026 (P6) with myelofibrosis
from Ph+ CML was transplanted and appeared morphologically and karyotypically unchanged at relapse.
In order to determine whether the composition of
pre- and post-BMT leukemic stimulator/target cells was
comparable, we performed differential counts of stained
cytospins. No important differences in the cellular composition of the leukemic stimulator/target cells were found
(Table 3).
Stimulation of third party lymphocytes by leukemia cells
before and after BMT
Paired samples of Ficoll-purified leukemia cells obtained
before BMT and at relapse after BMT were tested for their
ability to stimulate proliferation of lymphocytes from a
third party responder (D1 in Table 2). Proliferation against
K562 cells (not expressing HLA class I and II) was used
as a negative control to calculate the stimulation index
(SI = c.p.m. of responder vs leukemia cells/c.p.m. of
responder to K562 cells). Results are shown in Figure 1.
K562 cells induced background level proliferation
Mechanisms of escape from GVL by relapsed leukemia cells
S Dermime et al
Table 3
Differential counts of leukemic stimulators/targets
Patients (UPN)
status
P1(029)
P2(005)
P3(025)
P4(019)
P5(017)
Pre
Post
Pre
Post
Pre
Post
Pre
Post
Pre
Post
77
79
86
96
28
17
38
24
91a
92a
% Monocytes
—
—
1
—
51
57
23
38
—
—
% Lymphocytes
—
3
5
—
11
9
18
19
5
3
% Polymorphs
(NEB)
18
14
8
4
10
17
18
18
4
5
Cell type
% Blasts/myelocytes
a
Monocytoid blasts and abnormal monocytes.
50
3HTdR incorporation (c.p.m. × 103)
992
45
Pre-BMT
Relapse
14
12
10
8
6
4
2
0
40
35
30
P2 relapse
P2 relapse +
IFN- γ
25
20
15
10
5
0
P1
P2
P3
P4
P5
K562
Stimulators
Figure 1 Proliferative responses of pre-BMT and relapsed leukemia
cells to third-party T lymphocytes D1. Responder cells (10 5/well) cultured
for 5 days with 5000 cGY irradiated leukemia cells (105 /well), incubated
for additional 18 h with 3HTdR, and harvested. Relapsed leukemia cells
from P2 were incubated with IFN-g (103 U/ml) for 24 h before the test
(inset). Controls were responder cells with 10 000 cGY irradiated K562
cells. Counts per minute (c.p.m.) values represent the mean of six
replicates ± s.d.
(c.p.m. , 100 c.p.m.). Leukemias pre-transplant differed in
their ability to stimulate a third party with absolute c.p.m.
varying from 6000 ± 1000 to 40 000 ± 2000. Post-transplant stimulation was generally lower from , 100 to
27 5000 ± 2000 c.p.m. Leukemia cells at relapse post-BMT
induced lower proliferative responses than the pre-transplant leukemia in four patients with a ratio of pre:post
stimulation indices of 1:0.04 (P1), 1:0.59 (P2), 1:0.7 (P3)
and 1:0.07 (P4). One patient P5 who was transplanted in
blast crisis of CML and relapsed into blast crisis induced
comparable low proliferation before and after transplant
(6000 c.p.m.).
Pre-transplant and post-transplant leukemia samples
were incubated with IFN-g (103 U/ml) for 24 h to test for
induction of stimulatory ability to the third party lymphocytes. IFN-g treatment of P2 relapsed cells induced a sevenfold increase in stimulatory capacity (Figure 1 inset). There
was no increase in the stimulatory capacity of the other
relapsed leukemias.
Susceptibility to lysis of leukemia cells by CTLs before
and after BMT
Irradiated pre-transplant leukemia samples were used to
stimulate the lymphocytes from the HLA-mismatched
responder D1. After 10 days of culture with IL-2 addition
and re-feeding, the responder cells were tested for their
ability to lyse paired pre- and post-transplant leukemia
samples. Results are shown in Figure 2. At an E:T ratio of
24:1 there was a 27 to 44% lysis of pre-transplant targets.
Lysis of post-transplant leukemias was markedly reduced
in three patients P1, P2 and P5 (4.6, 6.6 and 3.6%
respectively). There was no change in the target susceptibility in patients P3 and P4. In patient P2, incubation of
the post-transplant but not the pre-transplant leukemia targets with IFN-g induced an increase in the lysis of the posttransplant leukemia to almost pre-BMT levels (E:T at 24:1
rose from 6.6 to 30%). There was no increase in the lytic
activity against other pre-BMT and relapsed leukemias.
Susceptibility to NK cell-mediated lysis of leukemia cells
before and after BMT
Fresh normal PBMC was used as a source of NK cells.
Standard NK assays were carried out at three E:T ratios,
against paired pre- and post-transplant leukemia cells using
K562 cells as a positive control. The results for four
patients tested are shown in Figure 3. In three both preand post-BMT leukemia samples gave comparable results
and were lysed more readily than K562 cells. In one (P4)
the relapsed sample was lysed less effectively than K562
cells and considerably less than the pre-transplant leukemia.
Phenotype of leukemia cells before and after BMT
The alterations in stimulatory ability and target susceptibility of leukemia at relapse after BMT prompted us to
examine the possibility that escape from immune surveillance had occurred by downregulation in surface molecules
involved in immune responses. We therefore compared
expression of MHC (class I and class II), CD45 (ICAM1), CD80 (B7.1) and CD95 (Fas) in CD33+ gated cell populations of pre-transplant and post-transplant relapsed leukemias. An additional patient (P6) was also studied. Results
of flow cytometry were expressed as % positive cells and
Mechanisms of escape from GVL by relapsed leukemia cells
S Dermime et al
993
50
P2
P2 (IFN)
Pre-BMT
Pre-BMT+ IFN-γ
Relapse
Relapse+IFN-γ
K562 cells
40
30
20
10
0
0
6
12
18
12
18
24
50
P3
% Lysis
6
0
24
50
P4
40
40
30
30
20
20
10
10
0
0
0
6
12
18
0
24
6
12
18
24
6
12
18
24
50
50
P5
P1
40
40
30
30
20
20
10
10
0
0
0
6
12
18
24
0
E:T ratio
Pre-BMT
Relapse
K562 cells
Figure 2 Susceptibility to cell-mediated cytotoxicity of pre-BMT and relapsed leukemia cells by normal third-party lymphocytes from D1. Effector
cells at dilutions of 6 × 103 –48 × 103 cell/well/20 ml were plated with 2 × 103 pre-BMT or relapsed leukemia cells as targets in a 4-h cytotoxic assay.
Six replicates were used for each dilution. Leukemia cells from P2 were incubated with IFN-g (103 U/ml) for 24 h before the test (inset). K562 target
cells were used as a control.
fluorescent intensity (FI) and calibrated against EBV LCL
(strong positivity) and K562 (weak to zero positivity).
HLA class I expression: With the exception of K562 cells
all samples tested contained more than 5% MHC class I
positive cells and in most cases the percentage of positive
cells was over 90% (Figure 4a). The fluorescence intensity
(FI) of MHC class I was in general much lower than that
of EBV-LCL cells. In P1, P2 and P3, FI was in the region
of 30% of EBV-LCL pre-transplant, and fell to barely
detectable levels after BMT. In P3, the post-BMT sample
also showed a major reduction in the percentage of cells
expressing MHC class I. P4 and P5 showed no difference
in FI before and after BMT. An increase in MHC class I
expression from 60 to 100% post-transplant with a corresponding increase in FI occurred in P6. Incubation with
IFN-g modestly increased FI in most samples.
HLA class II DR expression: HLA class II DR was generally less well expressed on leukemia cells, both in terms of
% positive cells and their FI, compared with EBV-LCL
(Figure 4b). Only patient P5 expressed high levels of DR.
Comparing pre- and post-relapse samples, there were three
discernable alterations in HLA DR expression: (1) lower
percentage of DR expressing cells at relapse after BMT
with almost undetectable FI at relapse (P1, P2 and P3); (2)
slightly higher percentage of cells expressing HLA DR
after BMT but with extremely low FI (P4 and P6); and (3)
P5 with high DR expression and high FI pre- and postBMT. IFN-g had no effect on the percentage of cells
expressing HLA DR but increased FI in P5 (pre- and postBMT) and P1 post-BMT.
HLA class II DQ: Compared with EBV-LCL the % and
the FI of leukemia cells expressing HLA DQ was generally
Mechanisms of escape from GVL by relapsed leukemia cells
S Dermime et al
994
100
100
% Lysis
P1
P2
80
80
60
60
40
40
20
20
0
0
0
100
25
50
75
100
0
100
P4
80
80
60
60
40
40
20
20
0
25
50
75
100
25
50
75
100
P5
0
0
25
50
75
100
0
E:T ratio
Pre-BMT;
Relapse;
K562 cells
Figure 3 Lytic susceptibility of pre-BMT and relapsed leukemia cells to NK cells from D2. Effector cells at dilutions to achieve 25 × 103–200 × 103
cell/well/20 ml were plated with 2 × 103 pre-BMT or relapsed leukemia cells as targets in a 4-h cytotoxic assay. Six replicates were used for each dilution.
K562 target cells were used as a control.
very low (Figure 4c). Only P3 and P5 had more than 10%
cells expressing DQ, with FI above the background for
K562 cells. Comparing pre- and post-BMT samples, there
were two changes: (1) low to undetectable % expression
and FI (P1, P2, P4 and P6); and (2) moderate % expression
and FI (P3 and P5). DQ expression in P3 was lower at
relapse post-BMT, while in P5 expression and FI increased
at relapse. IFN-g did not generally affect DQ expression
or FI. The exception was P2 who showed an increase in %
cells expressing DQ after but not before BMT.
CD54 (ICAM-1) expression: Expression of ICAM-1 both
in percentage of marked cells and FI was variable (Figure
4d). Some leukemias marked as strongly as EBV-LCL,
while in others expression was less than on K562 cells.
Comparing pre- and post-BMT samples, three variations
were observed: (1) low % expression pre-BMT, with
increase at relapse (P1, P2 and P3); (2) lower % expression
post-BMT (P4 and P6) and (3) high expression with no
change (P5). There were no clear differences in FI before
and after BMT. In general, IFN-g did not alter ICAM-1
expression in leukemia cells. In P1 post-BMT and P5 preBMT however IFN-g increased FI.
CD80 (B7.1) expression: B7.1 was strongly expressed on
EBV LCL but was less than 10% in K562 cells. Expression
was variable in the leukemias (Figure 4e). Comparing pre-
and post-relapse samples P1, P4, P5 and P6 had lower %
expression and usually lower FI after BMT, whereas P2
and P3 showed increased % expression after BMT but
decreased FI. IFN-g modestly increased B7.1 expression
and FI in some samples.
CD95 (Fas) expression: Fas was variably detectable on
leukemia cells. Notably P5 had extremely strong Fas
expression (Figure 4f). Comparing pre- and post-BMT
samples P1, P3, and P5 had moderate to high expression
(20–100%) and FI above 20 which did not change at
relapse. P2 and P6 showed an increased % of cells expressing Fas at relapse with no change in FI. P4 Fas expression
fell from around 80 to 35% post-BMT with no change in
FI. IFN-g increased Fas expression in P1 post-relapse cells.
Correlation of changes in phenotype with immune
function and outcome of immune manipulation
Table 4 summarizes the alterations in immune characteristics of the six patients tested and correlates results with
changes in immune phenotype and clinical outcome.
Behavior at relapse can be categorized into two broad
groups.
Loss in MHC class I and II expression (P1, P2 and
P3): Reduction in stimulatory capacity of relapsed cells
Mechanisms of escape from GVL by relapsed leukemia cells
S Dermime et al
995
a
LCL
100
Pre-BMT
Relapse
80
–IFN-γ
40
% Positive cells
Fluorescence intensity (Fl)
60
20
0
K562
P1 P2 P3 P6 P4 P5
LCL
100
80
60
40
+IFN-γ
20
0
K562
100
LCL
1800
1600
1400
1200
1000
800
600
400
200
0
LCL
K562
P1 P2 P3 P6 P4 P5
1800
1600
1400
1200
1000
800
600
400
200
0
LCL
K562
b
2400
2200
Pre-BMT 2000
Relapse 1800
1600
80
–IFN-γ
% Positive cells
40
20
K562
0
P1 P2 P3 P6 P4 P5
LCL
100
80
60
40
Fluorescence intensity (FI)
60
+IFN-γ
20
K562
0
LCL
1400
1200
1000
800
600
400
200
0
K562
P1 P2 P3 P6 P4 P5
2400
2200
2000
1800
1600
1400
1200
1000
800
600
400
200
0
K562
320
LCL
LCL
c
LCL
100
Pre-BMT 280
Relapse 240
80
–IFN-γ
% Positive cells
40
20
K562
0
P1 P2 P3 P6 P4 P5
100
LCL
80
60
40
+IFN-γ
20
0
K562
Fluorescence intensity (FI)
60
200
160
120
80
40
0
K562
P1 P2 P3 P6 P4 P5
320
280
240
200
160
120
80
40
0
LCL
K562
Figure 4 HLA class I (a), class II DR (b), class II DQ (c), ICAM-1 (d), B7.1 (e) and Fas (f) expression by pre-BMT and relapsed leukemia cells
analyzed by flow cytometry. Percentage positive cells (%) and fluorescence intensity (FI, median channel of fluorescence (log)) were measured in
untreated (−IFNg) and IFN-g treated (+IFNg) cells. Mean ± s.d. of percentage marked cells and FI from three × LCL lines and K562 were used as high
and low controls (horizontal grey bars). Differences between pre- and post-BMT and + or − IFNg samples were considered significant if they exceeded
2 s.d. of LCL controls.
and loss of susceptibility to CTL, but not NK cytotoxicity
in the two patients tested, was associated with no significant
change, or an increase in B7.1, ICAM-1 and Fas. None of
these three patients responded to treatment with donor lymphocytes.
No change (P4) or an increase in MHC class I and II
expression (P5 and P6): Loss of susceptibility to CTL or
NK cells, with low or reduced stimulatory capacity, was
associated with loss of B7.1, no change or loss or ICAM1 and variable changes in Fas. Two out of three patients
Mechanisms of escape from GVL by relapsed leukemia cells
S Dermime et al
996
d
100
LCL
60
% Positive cells
40
20
0
P1 P2 P3 P6 P4 P5
100
LCL
80
60
40
+IFN-γ
20
Fluorescence intensity (FI)
120
Pre-BMT 100
Relapse
80
60
–IFN-γ
40
20
K562
0
80
K562
0
LCL
K562
P1 P2 P3 P6 P4 P5
120
100
80
60
40
20
0
LCL
K562
e
100
100
LCL
60
–IFN-γ
% Positive cells
40
20
0
K562
P1 P2 P3 P6 P4 P5
100
LCL
80
60
40
Fluorescence intensity (FI)
Pre-BMT 80
Relapse
80
+IFN-γ
20
LCL
40
K562
20
0
P1 P2 P3 P6 P4 P5
100
80
60
LCL
40
K562
20
K562
0
60
0
f
100
100
LCL
80
60
–IFN-γ
% Positive cells
40
20
0
100
P1 P2 P3 P6 P4 P5
K562
LCL
80
60
+IFN-γ
40
Figure 4
60
40
K562
LCL
K562
20
0
P1 P2 P3 P6 P4 P5
100
80
60
40
20
20
0
Fluorescence intensity (FI)
Pre-BMT 80
Relapse
LCL
K562
0
(Continued).
responded to immune manipulation (one to DLT (P4)) and
one to withdrawal of cyclosporine immunosuppression.
Discussion
It has long been considered that malignant cells may
undergo clonal selection to escape from immune regulation
and there is circumstantial evidence supporting this hypothesis.10–12 The demonstration of a GVL effect following allo-
geneic BMT and the effect of immune manipulation to
induce remissions in patients relapsing after BMT (T celldepletion, DLT, stopping CsA, IFN-a) strongly supports
a role of the donor immune system in regulating residual
leukemia after BMT. Leukemic relapse following allogeneic BMT is usually attributed to a failure of the preparative regimen or the donor immune system to control
residual disease. An alternative possibility however is that
the leukemia may have evolved to escape from immune
control under the strong negative selection pressure of an
Mechanisms of escape from GVL by relapsed leukemia cells
S Dermime et al
Table 4
Correlation of changes in phenotype with immunogenicity of leukemia and outcome of immune manipulation with donor lymphocytes or
withdrawal of cyclosporine
Diagnosis at relapse
Phenotype
Functional tests
Response to treatment
P1
(029)
AML M4
P2
(005)
MDSt
P3
(025)
CML AP
P4
(019)
CML CP
P5
(017)
CML BC
P6
(026)
CML MF
MHC class I
HLA DR
HLA DQ
↓
↓
L
↓
↓
L
↓
↓
↓
0
0
0
0
0
0
↑
↑
0
ICAM-1 (CD54)
B7.1 (CD80)
fas (CD95)
↑
L
0
↑
↑
↑
↑
↑
0
0
↓
0
↓
↓
L
↓
↓
↑
stimulation
CTL lysis
NK lysis
↓
↓
0
↓
↓
0
↓
0
NT
L
↓
0
↓
↑
↓
NT
NT
NT
No
No
No
No
Yes
Donor lymphocytes
Yes
Cyclosporine withdrawal
At relapse: ↑ = increased; ↓ = decreased; 0 = no change; L = no change but low; NT = not tested.
alloreactive donor immune system. Only one report has
associated leukemic relapse after allogeneic BMT with
immune escape of the leukemia.12 In this study, relapse
occurred after an HLA-haploidentical BMT in a patient
with acute lymphocytic leukemia. The weak mixed lymphocyte culture and undetectable primary cell-mediated
cytotoxicity responses of the donor to the post-BMT relapse
cells suggested that the relapsed leukemia may have been
selected in vivo for resistance to the donor immunity.
We argued that immune escape leading to relapse after
BMT could occur either because the leukemia becomes
nonstimulatory to donor T cells or ceases to be susceptible
to destruction from effector T cells or NK cells. We therefore examined six patients with various myeloid leukemias
who relapsed with leukemia after HLA-matched sibling
BMT, comparing pre-transplant leukemia with the relapsed
leukemia, for its ability to stimulate an immune response
and to be a target for T cell and NK cell attack. We showed
that in four of five leukemias, the ability to stimulate a third
party lymphocyte proliferative responses was significantly
reduced. In the fifth patient the stimulatory capacity of the
leukemia was low pre-transplant and remained unchanged.
By using the normal donor as an HLA-matched control for
the third party responder, we showed that the response was
significantly lower than that of a normal stimulator cell.
Using a similar approach we used pre-transplant leukemia
cells to stimulate CTL from an HLA-mismatched normal
responder. (We chose pre-transplant cells to induce CTL,
since they were shown to be capable of stimulating a third
party proliferative response.) After 10 days, CTL generated
in this way had no LAK activity as demonstrated by their
inability to lyse K562 cells. However they showed specific
cytotoxic activity to the stimulating leukemia. Using a
second normal responder, the susceptibility to NKmediated lysis was also tested. Leukemia cells from all
patients before the BMT were susceptible to CTL lysis and
more susceptible to NK lysis than K562 cells. In three of
five patients tested, there was a reduction in the ability of
CTL to lyse the post-transplant target, with conserved susceptibility to NK lysis. One patient at relapse showed no
change in target susceptibility and one showed an increased
susceptibility to CTL lysis, but a reduced sensitivity to NK
lysis with cells that were more resistant than K562 cells.
Since there were no important differences in the composition of the stimulator/targets before and after relapse, the
observed differences in immunogenicity between pre-BMT
and relapsed leukemias could not be attributed to the
stimulator/target cell composition (Table 3).
In order further to define the changes in susceptibility
to immune attack, we quantitated major surface molecules
involved in the immune response on pre- and post-BMT
leukemia samples. CD33+ cell fractions from CML and
AML and MDS pre-transplant, showed some variation in
the percentage of marked cells and fluorescence intensity.
Mostly the expression of MHC molecules, costimulatory
molecules and Fas, were in an intermediate range between
the highly expressing EBV LCL controls and the low or
absent expression found in K562 cells. Leukemias relapsing
after transplantation showed two broad patterns of alteration in surface molecule expression: (1) downregulation
of MHC class I and II with upregulation of ICAM-1 and
upregulation or high expression of B7.1; and (2) increase
or no change in MHC expression but decrease or no change
in ICAM-1 and B7.1. Expression of Fas antigen was variably altered at relapse (in four it was unchanged or fell but
remained highly expressed; in two it was increased). Both
loss of MHC expression or loss of costimulation from
ICAM-1 or B7.1 could explain the decreased ability of
relapsed leukemia to stimulate T cell responses. These
findings did not however readily explain the reduced susceptibility of the leukemias to CTL or NK mediated lysis.
The decreased ability of the relapsed leukemia to stimulate an immune response and the increased ability to avoid
cytotoxic damage from CTL or NK cells were consistent
with the hypothesis that immune escape had occurred by
clonal evolution leading to relapse. In support of clonal
evolution in the leukemia, was the observation of new
chromosomal abnormalities in P2, P3 and P4 and a change
in morphology in P1 and P2. The alteration in the leukemia
cell type could be the result of selection of pre-existing
leukemia sub-clones remaining after the preparative regimen or to mutations occurring after BMT. It is of course
997
Mechanisms of escape from GVL by relapsed leukemia cells
S Dermime et al
998
possible that many other factors involved in target susceptibility to lysis could also have altered to allow the leukemia
to relapse, or that relapse occurred for non-immunological reasons.
The existing data are insufficient to link the in vitro
changes that we found with responsiveness of the leukemia
to treatment with donor lymphocytes, withdrawal of
immunosuppressive therapy or use of cytokines to alter the
immune response. It is of interest however that none of the
three patients who showed downregulation of MHC molecules and upregulation of ICAM-1 and B7.1 responded to
donor lymphocyte infusions with or without the addition of
IFN-a, while two of the patients who showed MHC upregulation and downregulation of ICAM-1 and B7.1 achieved
a further remission (one following cessation of CsA and
one following DLT and IFN-a).
Finally we investigated the potential of immune modulation to repair defects of stimulatory capacity or target susceptibility. Since IFN-g upregulates MHC molecules30 we
tested its effect on upregulating surface molecule
expression of relapse leukemia cells and altering immunostimulation and susceptibility to lysis. IFN-g upregulated
MHC class I and II expression in some relapsed leukemias
tested and variably upregulated ICAM-1 and B7.1
expression. One patient P2 showed a significant response
with a marked increase in stimulatory ability of the relapsed
leukemia and return of the target susceptibility to pre-transplant levels. These changes were accompanied by small
increases in ICAM-1 and MHC expression.
In conclusion these studies suggest that escape from
immune regulation may be a common accompaniment of
clonal evolution leading to relapse after allogeneic BMT.
This escape from immune surveillance can involve both the
stimulatory and the effector arm of the immune response
and alter susceptibility to T cells and NK cells. Understanding the mechanisms of immune escape may lead to
improved approaches to immune modulation of the alloresponse and better treatments for leukemias that relapse
after BMT.
Acknowledgements
We would like to thank Toni Simonis for HLA serotyping and
DNA typing, and Dr Carlo Gambacorti-Passerini and Dr Neal
Young for helpful discussion.
References
1 Apperley JF, Jones L, Hale G, Goldman GM. Bone marrow
transplantation for chronic myelogenous leukaemia: T-cell
depletion with Campath-1 reduces the incidence of acute graftversus-host disease but may increase the risk of leukaemia
relapse. Bone Marrow Transplant 1986; 1: 53–66.
2 Weiden PL, Horowitz MM. Graft-vs-leukaemia effects clinical bone marrow transplantation. In: Burakoff HJ, Deeg HJ,
Ferrara I, Atkinson K (eds). Graft-vs-Host Disease; Immunology, Pathophysiology, and Treatment. Dekker: New York,
NY, 1990, pp 691–708.
3 Horowitz MM, Gale RP, Sondel PM et al. Graft-versus-leukaemia reactions after bone marrow transplantation. Blood
1990; 75: 555–562.
4 Horowitz MM, Truitt RL. Graft-versus-leukaemia effects of
bone marrow transplantation. In: Atkinson K (ed). Clinical
Bone Marrow Transplantation. Cambridge University Press:
Cambridge, 1994, pp 704–714.
5 Barrett AJ, Malkovska V. Graft-versus-leukaemia: understanding and using the alloimmune response to treat haematological malignancies. Br J Haematol 1996; 93: 754–761.
6 Kolb HJ, Mittermuller J, Clemm C et al. Donor leukocyte
transfusions for treatment of recurrent CML in marrow transplant patients. Blood 1990; 76: 2462–2465.
7 Marmont AM, Horowitz MM, Gale RP et al. T-cell depletion
of HLA-identical transplants in leukaemia. Blood 1991; 78:
2120–2130.
8 Antin JH. Graft-versus-leukaemia: no longer an epiphenomenon. Blood 1993; 82: 2273–2277.
9 Truitt RL, Johnson BD. Principles of graft-vs-leukaemia reactivity. Biol Blood Marrow Transplant 1995; 1: 61–68.
10 Bosslet K, Schirrmacher V. Escape of metastasizing clonal
tumor cell variants from tumor specific cytolytic T lymphocytes. J Exp Med 1981; 154: 557–562.
11 Gronberg A, Kiessling R, Eriksson E, Hansson M. Variants
from a MLV-induced lymphoma selected for decreased sensitivity to NK. J Immunol 1981; 127: 1734–1739.
12 Sondel PM, Hank JA, Molanda J et al. Relapse of host
leukaemic lymphoblasts following engraftment by an HLAmismatched marrow transplant: mechanisms of escape from
the graft-versus-leukaemia effect. Exp Hematol 1985; 13:
782–790.
13 Taylor GM, Fergusson W, Dyer PA et al. An HLA-DR negative acute leukaemia which stimulates MLC and CMC
responses. Cancer Immunol Immunother 1983; 16: 117–122.
14 Doyle A, Martin WJ, Funa K et al. Markedly decreased
expression of class I histocompatibility antigens, protein, and
mRNA in human small-cell lung cancer. J Exp Med 1985;
161: 1135–1151.
15 Lassam N, Jay G. Suppression of MHC class I RNA in highly
oncogenic cells occurs at the level of transcription initiation.
J Immunol 1989; 143: 3792–3797.
16 Ferrone S, Marincola F. Loss of HLA class I antigens by
melanoma cells: molecular mechanisms, functional significance and clinical relevance. Immunol Today 1995; 16:
487–494.
17 Boon T, DePlaen E, Lurquin C et al. Identification of tumor
rejection antigens recognized by T lymphocytes. Cancer Surv
1992; 13: 23–37.
18 Alexander M, Bennicelli J, Guerry D. Defective antigen presentation by human melanoma cell lines cultured from
advanced, but not biologically early disease. J Immunol 1989;
142: 4070–4078.
19 Restifo NP, Esquivel F, Kawakami Y et al. Identification of
human cancers deficient in antigen processing. J Exp Med
1993; 177: 265–272.
20 Trojan J, Johnson TR, Rudin SD et al. Treatment and prevention of rat glioblastoma by immunogenic C6 cells expressing
anti-sense insulin-like growth factor I RNA. Science 1993;
259: 94–97.
21 Chen L, Ashe S, Brady WA et al. Costimulation of antitumor
immunity by the B7 counter receptor for the T lymphocyte
molecules CD28 and CTLA-4. Cell 1992; 71: 1093–1102.
22 Townsend SE, Allison JP. Tumor rejection after direct
costimulation of CD8+ T cells by B7-transfected melanoma
cells. Science 1993; 259: 368–370.
23 Matulonis UA, Dosiou C, Lamont C et al. Role of B7-1 in
mediating an immune response to myeloid leukaemia cells.
Blood 1995; 85: 2507–2515.
24 Jonjic N, Alberti S, Bernasconi S et al. Heterogeneous susceptibility of human melanoma clones to monocyte cytotoxicity:
Mechanisms of escape from GVL by relapsed leukemia cells
S Dermime et al
role of ICAM-1 defined by antibody blocking and gene transfer. Eur J Immunol 1992; 22: 2255–2260.
25 Anichini A, Mortarini R, Alberti S et al. T cell-receptor
engagement and tumor ICAM-1 up-regulation are required to
by-pass low susceptibility of melanoma cells to autologous
CTL-mediated lysis. Int J Cancer 1993; 53: 994–1001.
26 Panayiotidis P, Ganeshaguru L, Foroni L, Hoffbrand AV.
Expression and function of the Fas antigen in B chronic lymphocytic leukaemia and hairy cell leukaemia. Leukemia 1995;
9: 1227–1232.
27 Munker R, Lubbert M, Yonehara S et al. Expression of the
Fas antigen on primary human leukaemia cells. Ann Hematol
1995; 70: 15–17.
28 Mavroudis DA, Jiang YZ, Hensel N et al. Specific depletion
of alloreactivity against haplotype mismatched related individuals by a recombinant immunotoxin: a new approach to graftversus-host disease prophylaxis in haploidentical bone marrow
transplantation. Bone Marrow Transplant 1996; 17: 793–799.
29 Jiang YZ, Mavroudis DA, Dermime S et al. Alloreactive
CD4+ T lymphocytes can exert cytotoxicity to chronic myeloid
leukaemia cells processing and presenting exogenous antigen.
Br J Haematol 1996; 93: 606–612.
30 Hayashi H, Tanaka K, Jay F et al. Modulation of the tumorigenicity of human adenovirus-12-transformed cells by interferon. Cell 1985; 43: 263–267.
999