Elevated serum HLA class I levels coincide with acute and chronic
graft-versus-host disease
LM Liem
1, CA Koelman
1, IIN Doxiadis
1, JC van Houwelingen
2, E Goulmy
1and FHJ Claas
1 1Department of Immunohematology and Blood Bank and2Department of Medical Statistics, Leiden University Hospital, Leiden, The NetherlandsSummary: More than two decades ago, the presence of naturally
occurring soluble HLA class I molecules (sHLA-I) in the
The ability to predict the likely occurrence of graft- sera of healthy individuals was reported by van Rood et al2
versus-host-disease (GVHD) after BMT would be and Charlton and Zmijewski.3 Both groups defined these
extremely valuable. We performed a retrospective study sHLA-I molecules by inhibition of HLA-specific
alloanti-on the correlatialloanti-on between soluble HLA class I (sHLA- sera in a complement-dependent cytotoxicity assay.
sHLA-I) levels and GVHD in the sera of 34 patients receiving I can be detected in virtually all body fluids.4Serum HLA
an allogeneic BMT and in the sera of 12 patients receiv- levels are fairly constant in normal individuals and range
ing an autologous BMT. sHLA-I levels measured pre- between 0.5 and 25mg/ml.5–9
and at different times post-BMT were correlated with Different lines of evidence suggest that sHLA-I is
the occurrence of post-BMT complications, ie acute increased during immunologically active periods, eg during graft-versus-host disease (aGVHD), chronic graft-ver- the course of viral infections,9–11 after vaccination12 and
sus-host disease (cGVHD), infections and relapse. No during rejection episodes.13–18 The liver is a major source
changes in sHLA-I levels (DsHLA-I) occurred in auto- of sHLA-I as has been shown in liver transplant recipients, logous and allogeneic BMT patients without GVHD. In where up to 50% of the total sHLA-I found in the recipient contrast, sHLA-I reached high levels in patients suffer- after transplantation is of donor-type.17 Analysis of the
ing from GVHD. Increased sHLA-I levels correlated post-transplantation period revealed a correlation between strongly with episodes of both acute and chronic GVHD
increased levels of both donor-type serum sHLA-I and total
(P=0.004 and P=0.005, respectively). Also during
serum sHLA-I with rejection episodes.13,14In kidney and
relapse increased sHLA-I levels were found (P=0.032).
heart transplant recipients increased sHLA-I levels are also
During infections sHLA-I levels increased, although not
correlated with rejection episodes.15–18In kidney transplant
significantly. Kinetic studies gave no evidence that the
recipients, however, this correlation is still controversial.19
increase in sHLA-I levels preceded the clinical
occur-sHLA-I levels have also been analyzed in recipients of
rence of aGVHD or of cGVHD. A slight, but significant
BMT. Tsuji et al20 and Westhoff et al21 were the first to
correlation was found between total blood bilirubin
lev-note an increase of sHLA-I levels coinciding with the
clini-els and sHLA-I levclini-els in patients suffering from GVHD
cal signs of GVHD in a small series of patients. The effect
(P=0.037), indicating the contribution of the liver as a
of infections on sHLA-I levels, however, were not
investi-source of sHLA-I. We conclude that measurements of
gated. Recently, Puppo et al22 reported their results on
sHLA-I levels do not function as a predictive parameter
sHLA-I levels after allogeneic BMT in a prospective study.
for GVHD, but can be valuable for the monitoring of
sHLA-I levels remained relatively stable in patients with
GVHD after BMT.
acute GVHD grade I without infection. In patients suffering
Keywords: sHLA class I; bone marrow transplantation;
from acute GVHD grades II–IV, however, sHLA-I levels graft-versus-host disease
started to rise a week before the onset of acute GVHD grades II–IV, reached their peak level at the onset and stayed elevated during the following 2 weeks. In contrast, Bone marrow transplantation (BMT) is an established sHLA-I levels increased significantly 2 weeks after the therapy for hematological diseases. Development of graft- onset of infections, although to a lower extent than versus-host-disease (GVHD) is still a severe complication during GVHD.
occurring in approximately 35% of recipients of HLA- Here we report our results of a retrospective analysis of genotypically identical marrow.1 A parameter for early
sHLA-I levels measured pre- and at different times post-diagnosis of GVHD would therefore be of great value. BMT in sera of 34 recipients of allogeneic BMT. In addition, 12 recipients of autologous BMT were analyzed to study a possible transplant-related effect. The putative
Correspondence: Dr E Goulmy, Leiden University Hospital, Dept of correlation of sHLA-I levels with acute GVHD, chronic Immunohematology and Blood Bank, PO Box 9600, 2300 RC Leiden,
GVHD, relapse and infections on the shLA-I levels
post-The Netherlands
The first two authors have contributed equally to this paper. BMT were statistically assayed.
228
Materials and methods Complications after BMT
GVHD was diagnosed according to established criteria.23
Patients Relapse of the original disease was confirmed histologically
by bone marrow aspirate and, if possible, genetically by Forty-six adult patients who underwent BMT between 1978
the presence of Philadelphia chromosome-positive cells. and 1990 in the Leiden University Hospital were included
Infections comprised both viral (CMV, VZV, HSV), bac-in this study. Thirty-three patients received bone marrow
terial (sepsis, pneumonitis) and generalized candida infec-from an HLA-identical sibling, one patient (UPN 30)
tions. received bone marrow from her HLA-identical father and
12 patients received autologous bone marrow. Of the
Sera
patients receiving allogeneic bone marrow (21 females, 13
males) the mean age was 27 years (range 16–42). Table 1 Serum samples were collected before BMT and after BMT. Post-BMT serum samples were collected systematically for summarizes relevant clinical information on the allogeneic
BMT patients.The mean age of the patients receiving auto- the first 3 months post-BMT, and thereafter incidentally up to 3 years post-BMT. Serum samples were also collected logous bone marrow (five females and seven males) was
37 (range 20–58). Table 2 summarizes the data on the auto- from BMT donors and healthy blood donors. All sera were stored at −30°C until further use. For this study, we logous BMT patients. Normal levels of sHLA-I were
determined in bone marrow donors and healthy blood attempted to include at least one sample for every 10-day period post-BMT until day 100 and during complications. donors (n=36).
Table 1 Clinical data of the allogeneic BMT patients included in this study
UPNa Diagnosisb BMT Conditioningc Prophylaxisd aGVHD cGVHD Relapse Survivalf HLA class I typing (onset)e
1 AML-M4 08.04.82 Cy/TBI MTX 0 0 no alive A2,3 B7
2 AML-M1/2 09.12.82 Cy/TBI MTX 0 0 no alive A1,36 B57,7 Cw6
3 AML-M2 11.02.82 Cy/TBI MTX 0 0 no alive A29,30 B8,44 Cw7
4 SAA II 14.10.82 Cy/TLI MTX 0 0 no +28 A1,2 B37 Cw6
5 AML-M3 06.05.82 Cy/TBI MTX 0 0 no alive A2,31 B58,40 Cw3,w7
6 AML-M2 21.11.79 Cy/TBI MTX 0 0 no alive A2,31 B44,56 Cw5
7 AML-M2 15.08.85 Cy/TBI CsA/MTX 0 0 yes +513 A2, B44,62 Cw3
8 ALL 16.06.83 Cy/TBI MTX 0 0 yes +538 A2,28 B44,4 Cw5
9 AML-M4 25.05.84 Cy/YBI CsA 0 0 yes +223 A1,30 B8,13 Cw6,w7
10 AML-M1 26.11.81 Cy/TBI MTX 0 0 yes +609 A11,28 B8,35 Cw4
11 AML-M2 27.02.80 Cy/TBI MTX I(+8) 0 no alive A3,24 B7,35 Cw4
12 AML-M2 31.01.85 Cy/TBI CsA I(+11) 0 no alive A2,3 B7,37 Cw6,w7
13 ALL 19.07.84 Cy/TBI CsA I(+14) 0 yes +191 A2, B7 Cw7
14 B-ALL 0.7.07.83 Cy/TBI MTX II 0 no alive A1,3 B8,39 Cw6,w7
15 AML-M4 22.11.84 Cy/TBI CsA II(+17) 0 no +175 A2,3 B7,18 Cw5,w7
16 AML-M2 17.09.81 Cy/TBI MTX II(+23) 0 no alive A1,11 B8,60 Cw3,w7
17 SAA III 27.05.82 Cy/TLI MTX II(+31) 0 no +143 A1,3 B7,13 Cw6
18 AML-M2 12.01.84 Cy/TBI MTX III(+54) 0 no alive A2 B8,15 Cw3,w7
19 SAA III 25.06.81 Cy/TLI MTX 0 severe no +406 A2,24 B7,55 Cw3
20 AML-M4 28.04.83 Cy/TBI MTX I(+33) mild no alive A3 B35 Cw4
21 CML (Ph+) 16.02.84 Cy/TBI CsA II(+11) mild no alive A2,3 B7,44 Cw5,w7
22 CML (Ph+) 10.02.83 Cy/TBI MTX II(+16) mild no +420 A2,3 B18,44 Cw7
23 AML-M4 24.01.85 Cy/TBI CsA II(+17) severe no alive A2,31 B7,41 Cw7
24 ALL 27.05.82 Cy/TBI MTX II(+19) mild no alive A24,28 B7,35 Cw4
25 AML-M1 05.01.84 Cy/TBI MTX II(+28) severe no +497 A3,19 B5,7 Cw7
26 AML-M1 12.04.84 Cy/TBI CsA III(+11) severe no +139 A3,32 B7,44 Cw7
27 SAA III 18.11.82 Cy/TLI MTX III(+12) severe no +191 A2,24 B27,62 Cw3
28 AML-M2 21.02.85 Cy/TBI CsA III(+12) severe no +269 A1,3 B7,8 Cw7
29 AML-M2 15.01.81 Cy/TBI MTX III(+24) severe no alive A2 B27,62 Cw1,w3
30 AML-M2 05.08.82 Cy/TBI MTX III(+24) severe no +155 A2,3 B27,47 Cw3,w3
31 AML-M4 03.12.81 Cy/TBI MTX III(+25) severe no +104 A2,29 B44,49
32 SAA 18.01.79 Cy/TLI MTX III(+35) severe no alive A1,2 B7,8 Cw7
33 SAA III 29.09.78 Cy/TLI MTX IV(+15) severe no +268 A2 B8,40 Cw3
34 AML-M1 18.08.83 Cy/TBI MTX II(+35) severe yes alive A2,3 B35,44 Cw4,w5
aUPN=unique patient number.
bDiagnosis of the patients: AML=acute myeloid leukemia with FAB classification; ALL=acute lymphoblastic leukemia; SAA=severe aplastic anemia; CML=chronic myeloid leukemia.
cConditioning regimen: Cy/TBI=cyclophosphamide and total body irradiation; Cy/TLI=cyclophosphamide and total lymph node irradiation. dGVHD prophylaxis: CsA=cyclosporin A; MTX=methotrexate.
eDay of onset of aGVHD.
229 Table 2 Clinical data of the autologous BMT patients included in this study
UPNa Diagnosisb BMT Conditioningc Relapse Survivald HLA class I typing
43 AML-3 7.6.89 Cy/TBI no alive A3,28 B7,60 Cw3,w7
44 M. Hodgkin 17.11.88 Hovon 8 no alive A1,2 B4,37 Cw6
45 AML-M3 8.6.88 Cy/TBI no alive A24,11 B8,50 Cw6,w7
46 AML-M1/2 29.6.88 Cy/TBI no alive A2 B39,47 Cw6
47 AML (undiff.) 15.11.89 Cy/TBI yes +1514 A2,32 B51,60 Cw2,w3
48 NHL 18.10.89 BEAC yes +1121 A2,3 B7,62 Cw4,w7
49 NHL 29.3.90 BEAC no alive A2,32 B51,14 Cw1
50 NHL 9.8.89 Cy/TBI no alive A11,28 B55,35 Cw9,4
51 AML-M5B 7.2.90 Cy/TBI no alive A1,2 B62,15 Cw9
52 M. Hodgkin 20.7.88 Hovon 8 no alive A2,24 B18,60 Cw10,w7
53 M. Hodgkin 2.11.89 Hovon 8 yes +677 A2 B7,56 Cw1,w7
54 NHL 3.5.89 CY/TBI no alive A24,66 B35,41
aUPN=Unique patient number.
bDiagnosis of the patients: AML=acute myeloid leukemia with FAB classification; M. Hodgkin=morbus Hodgkin; NHL=non-Hodgkin lymphoma. cConditioning regimen: Cy/TBI=cyclophosphamide and total body irradiation; Hovon 8=BCNU, etoposide and cyclophosphamide; BEAC=BCNU, etoposide, cyclophosphamide and Ara-C.
dDay of death after BMT. Survival=alive means: follow-up is>5 years post-BMT.
sHLA-I ELISA Statistical analysis
Serum levels of sHLA-I were assessed using the following
Since serum sHLA-I levels vary widely between individ-method. Microtiter plates (Greiner, Rijn, The Netherlands)
uals, the values are reported in figures as changes in sHLA-were coated with goat-anti-mouse IgG (Jackson, West
I level using the formula: DsHLA-I=[post-BMT Grove, PA, USA), diluted 1:500 for 1 h at room
tempera-concentration]−[pre-BMT concentration].24To test for
dif-ture. Plates were then incubated with the monoclonal
anti-ferences in sHLA-I levels of healthy controls and pre-BMT body TP25.99, directed against the a3 domain of HLA
sHLA-I levels of BMT patients, t-tests were used. Differ-class I (a kind gift of Prof Ferrone, Valhalla, New York,
ences between autologous BMT patients and allogeneic USA) (5mg/ml, 60 min at 37°C), washed and blocked with
BMT patients in changes in sHLA-I levels from pre- to phosphate-buffered saline (PBS)/0.05% Tween/1% bovine
post-BMT were determined with the repeated measure-serum albumin (BSA) for 30 min at 37°C). Twomg EDTA
ments multivariate analysis of variance (MANOVA). For were added to 100ml serum and briefly centrifuged. Plates
this analysis, data were only used if no aGVHD or cGVHD were washed, and sera, diluted 1:80 or 1:200 in PBS/0.05%
was present. Technically speaking, in this MANOVA the Tween/1% BSA, were incubated for 16–18 h (100ml/well,
interaction between the BMT group and pre-post treatment 4°C). After washing the plates with PBS/0.05% Tween,
is tested after correction for the factor ‘patients’. By taking 100ml rabbit-anti-human b2m (ACSC) were added to the
‘patients’ as a factor in the MANOVA, a correction for wells (1:10 000, 60 min, 37°C). Finally, plates were washed
general patient levels is carried out. To determine the corre-and incubated with a goat-anti-rabbit Ig conjugated with
lation of different clinical complications with changes in horse radish peroxidase (Sigma, Bornem, Belgium)
(1:1000, 60 min, 37°C). Bound antibody was detected by sHLA-I levels within the allogeneic BMT group, compli-using 1,2-phenylenediamine (OPD; DAKO, Glostrup, cations were coded into the following dichotomous vari-Denmark) as a substrate. The reaction was stopped with ables: ‘relapse’=absence (0) vs presence of relapse (1), 2.5 n H2SO4and absorbance was read at 492 nm. Different ‘aGVHD’=grades 0–1 (0) vs grades II–IV (1),
dilutions of a control serum with a known concentration of ‘cGVHD’=absence (0) vs mild or severe cGVHD (1) and sHLA-I were used as standard. All samples from an indi- ‘infection’=absence (0) vs bacterial, viral or fungal infec-vidual were tested in the same experiment to minimize tion (1). The statement used in the MANOVA is the follow-inter-assay variation. ing: DsHLA-I BY patients (1 34) WITH relapse, aGVHD, cGVHD, infection. Thus only the changes within patients are used and not the differences between patient groups.
sHLA-I levels and liver GVHD
This analysis can be seen as an extension of the well-known To evaluate the contribution of liver GVHD to increased paired t-test. The correlation between changes in sHLA-I sHLA-I levels, total blood bilirubin levels were used as a levels and changes in bilirubin levels within selected marker for the occurrence of liver GVHD.23For this
analy-patients was obtained in a truly multivariate MANOVA sis patients were selected who suffered from GVHD grades with ‘patients’ as factor and 10log(bili) and DsHLA-I as
II–IV and whose bilirubin levels were known (n=13). Bili- outcome variables. P values ,0.05 were accepted as rubin data were converted to their10log-value (10log(bili)).
significant. All data on sHLA-I and bilirubin levels from these patients
230 20 15 10 5 0 -5 -10 0 100 200 300 400 UPN 43 UPN 44 UPN 45 UPN 46 UPN 47 UPN 48 UPN 49 UPN 50 UPN 51 UPN 52 UPN 53 UPN 54 a Days post-BMT ∆ sHLA-I ( g/ml) µ µ 20 15 10 5 0 -5 -10 0 100 100 300 400 UPN 1 UPN 2 UPN 3 UPN 4 UPN 5 UPN 6 UPN 11 UPN 12 b ∆ sHLA-I ( g/ml) Days post-BMT
Figure 1 Kinetics ofDsHLA-I in autologous BMT patients (a), allogeneic BMT patients with aGVHD grades 0–1 (b), relapsed allogeneic BMT patients (c) and allogeneic BMT patients with aGVHD grades II–IV or cGVHD (d). Each line represents one patient. The dashed line isDsHLA=0. Day 0 is day of BMT. The grey area is the area of normal fluctuation, as is assessed in the autologous BMT group.
Results 0.94mg/ml and 18.73mg/ml (mean 4.81±4.30mg/ml).
The pre-BMT sHLA-I levels of 12 patients receiving
auto-sHLA-I class I levels in healthy individuals and in BMT
logous BMT ranged between 1.64mg/ml and 17.33mg/ml
recipients before BMT
231 20 15 10 5 0 -5 -10 0 100 200 300 400 ∆ sHLA-I ( g/ml) UPN 7 UPN 8 UPN 9 UPN 10 UPN 13 UPN 34 c Days post-BMT µ * * * * 20 15 10 5 0 -5 -10 0 100 200 300 400 UPN 14 UPN 15 UPN 16 UPN 17 UPN 18 UPN 19 UPN 20 UPN 21 UPN 22 UPN 23 UPN 24 UPN 25 UPN 26 UPN 27 UPN 28 UPN 29 UPN 30 UPN 31 UPN 32 UPN 33 UPN 34 d ∆ sHLA-I ( g/ml) Days post-BMT µ Figure 1 (Continued).
sHLA-I levels of healthy blood and bone marrow donors 4.5mg/ml of the pre-BMT level with the exception of
DI of two patients (UPNs 45 and 54) whose
sHLA-and autologous or allogeneic BMT patients pre-BMT were
not significant (data not shown). I decreased post-BMT. sHLA-I levels of both patients returned to pre-BMT value after approximately 60 days post-BMT (Figure 1a). No significant difference in
uncom-Post-BMT sHLA-I levels in autologous and allogeneic
plicated post-BMT DsHLA-I levels was found between
BMT patients
autologous and allogeneic BMT recipients.
232
Table 3 Mean and s.d. values in the absence and presence of differ- Table 4 Results of the repeated measurements MANOVA analysis ent complications
Complications B s.e. P value
No. No. Mean s.d.
patients samples DsHLA-I Relapse 2.24 1.04 0.032
aGVHDs II–IV 1.74 0.59 0.004
cGVHD 1.96 0.69 0.005
No relapse 29 175 1.27 3.01
Relapse absence 5 19 −0.30 2.26 Infection −0.07 0.45 0.872
presence 12 1.27 2.76
All complications (relapse, aGVHD II–IV, cGVHD and infection) were
No aGVHD 17 90 0.84 2.84
coded as dichotomous variables as described in the Material and methods
aGVHD grades 0–1 17 75 0.82 2.50
section and were entered as covariate in the analysis. For each
compli-grades II–IV 41 2.28 3.69
cation, regression coefficient (B), standard error (s.e.) and P value are No cGVHD 25 128 0.75 3.27 given. Since the variables are dichotomous, B represents the effect of the cGVHD absence 9 48 1.27 2.33 presence vs the absence of a complication onDsHLA-I levels. These multi-presence 30 2.49 1.98 variately corrected effects are very similar to the differences between the mean sHLA-I levels in the absence or presence of a complication in the
No infections 8 25 1.27 3.06
Infections absence 26 121 0.98 2.76 group of patients who suffered from that complication.
presence 60 1.35 3.34
sHLA-I levels increase more than during bacterial or fungal
For each complication the total allogeneic BMT group is divided into
infections, although none of these increases was
signifi-patients who did not suffer at all from that complication during any of
the time-points measured and patients who suffered from that complication cant (Figure 2).
during at least one post-BMT time-point measured. For each category the Increases in sHLA-I corresponded to periods of active number of patients, the number of samples and the mean and s.d. are
GVHD. However, no evidence was found in our patient
given. In the patients who suffered from a complication during one or
group that sHLA-I were raised before the clinical signs of
more post-BMT time-points measured, the number of samples, means and
s.d. are given for data obtained in the absence or presence of that compli- GVHD. No correlation could be found between the severity
cation. of aGVHD and theDsHLA-I (data not shown).
The effect of liver GVHD on sHLA-I levels
DsHLA-I levels also remained within 4.5mg/ml of the
pre-BMT level (Figure 1b). In two patients with aGVHD grade We investigated whether GVHD of the liver has a signifi-I (UPNs 11 and 12) the sHLA-signifi-I levels increased above this cant influence on the sHLA-I levels. Blood bilirubin levels threshold around day 200. In both cases no sign of GVHD were used as a marker for liver GVHD and correlated with was found, but the increase in sHLA-I level correspon- DHLA-I levels. The results are shown in Figure 3. denced to a viral infection. Recipients of allogeneic marrow MANOVA analysis revealed a slight, but significant corre-who relapsed maintained stable levels (Figure 1c), which lation between bilirubin and DsHLA-I levels (R=0.282, was also observed in recipients of autologous marrow who P=0.037).
relapsed (Figure 1a, UPNs 47, 48 and 53). In most patients suffering from either aGVHD grades II–IV or cGVHD,
sHLA-I levels reached levels above the threshold of Discussion 4.5mg/ml (Figure 1d).
Our results show that the conditioning regimen and BMT itself do not have a significant effect on sHLA-I levels in
sHLA-I levels during post-BMT complications
autologous or allogeneic BMT patients. After BMT the The effects of post-BMT complications were further
ana-lyzed in the allogeneic recipients. This group can be div-ided into patients who never suffered from a given compli-cation and patients who suffered from that complicompli-cation during at least one post-BMT time-point measured. Data from the latter group is obtained in the absence or presence of that complication. Summary statistics of these subgroups are given in Table 3.
Changes in sHLA-I levels during different complications were analyzed in the MANOVA analysis as described in the Material and methods section. The results from the MANOVA analysis are summarized in Table 4. Signifi-cantly, increased sHLA-I levels were found during aGVHD and cGVHD (P=0.004 and P=0.005, respectively) and during relapse (P=0.032). Although sHLA-I levels rise during infection, this increase is not significant. Also, in the autologous BMT patients, infections do not cause a
sig-2.5 2.0 1.5 1.0 0.5 0.0 -0.5
No Bacterial Viral Fungal
P = 0.695 P = 0.487 P = 0.865
∆
sHLA-I ( g/ml)
µ
nificant change in sHLA-I levels. Further analysis of differ- Figure 2 DsHLA-I during different kinds of infections. Viral infections
increase sHLA-I levels the most, although the increase is not significant.
233 In our retrospective study, infections did not have a
sig-nificant influence on sHLA-I levels. Since viral infections upregulate IFNg production, that in turn can increase the release of sHLA-I, and increased sHLA-I levels during viral infections are to be expected. When analyzing the different kinds of infections separately, viral infections increase sHLA-I levels the most, although not significantly. In our study, infections have only been recorded when the patient showed signs of active infection. If sHLA-I levels do not start rising during the active infection, but earlier or later in the follow-up, no correlation is found in our analysis. In a prospectively designed study, Puppo et al22 found an
increase in sHLA-I levels 2 weeks after onset of infection. However, they only considered patients suffering from GVHD and compared sHLA-I levels in patients with and
1000 100 10 1 -10 -5 0 5 10 -15 R = 0.282 P = 0.037 Bilirubin ( mol/l) µ ∆sHLA-I ( µg/ml)
without infection. Therefore, the difference between the lat-Figure 3 Scatterplot ofDsHLA-I (x-axis) and total blood bilirubin levels ter results and ours may well lie in the design of the studies. (y-axis, log-scale). The correlation coefficient and the P value, obtained
Our data show that GVHD and infections differ in their
in the MANOVA as described in the Material and methods section, show
a weak, but statistically significant correlation. ability to increase sHLA-I. While GVHD episodes have a profound effect on sHLA-I levels, the effect of infections is less prominent. Puppo et al22 showed in their group of
sHLA-I levels of autologous and allogeneic BMT patients
with GVHD grades 0–I remain fairly stable. In patients suf- BMT patients that sHLA-I levels are higher during GVHD than during infections. Both GVHD and infections lead to fering from aGVHD grades II–IV or cGVHD, significantly
increased sHLA-I levels were found. sHLA-I levels during an activation of the immune system. During GVHD, how-ever, more tissue damage will occur than during infections, relapse were also significantly increased. This group,
how-ever, consisted of six patients, of whom one relapsed while suggesting that release of sHLA-I molecules by damaged tissue contributes significantly to increased sHLA-I levels suffering from severe cGVHD. Therefore, these results
should be interpreted with care. Although an increment of during GVHD.
The biological function of sHLA-I is still unknown, I level was observed, infections did not affect
sHLA-I levels significantly. though it has been shown that sHLA-I can down-regulate both cellular and humoral responses.24,28,29A recent study
The increase in sHLA-I during GVHD may be explained
by two mechanisms. Firstly, activation of donor bone mar- by Zavazava and Kro¨nke30showed that sHLA acts in vitro
on T cells by promoting apoptosis of activated, Fas-row-derived cells in the course of GVHD may lead to
enhanced release of sHLA-I by these cells. Enhanced pro- expressing T cells through the upregulation of Fas ligand in an allospecific fashion. Although the studies performed duction of endogenous cytokines might underlie the
increased sHLA-I production by activated donor-derived so far in BMT patients show that increased sHLA-I levels correlate well with GVHD activity, its modulating proper-immune cells. Therapeutic use of recombinant
interferon-g(IFNg) has been shown to increase sHLA-I levels25,26and ties in the GVH reactivities remain to be established.
it has been suggested that tumor necrosis factora(TNFa) might have a similar effect.9 Since both IFNg and TNFa
have been implicated in the pathogenesis of GVHD,27the
Acknowledgements
increase in sHLA-I levels might be a reflection of an
increased production of both cytokines. Secondly, sHLA-I We thank Prof Dr R Willemze, Dr A Brand and Mrs L Huige for may be released from patient tissue damaged during the collection of serum samples of hematological patients. We thank Dr M Oudshoorn for critically reviewing this manuscript. GVHD. The correlation between increased donor sHLA-I
This work was supported by the JA Cohen Institute for Radiopa-and rejection episodes after solid organ transplantation
thology and Radiation Protection (IRS), by EC grant BIO2 CT92 strongly suggests an important role for the latter mechanism
0300, by a grant from the Dutch Heart Foundation and by a grant in rejection.13,15–17 In patients suffering from GVHD,
from the MACROPA Foundation. sHLA-I levels correlated significantly with blood bilirubin
levels, implying that GVHD of the liver may contribute to sHLA-I levels. No correlation, however, could be found
References
between sHLA-I levels and the severity of GVHD. Detailed analysis of our retrospective study gave no indication of
1 Beatty PG, Hansen JA, Lonton GM et al. Marrow transplan-an increase of sHLA-I levels before the onset of GVHD.
tation from HLA-matched unrelated donors for treatment of Although it has been suggested that sHLA-I levels start
hematologic malignancies. Transplantation 1991; 51: 443– rising before the onset of aGVHD, the actual evidence
447. remains scarce. Tsuji et al20only noticed a rise in
sHLA-2 Van Rood JJ, van Leeuwen A, van Santen MCT. Anti-HLA-I levels after the onset of GVHD. Westhoff et al21showed
A2 inhibitor in normal serum. Nature 1970; 226: 366–367. that GVHD is preceded by a rise in sHLA-I levels in some 3 Charlton RK, Zmijewski CM. Soluble HLA-A7 antigen: patients and Puppo et al22found significantly increased
lev-Localization in the b-lipoprotein fraction of human serum.
Science 1970; 170: 636–637.
234
4 Singh PB, Brown RE, Roser B. Class I transplantation anti- and kidney allograft recipients. Tissue Antigens 1993; 42:
20–26. gens in solution in body fluids and in the urine. Individuality
17 Claas F, Jankowska-Gan E, DeVito LD et al. Monitoring of signals to the environment. J Exp Med 1988; 168: 195–211.
heart transplant rejection using a donor-specific soluble class 5 Krangel MS. Two forms of HLA class I molecules in human
I ELISA. Hum Immunol 1993; 37: 121. plasma. Hum Immunol 1987; 20: 155–165.
18 Schroeder TJ, Pouletty P, Shimizu RM et al. Soluble class I 6 Kao KJ. Plasma and platelet HLA in normal individuals:
quan-HLA antigens in serum following renal transplantation. Hum titation by competitive enzyme-linked immunoassay. Blood
Immunol 1993; 37: 130. 1987; 70: 282–286.
19 Drouet M, Petit B, Peyronnet P et al. Soluble HLA class I 7 Kao KJ, Scornik JC, Riley WJ, McQueen CF. Association
antigens in kidney allograft recipients. Transplant Proc 1995; between HLA phenotype and HLA concentration in plasma or
27: 1671. platelets. Hum Immunol 1988; 21: 115–124.
20 Tsuji K, Kusama M, Nakatsuji T, Kato S. Biological signifi-8 Doxiadis I, Westhoff U, Grosse-Wilde H. Quantification of
cance of Ss (serum soluble) HLA-class I antigens in bone mar-soluble HLA class I gene products by an enzyme linked
row transplantation. Tokai J Exp Clin Med 1985; 10: 169–174. immunosorbent assay. Blut 1989; 59: 449–454.
21 Westhoff U, Doxiadis I, Beelen DW et al. Soluble HLA class 9 Ferrone S, Yamamura M, Grosse-Wilde H, Pouletty P.
Sum-I concentrations and GVHD after allogeneic marrow trans-mary of the serum soluble HLA class I antigen component.
plantation. Transplantation 1989; 48: 891–893. In: Tsuji K, Aizawa M, Sasazuki T (eds). Proceedings 11th
22 Puppo F, Brenci S, Ghio M et al. Serum HLA class I levels International Histocompatibility Workshop. Oxford Science
in allogeneic bone marrow transplantation: a possible marker Publications: New York; 1992; pp 1057–1061.
of acute GVHD. Bone Marrow Transplant 1996; 17: 735–738. 10 Alvarez-Cemen˜o JC, Echevarra JM, Villar LM et al. Soluble
23 Thomas ED, Storb R, Clift RA et al. Bone marrow transplan-class I antigens in serum and CSF of patients with
Varicella-tation. New Engl J Med 1975; 292: 832–843. Zoster virus meningitis. J Neurol Neurosurg Psychiatry 1989;
24 Auzlitzky WE, Grosse-Wilde H, Westhoff U et al. Enhanced 52: 1194–1196.
serum levels of soluble class I molecules are induced by treat-11 Puppo F, Brenci S, Lanza L et al. Increased level of serum
ment with recombinant interferon-gamma. Clin Exp Immunol class I antigens in HIV infection. Correlation with disease
pro-1991; 86: 236–239. gression. Hum Immunol 1994; 40: 259–266.
25 Hillebrand K, Moritz T, Westhoff U et al. HLA class I and 12 Saririan K, Attiyeh E, Contini P, Russo C. sHLA-I class I
beta-2-microglobulin plasma concentrations during interferon levels in elderly human following influenza vaccination. Hum
treatment of chronic myelogenous leukemia. Vox Sang 1994;
Immunol 1993; 37: 129. 67: 310–314.
13 Davies HffS, Pollard SG, Caine RY. Soluble HLA antigens 26 Ferrara JLM. Cytokine dysregulation as a mechanism of
graft-in the circulation of liver graft recipients. Transplantation versus-host-disease. Curr Opin Immunol 1993; 5: 794–799.
1989; 47: 524–527. 27 Hausmann R, Zavazava N, Mu¨ller-Rochholtz W. Interaction
14 Puppo F, Pellicci R, Brenci S et al. HLA class-I soluble anti- of purified HLA class I molecules with alloreactive CTL.
gen serum levels in liver transplantation. A predictor marker Transplant Proc 1991; 23: 2255–2257.
of acute rejection. Hum Immunol 1994; 40: 166–170. 28 Buelow R, Burlingham WJ, Clayberger C. Immunomodulation
15 DeVito-Haynes LD, Jankowska-Gan E, Sollinger HW, by soluble HLA class I. Transplantation 1995; 59: 649–654.
Knechtle SJ, Burlingham WJ. Monitoring of kidney and sim- 29 Carbone E, Terrazzano G, Colonna M et al. Natural killer
ultaneous pancreas–kidney transplantation rejection by release clones recognize specific soluble HLA class I molecules. Eur
of donor specific, soluble HLA class I. Hum Immunol 1994; J Immunol 1996; 26: 683–689.
40: 191–201. 30 Zavazava N, Kro¨nke M. Soluble HLA class I molecules
16 Zavazava N, Bo¨ttcher H, Mu¨ller Ruchholtz W. Soluble MHC induce apoptosis in alloreactive cytotoxic T lymphocytes. Nat
