Azacitidine Induces Expansion of Regulatory T Cells and Tumour Specific CD8+T Lymphocytes After Allogeneic Stem Cell Transplantation: A Strategy for Epigenetic Manipulation of a Graft-Versus-Leukemia Response

Abstract 152 Acute graft-versus-host disease (GVHD) is the primary limitation of allogeneic hematopoietic cell transplantation (HCT). Current diagnostic tests do not predict a patient's response to therapy, particularly at GVHD onset, when risk-stratification is most beneficial. We hypothesized that biomarkers discovered with an intact proteomic analysis system (IPAS) approach that we have previously described ( Paczesny et al . 2010) will discriminate between therapy responsive (resolution or improvement of GVHD by day (D) 28 post-therapy initiation) and unresponsive (absence of complete or partial response) patients and predict survival in patients receiving GVHD therapy. We first performed an IPAS comparing pooled plasma taken at D16 ± 5 post-therapy from 10 responders (R) and 10 non-responders (NR). Ten candidate biomarkers with an NR/R ratio of > 1.5 in the IPAS were measurable by ELISA. We therefore measured concentrations in the 20 individual plasma aliquots. Five were significantly increased in NR vs. R, with an area under the receiver operator characteristic curve of ≥ 0.85 (ST2, IL1sRII, MIF, LYVE, and Lipocalin). These were then measured at therapy initiation (D0), with 6 previously validated diagnostic biomarkers of GVHD (IL2Rα, TNFR1, HGF, IL8, Elafin, a skin-specific marker, and Reg3α, a gut-specific marker) in plasma samples from a validation set of 381 patients with acute GVHD grade 1–4 at onset and treated with systemic steroids at the University of Michigan from 2000 to 2011. Preliminary analyses (not shown) determined that D0 measurements predicted D28 non-response and D180 OS. HLA match (match vs. mismatched; Odds Ratio (OR) 1.5, p = 0.07), conditioning intensity (full vs. reduced; OR 1.7, p = 0.04), and GVHD onset grade (grade 3–4 vs. grade 1–2; OR 2.2, p = 0.001) predicted day 28 non-response in univariate analysis, while age at transplant (≥ 55 years vs. < 55 years), donor (unrelated vs. related), and stem cell source (peripheral blood vs. bone marrow/cord) did not. After adjustment for the 3 clinical characteristics which predicted D28 response, multivariate analysis of the 11 protein concentrations showed that 3 predicted D28 response (ST2, p = 0.001; IL1sRII, p = 0.07; and IL8, p = 0.03) and 7 predicted post-therapy D180 OS (ST2, p = 0.003; IL1sRII, p = 0.07; IL8, p = 0.05; Elafin, p = 0.06; MIF, p = 0.04; TNFR, p = 0.03; and Reg3α, p = 0.002 in gut-GVHD subset). Using logistic regression, we examined the ability of both the 7 biomarkers and ST2 alone to predict for D28 non-response, as ST2 was the most significant marker in all previous analyses. ST2 is the IL33 receptor, a member of the IL1/ Toll-like receptor superfamily, which promotes a Th2-type immune response in diseases such as arthritis and asthma ( Kakkar et al. 2008 ). A high biomarker value was defined as a plasma concentration greater than 50% above the median value of the responders' group. A high panel was defined as having at least 5 of 7 high biomarkers. Patients with high ST2 levels were 2.6 times more likely not to respond to therapy independent of the aforementioned significant clinical characteristics (p < 0.001) while patients with a high panel were only 1.9 times more likely not to respond (p = 0.004). Thus, only ST2 measurement was used for further analyses. Because ST2 concentrations correlated with response, we hypothesized that ST2 would predict D180 non-relapse mortality (NRM), independent of GVHD onset grade, the strongest clinical predictor of NRM (20% for GVHD grade 1–2 vs. 50% for GVHD grade 3–4, Hazard ratio (HR) 3.0, p < 0.001). NRM cumulative incidence curves and HR for the 4 risk categories of low ST2 / grade 1–2, low ST2 / grade 3–4, high ST2 / grade 1–2, and high ST2 / grade 3–4 are shown in Figure 1. Interestingly, patients presenting with high clinical grade and low ST2 had a good prognosis, suggesting that ST2 provides important prognostic information at initiation of therapy above the clinical grade. ![Figure][1] In conclusion, soluble ST2, the form measured by ELISA, is a decoy receptor that drives the Th2 phenotype toward Th1, a mechanism by which it may act in the pathophysiology of resistant GVHD. ST2 concentrations obtained at initiation of GVHD therapy significantly enhance the accuracy of outcome prediction independent of GVHD grade. Measurement of ST2 may allow for early identification of patients at risk for subsequent non-response and mortality, and may provide a promising target for novel therapeutic interventions. Disclosures: No relevant conflicts of interest to declare. [1]: pending:yes