12th Congress of the European Hematology ... - Haematologica
12th Congress of the European Hematology ... - Haematologica
12th Congress of the European Hematology ... - Haematologica
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
Genomics and proteomics<br />
0158<br />
PLASMA PROTEOMIC PROFILES MAY PREDICT EARLY ACUTE GRAFT-VS.-HOST DISEASE<br />
FOLLOWING REDUCED INTENSITY CONDITIONING ALLOGENEIC HLA-IDENTICAL SIBLING<br />
TRANSPLANTATION<br />
M. Mohty, 1 M. Mohty, 2 A. Goncalves, 2 E. Esterni, 2 Y. Toiron, 2<br />
B. Gaugler, 2 C. Faucher, 2 J. El-Cheikh, 2 S. Furst, 2 J.A. Gastaut, 2 P. Viens, 2<br />
D. Olive, 2 C. Chabannon, 2 J.P. Borg, 2 D. Blaise2 1 2 Imperial College, LONDON, United Kingdom, Institut Paoli-Calmettes,<br />
MARSEILLES, France<br />
Background. Modern approaches to predict <strong>the</strong> occurrence/severity <strong>of</strong><br />
aGVHD are needed. Aims. This study aimed to identify a plasma protein<br />
signature correlating with occurrence <strong>of</strong> early aGVHD. Methods. We performed<br />
Surface-Enhanced Laser Desorption/Ionization time (SELDI) <strong>of</strong><br />
flight mass spectrometry pr<strong>of</strong>iling <strong>of</strong> plasma from 88 pts who received<br />
RIC allo-SCT from HLA-identical siblings. Results. Median age <strong>of</strong> pts<br />
was 51 (range, 18-70) y. 41 pts (47%) had a myeloid malignancy, whereas<br />
30 (34%) had a lymphoid malignancy. 17 pts (19%) were treated for<br />
metastatic non-hematological malignancies. The RIC regimen included<br />
fludarabine, busulfan and ATG in 53 pts (60%) and low dose irradiation<br />
in 35 pts (40%). With a median FU <strong>of</strong> 400 (range, 127-829) d, 20 pts<br />
(23%) had early (prior to day 35 after allo-SCT) grade 2-4 acute GVHD<br />
(12 grade 2 and 8 grade 3-4). Denatured plasma samples (collected at a<br />
median <strong>of</strong> 28 d.) were incubated with H50 and CM10 ProteinChip arrays<br />
and subjected to SELDI analysis. Pts population was divided into a training<br />
(n=59) and a validation set (n=29). In <strong>the</strong> training set, 36 protein<br />
peaks were differentially expressed according to early aGVHD occurrence.<br />
By combining partial least squares and logistic regression methods,<br />
we built a multiprotein model that correctly predicted outcome in<br />
96% <strong>of</strong> pts (14/14 patients with early aGVHD; specificity, 96%). The<br />
observed correct prediction rate in <strong>the</strong> validation set was 69% with a<br />
sensitivity <strong>of</strong> 67%, and a specificity <strong>of</strong> 70%. While negative predictive<br />
value <strong>of</strong> <strong>the</strong> model was only 36%, predictive positive value was estimated<br />
to 89% in <strong>the</strong> validation set. The performances <strong>of</strong> <strong>the</strong> model remained<br />
very similar after iterative (500 times) random resampling (correct prediction<br />
rate: 74%, median sensitivity: 48%, median specificity: validation<br />
set: 83%). Univariate and multivariate analyses <strong>of</strong> known risk factors<br />
(demographic features, diagnoses and transplant procedures) for<br />
early grade 2-4 aGVHD did not show any statistically significant difference<br />
between <strong>the</strong> group <strong>of</strong> 20 patients who had early grade 2-4 aGVHD<br />
as compared to <strong>the</strong> remaining patients, and suggested that <strong>the</strong> multiprotein<br />
index is likely to be <strong>the</strong> only independent prognostic parameter.<br />
Major components <strong>of</strong> this multiprotein index are currently being characterized<br />
and will be presented. Conclusions. Obviously, larger prospective<br />
studies are still needed, but our results already suggest that proteomic<br />
analysis <strong>of</strong> plasma will prove increasingly important in <strong>the</strong> early<br />
and clinical diagnosis <strong>of</strong> aGVHD.<br />
0159<br />
PROTEOME PROFILING IDENTIFIES APOLIPOPROTEIN A1 AS A SERUM MARKER<br />
CORRELATED TO JAK2 V617F BURDEN IN POLYCYTHEMIA VERA AT DIAGNOSIS<br />
P. Mossuz, 1 A. Bouamrani, 2 S. Brugière, 3 M. Arlotto, 2 S. Hermouet, 4<br />
E. Lippert, 5 F. Laporte, 1 F. Girodon, 6 I. Dobo, 7 V. Praloran,5 J. Garin, 3<br />
J.Y. Cahn, 1 F. Berger1 1 CHU Grenoble, GRENOBLE; 2 INSERM U318, GRENOBLE; 3 CEA, UNIT-<br />
M 201, GRENOBLE; 4 CHU Nantes, NANTES; 5 CHU Bordeaux, BOR-<br />
DEAUX; 6 CHU Dijon, DIJON; 7 CHU Angers, ANGERS, France<br />
Background. Polycy<strong>the</strong>mia Vera (PV) is a myeloproliferative disorder<br />
(MPD) originating from a multipotent haematopoietic progenitor cell.<br />
The great majority <strong>of</strong> PV are characterized by a recurrent acquired gain<strong>of</strong>-function<br />
mutation <strong>of</strong> <strong>the</strong> JAK2 protein (JAK2-V617F). Modifications <strong>of</strong><br />
JAK2-V617F burden are associated with changes <strong>of</strong> PV phenotype and<br />
probably impact on <strong>the</strong> risk <strong>of</strong> complications. However, we don’t know<br />
whe<strong>the</strong>r if <strong>the</strong> proportion <strong>of</strong> JAK2-V617F allele modifies certain serum<br />
proteins. Aims. The purpose <strong>of</strong> our study was to generate serum proteome<br />
pr<strong>of</strong>iles <strong>of</strong> PV patients to discover and identify novel serum protein<br />
correlated with <strong>the</strong> levels <strong>of</strong> granulocyte JAK2-V617F. Methods. PV<br />
serums were collected in a multicenter study and were randomly affected<br />
in two independent learning and validation sets. Proteome pr<strong>of</strong>iles<br />
were generated by SELDI-TOF mass spectrometry. JAK2 V617F status<br />
was determined by quantitative-PCR analysis <strong>of</strong> purified granulocytes.<br />
12 th <strong>Congress</strong> <strong>of</strong> <strong>the</strong> <strong>European</strong> <strong>Hematology</strong> Association<br />
Results. Unsupervised clustering analysis <strong>of</strong> <strong>the</strong> learning set showed that<br />
PV patients could be separated in two major subgroups tending to be different<br />
with respect to <strong>the</strong> mean percentage <strong>of</strong> mutated JAK2 (p=0.09), <strong>the</strong><br />
number <strong>of</strong> PRV-1 transcripts (p=0.08) and Ht (p=0.09). Comparative analysis<br />
<strong>of</strong> proteome pr<strong>of</strong>iles found significant difference (p grade<br />
II) and samples (n=50) <strong>of</strong> 23 patients without aGvHD. The application<br />
<strong>of</strong> <strong>the</strong> aGvHD specific pattern and calculated model allowed correct classification<br />
<strong>of</strong> blinded and prospectively collected urine samples with high<br />
accuracy: The model enabled <strong>the</strong> diagnosis <strong>of</strong> aGvHD > grade II with a<br />
sensitivity <strong>of</strong> more than 83% [95% CI 73.1 to 87.9]). High resolution proteome<br />
analysis with diagnostic peptide patterns may help to identify<br />
patients at risk <strong>of</strong> severe aGvHD development prior to clinical features<br />
(mean: 7 days, range: 1 to 13 days prior to clinical symptoms) in an unbiased<br />
laboratory based screening assay. O<strong>the</strong>r patterns (bacterial infection/septicaemia,<br />
CMV and EBV reactivation and infection) will be<br />
shown. Application <strong>of</strong> proteomic based patterns may lead to <strong>the</strong> establishment<br />
<strong>of</strong> a diagnostic tool suitable for pre-emptive <strong>the</strong>rapy <strong>of</strong> aGvHD<br />
based on proteomic patterns.<br />
haematologica/<strong>the</strong> hematology journal | 2007; 92(s1) | 57