ALBERT

Description: Background: Primary mediastinal (PMBL), GCB and non-GCB B-cell lymphoma are three major subtypes of DLBCL with distinct clinical outcomes. PMBL and DLBCL of GCB immunophenotype have a higher chance of cure than patients with non-GCB DLBCL when treated with conventional chemotherapy. In the relapsed/refractory setting, there is paucity of data regarding the impact of histological classification of DLBCL on survival following alloSCT in patients who were not eligible for or who have failed a prior autologous SCT (ASCT). Herein, we compare outcomes of alloSCT in these patients. Methods and Patients: We identified 101 de novo DLBCL patients who were treated at our center from January 1, 1998 to December 31, 2011. Seventeen (17%) tumors were PMBL. We determined the cell-of-origin in the remaining patients, based upon the Visco/Young and Choi immunohistochemical algorithms, and classified 62 (61%) as GCB and 22 (22%) as non-GCB tumors. GCB patients were older than non-GCB [median (range): 54 (21-70) vs 48(24-58) years, p=0.003)] and PMBL [26(19-48), P

Description: Abstract 4200 Graft-vs.-Host Disease (GVHD) remains a common complication following matched sibling and unrelated donor hematopoietic cell transplantation (HCT). Standard GVHD prophylaxis calls for prolonged immune suppression, typically with a calcineurin-inhibitor. Recently, post-transplant cyclophosphamide (CY) has been studied as sole prophylaxis in matched related and unrelated bone marrow (BM) transplant recipients following an ablative conditioning regimen with busulfan (Bu) and CY and demonstrated comparable rates of acute GVHD and lower rates of chronic GVHD as traditional GVHD prophylaxis regimens. We recently conducted a phase II study of post-transplant CY following a reduced-intensity conditioning (RIC) regimen of Busulfan (Bu) and Fludaribine (Flu) in matched related and unrelated donor transplants and performed a matched-control analysis comparing their results with patients who received traditional GVHD prophylaxis with tacrolimus and mini-dose methotrexate (MTX) during the same time period. Forty-nine (49) patients were enrolled onto this study. They received Flu at a dose of 40mg/m2 over 1hour followed by intravenous Bu over 3 hours targeting a daily AUC of 4,000 microMol-min on days –6 to –3. Recipients of unrelated transplants received ATG on days –3 to –1 (total dose 4 mg/kg). CY was given as sole GVHD prophylaxis at a dose of 50 mg/kg on days +3 and +4. During the same period of time, 133 patients received a RIC regimen with intravenous Bu/Flu or Flu and melphalan (Mel) and received GVHD prophylaxis with tacrolimus plus mini-dose MTX (10mg/m2 on day +1, 5mg/m2on days +3, +6, +11). Unrelated donor transplants also received ATG. A computer generated algorithm was used to identify a comparable control group from our departmental database matching, in order of priority, on age, diagnosis, disease status, donor (matched-related versus unrelated) and graft source (PB versus BM). Matched controls (control group) were successfully identified for 37 study patients (Post-Cy group). Results: The median age for the Post- CY group and control group was 61 (range, 39–72) and 62 years (range, 37–72). Eight-one (81) % of patients in both groups had AML or MDS, 3% had ALL and 16% had NHL or CLL. Fifty-nine (59) % of patients in both groups had unrelated donors and received ATG in the conditioning. Disease status for the Post-CY and control groups respectively were CR1: 14 and 14%, CR2: 8 and 11%, 〉CR2: 38 and 32% and Primary Induction Failure / Untreated: 40 and 32%. Seventy (70) % of the post-CY group received BM versus 48% of the control group, whereas sex mismatching (Male donor for Female patient) occurred in 22% of the post-Cy and 8% of the control group. The cumulative incidence of grade II-IV acute GVHD and chronic GVHD in the post-CY and control groups were: 46% versus 19% (Hazard Ratio (HR): 2.8, 95% CI, 1.1–6.7; p=0.02) and 14% versus 21% (HR: 0.8, 95% CI, 0.2–2.6, p=0.7). Grades III/IV acute GVHD occurred in 14% (95% CI, 6–32) of the patients in the post-CY group whereas there were no cases of grade III/IV in the control group (p=0.02). Overall, progression-free and non-relapse mortality at 2-years are shown in the table below. Conclusion: Post-transplant CY following RIC is associated with higher rates of acute GVHD, with resultant trends for higher non-relapse mortality and lower overall survival when compared to tacrolimus and mini-dose MTX. The use of post-transplant CY as a sole GVHD prophylaxis regimen should be avoided following RIC transplant in matched-related and unrelated donors. Disclosures: Off Label Use: azacitidine: off-label use as maintenance therapy following allogeneic stem cell transplant for MDS/AML.

Description: Abstract 3003FN2 Objective: Progressive disease (PD) remains the main cause of treatment failure following RIC allo-SCT in HL (Peggs KS et al, BJH 2008: 143; 468). We elected to explore the feasibility of adding gemcitabine (G), a highly active agent in relapsed/refractory HL (Santoro A et al, JCO 2000 :18; 2615), to our standard fludarabine (F) - melphalan (M) RIC in patients with relapsed and refractory HL undergoing allo-SCT, with the ultimate goal to augment cytoreduction and reduce the incidence of PD. Main G-associated non-hematologic toxicities include pulmonary, skin toxicities and mucositis. Patients and Methods: Between 8/07 and 3/11, fifteen consecutive HL patients (Unique Patient Number/UPNs 1–15) underwent an allo-SCT with the G-FM regimen. They had failed multiple conventional treatments (median prior chemotherapy regimens: 4; range 2–9), radiation therapy (6/15) and a prior auto-SCT (7/15). The median age was 33 years (range 20–46). Disease status at SCT was chemosensitive relapse (n=4), untreated relapse (n=1), induction failure chemosensitive (n=2), complete remission undetermined (CRU1 and CRU2) (n=8). The median time to PD after auto-SCT was 10 months (range 3–19). The donor was an HLA-identical sibling (n=10) or matched unrelated donor (MUD; n=5, with UPN 6 having a 9/10 match). The conditioning regimen was G 800 mg/m sq IV x1 at day -7 (two patients, UPN 6 and UPN 7 received G 1000 mg/m sq IV x2, day -5 and day -2 as part of a dose escalation attempt stopped for excess toxicity), F (32 mg/m sq IV x4, day -5 to day -2), M (70 mg/m sq IV x2, day -3 and day -2; total dose 140 mg/m sq). Thymoglobulin (4 mg/kg IV) was added in MUD allografts. Graft-vs-host disease (GVHD) prophylaxis included tacrolimus and mini-dose methotrexate (5 mg/m sq). Results: Myeloid recovery was prompt, with an absolute neutrophil count (ANC)〉500/mcL at day +12 (range 11–20). Median platelet recovery at 20K/mcL was at day +14 (range 9–26). Chimerism studies indicate 100% donor-derived engraftment in 13/13 evaluable patients (100%). Day 100/overall transplant-related mortality (TRM) were 2/15 (13%) and 2/15 (13%), respectively. Acute GVHD (grade II-IV) occurred in 8/14 evaluable patients, chronic GVHD in 7/13 evaluable patients (extensive in 7). Pulmonary toxicity was seen in four patients (26%). Grade 4 pulmonary toxicity was seen in UPN 6. Otherwise it was grade 1–2 (n=3). Cutaneous toxicity (skin rash, responsive to steroid therapy) was seen in five patients (33%: grade 3 n=1; grade 1–2 n=4). Mucositis was seen in nine patients (60%). It was grade 3 (n=1); grade 1–2 (n=8). Other organ toxicities were not seemingly markedly different from the ones seen with FM140 only. Three patients expired (graft rejection n=1, in UPN 6; CMV pneumonia n=1; PD n=1). Twelve patients are alive (ten progression-free, eleven in CR/CRU) with a median follow-up of 18 months (range 2–33). Actuarial rates of overall survival (OS) and progression-free survival (PFS) at 24 months are 87% (95% CI: 56–96) and 49% (95% CI: 18–74), respectively (see Figure). While a formal comparison with our historical experience with the FM140 regimen is clearly not possible, OS and PFS (actuarial estimates) at 24 months were 64% (95% CI: 49–76), and 32% (95% CI: 20–45), respectively (Anderlini et al, Haematol 2008; 93 :257). Conclusions: The addition of G to FM140 is feasible. Pulmonary, cutaneous toxicities, as well as mucositis, while clinically significant, were manageable and did not contribute to mortality in the patients treated with only one G dose. While quite encouraging, these data remain preliminary. The G-FM regimen deserves further study in a larger cohort of patients. Disclosures: Off Label Use: Gemcitabine, fludarabine and melphalan as part of conditioning regimen for allogeneic stem cell transplantation.

Description: Abstract 894 Background: We previously reported the use of fludarabine, cyclophosphamide and rituximab as a nonmyeloablative allogeneic conditioning for CLL and non-Hodgkin's lymphomas (Khouri et al, Blood 2008;111:5530). Bendamustine is a novel compound which was found to be effective in pts who were refractory to alkylating agents. In order to improve outcomes in nonmyeloablative stem cell transplantation (NST), we substituted cyclophosphamide with bendamustine in the conditioning. Methods: Bendamustine was given iv in an escalated dose of 70, 90, 110, 130 mg/m2 daily on days −5 to - 3 prior to NST, together with 30 mg/m2 of fludarabine given on the same days. Rituximab was given at a dose of 375 mg/m2 on day –13 and 1000 mg/m2 on days −6, +1, and +8, as previously described. Tacrolimus and methotrexate was used for GVHD prophylaxis. In addition, thymoglobulin of 1 mg/kg was given on days −2, and −1 in pts receiving an unrelated donor. Results: The study included 23 pts [Mantle cell=9, CLL=6 (2 with 17p- and 1 Richter's), Follicular =4, Diffuse large cell =3]. All pts had relapsed disease after the best conventional therapy (-ies) available. Median age was 60 (range, 30–70) years. Median prior treatments was 2 (range, 1–5); 3 pts (13%) had failed a prior autotransplant. At NST, 11 pts (48%) were in CR, 7 (31%) in PR, 1 (4%) induction failure/sensitive, and 4 pts (17%) had refractory disease. Fifteen pts (65%) received their transplants from HLA-compatible siblings and 8 (35%) from unrelated donors. Continual Reassessment Method based on target toxicity at day 30 post NST was used for dose finding. The number of pts who received the 70, 90, 110, 130 mg/m2 daily doses of bendamustine were 2, 3, 3, and 15 pts, respectively. No dose-limiting toxicity was observed. Fourteen pts (61%) did not nadir to an ANC〈 500; 3 pts did not require neupogen for ANC recovery; 19 (83%) did not experience a platelet count 〈 20,000/mm3. All pts engrafted donor cells. Median donor T cells at day 30 was 93%. Only one pt developed acute GVHD (it was grade 3); none had acute grade 2 or 4. Chronic extensive GVHD was observed in 2 of 22 (9%) evaluable pts. Fungal infection was the cause of the only death observed. Twenty pts (87%) achieved CR, 2 (9%) have ongoing PR, and 1 (4%) had SD. With a median follow-up time of 8 months (range, 3–25 months), the OS and PFS rates were 92% and 79%, respectively. Conclusions: Our results represent the first report to suggest that combining bendamustine at a dose up 130 mg/m2 daily × 3 days, together with fludarabine and rituximab is safe and constitutes a well tolerated conditioning for NST. We are treating pts with this regimen in the outpatient setting. The study is currently ongoing to assess its efficacy in a larger cohort of pts. Disclosures: Khouri: Cephalon: Research Funding. Off Label Use: Bendamustine use in SCT.

Description: Abstract 3537 Donor-recipient HLA mismatches are associated with increased morbidity and mortality after UD hematopoietic stem cell transplants (HSCT). We hypothesized that HLA-DP mismatches would worsen outcomes of HSCT using donors mismatched at HLA-A,-B,-C,-DRB1 or -DQB1 and evaluated 391 consecutive patients (pts) with myeloid malignancies treated at our institution with 0,1,2,3 mismatches out of 12 alleles typed by high resolution at HLA-A,-B,-C,-DR,-DQ,-DP loci. Eighty-one pts were 12/12, 180 pts were 11/12, 113 pts were 10/12, and 15 pts were 9/12 HLA match with the recipients. Characteristics of the 4 groups (12/12, 11/12, 10/12, 9/12) were similar except source of stem cells; 87% of pts with 9/12 donors received bone marrow versus 60–62% for the other 3 groups. Results: Two-year overall survival (OS) and progression-free survival (PFS) were 40%, 44%, 45%, 53% and 33%, 40%, 44%, 49%, respectively (p=NS). However, OS was significantly worse with increasing number of mismatches for patients with AML/MDS with poor-risk cytogenetics (p=0.005, HR 1.6, 95% CI 2.1–4.2). Except for the 9/12 group, pts had a significantly higher non-relapse mortality (NRM) (11%, 24%, 36%) and lower risk of progression (32%, 25%, 20%). In the 9/12 group, NRM was 27% and progression rate was 40%. Grade II-IV, III-IV aGVHD as well as cGVHD were also progressively worse with increasing number of mismatches. Gr II-IV and III-IV aGVHD rates were 35%, 37%, 41%, 69%, and 8%, 8%, 15%, 16%, respectively. Cumulative incidence of cGVHD was 35%, 39%, 44% and 61%, respectively. Compared with 11–12/12 donors, pts who received a 9–10/12 donor had significantly higher rates of gr III-IV aGVHD and cGVHD (p=0.03, HR 2.1, CI 1.1–3.7 and p=0.02, HR 1.5, CI 1.1–2.1, respectively). Univariate analysis revealed that there is less NRM with a 12/12 donor (vs. other) (p=0.008, HR 1.9, 95% CI 1.2–2.9), while in multivariate analysis, compared with a 12/12 donor, the use of a donor with mismatch was significantly associated with higher NRM [HR and 95%CI were 2.1 and 1.04–4.4 for 11/12 donor (p=0.04); 3.1 and 1.5–3.3 for 10/12 donor (p=0.003); 2.8 and 0.9–9.3 for a 9/12 donor (p=0.08), respectively] (Figure). In multivariate analysis, factors significantly associated with OS were disease status at transplant (active disease vs. not) (p

Description: Abstract 3445 Background: Relapsing AML/MDS after HSCT has a dismal prognosis, with few patients achieving long-term control of the malignancy. AZA is a hypomethylating agent that is moderately active against AML/MDS, and may have beneficial immunomodulatory effects after HSCT. We have shown that a significant minority of patients with recurrent disease respond to this drug. Here, we present long-term follow-up after salvage treatment regimens that included AZA, to treat AML/MDS that recurred after HSCT. Patients and Methods: Twenty-three patients received low-dose AZA for recurrence. Decision to use AZA was based on clinical assessment of slow progression of disease and relatively slower disease ‘tempo' and relatively small AML bulk. AZA cohort preparative regimens for 1st HSCT were myeloablative in 12 cases, and of reduced intensity in 11 cases. AZA was used prior to or without a 2nd HSCT (n=17), or after a 2nd HSCT (n=6). Outcomes were compared to controls (n=18) that relapsed ≥ 8 months after HSCT, and did not receive AZA (8 months representing the median disease free survival (DFS) for AZA-treated patients). The control group included all patients that relapsed ≥ 8 months after allogeneic HSCT using myeloablative busulfan 130 mg/m2 and fludarabine 40 mg/m2 for 4 days. AZA was studied as a time dependent variable. AZA and controls had similar baseline characteristics as described in the Table, although median DFS after the first HSCT was 8 (range: 2–51) and 17 (range: 7–59) months, favoring the control group (p=0.08). AZA was administered outpatient, with good tolerance. Fatigue and nausea were commonly observed toxicities. Doses were 8 mg/m2 (n=1), 16 mg/m2 (n=3), 24 mg/m2 (n=10), 32 mg/m2 (n=5), 40 mg/m2 (n=2), and 75 mg/m2 (n=2), administered subcutaneously for 5 days, in 28–32-day cycles. Results: Median number of cycles was 4 (range, 1–44). With a median follow-up of 18 months for AZA and control patients, median survival after relapse was 17 versus 6 months, respectively for AZA and control patients. 11 (48%) AZA patients are alive, while 2 (11%) control patients are alive. Two-year overall survival (OS) for AZA and control groups was 40% and 10%, respectively. AZA and controls had similar baseline characteristics as described in the Table. Conclusion: Low-dose AZA was a well tolerated outpatient treatment that may improve survival after AML/MDS recurrence in selected cases. Major determinants of survival in this setting, however, were remission duration after HSCT, and use of a 2nd HSCT. Disclosures: No relevant conflicts of interest to declare.

Description: Abstract 3510 Background: The outcome in FL pts relapsing after ASCT has not been well studied. While the use of nonmyeloablative allogeneic stem cell transplantation (NST) has been promising in this setting, one could not underestimate the results of selective inclusion criteria. Methods: All patients with FL who experienced a relapse after ASCT at The University of Texas M. D. Anderson Cancer Center between 1997 (the initiation of the rituximab and NST era) and 2007, were analyzed. Results: Fifty pts were identified. Fifteen (30%) pts underwent NST after ASCT relapse (Group A); 25 (50%) pts (Group B) met the eligibility criteria for NST but were not allotransplanted due to either physician/pt preference (n=14), lack of suitable donors (n=5), insurance requirements (n= 3), other causes (n=3). Ten (20%) pts (Group C) were not eligible for NST due to refractory disease or co-morbidities. At the time of progression after ASCT, the comparison of pts who were considered for NST showed that Group A pts and Group B pts had a comparable age [median 57 years(range, 45–64) vs 58 years (range, 40–70), p=0.7, respectively], serum LDH level (p=0.1), # of extranodal sites of disease (p=0.4), marrow involvement (p=0.1), and stage III/IV (p=0.2). The histology subtypes were also equally distributed in both groups (FL grades 1, 2, and 3 were found in 27%, 40% and 30% of Group A pts, respectively, and in 16%, 44% and 40% of Group B pts, respectively). The majority of pts in Groups A and B had a good ECOG PS of 0–1 (100% vs 96%, respectively). The median time from ASCT to progression was 16 months (range, 4–42) in Group A, and 19 months (range, 3–99) in Group B. Group A pts underwent their NST at a median of 8 months after their ASCT relapse. At their progression post ASCT, pts in Group B were treated with either rituximab as a single agent (n=12, 48%), or combination chemo-antibodies (n=7, 28%); the therapy received in the remaining 6 (24%) pts was unknown. With a median follow-up time of 49 months (range 23–113 months) for Group A, and 37 months (range, 17–130 months) for Group B, the actuarial survival rates at 4-year were 73%(%95 CI, 42–89) and 71% (%95 CI, 46–86), respectively, (Figure, p = 0.9). The causes of death in Group A were related to infection (1), organ failure (1), progression (2), and unknown (1). Five pts in Group B died of progression and one died of secondary leukemia. Pts in Group C had a median survival time of 7 months; only 2 pts of this group were still alive at the time of this analysis was made. Conclusions: Single institution results show that 30% of pts with FL relapsing after ASCT undertook an allotransplant. While allogeneic NST is an effective therapy for these pts, a significant proportion of pts can be observed for several years before an allotransplant should be considered. Disclosures: No relevant conflicts of interest to declare.

Description: Abstract 3053 Recovery of the immune system is critical for the success of allogeneic stem cell transplantation. Several groups, including ours, reported that faster lymphocyte recovery was associated with improved outcomes. There is limited information however, on factors affecting lymphocyte recovery and its influence on outcomes specifically after BMT. We retrospectively assessed 536 consecutive patients (pts) with acute leukemia (452 AML/MDS and 102 ALL) who underwent BMT (and engrafted) at our institution between 01/1999 and 12/2010 to determine 1) predictors of early lymphocyte recovery (ELR), and 2) influence on outcomes of ELR defined as achieving absolute lymphocyte count of 1000/μL (ALC1000) by day 100 after BMT. Characteristics of the study population including demographics, graft, disease, and transplant characteristics assessed are described in the Table. Conditioning regimens were classified as reduced intensity (RI), or high intensity (including TBI-, busulfan-, or melphalan-based) conditioning, as previously described. Cox's proportional hazards regression analysis was used to assess predictors of ELR and NRM on univariate and multivariate (MV) analysis. Pts who developed grade II-IV acute GVHD (aGVHD), received a second graft infusion, or relapsed before achieving ALC1000 were censored at the time of these events for the assessment of predictors of ELR. Only pts who were alive progression-free, and had not developed grade II-IV aGVHD by day 100 were eligible for the assessment of predictors of outcomes in landmark analysis starting on day 100 after BMT. Outcomes were assessed at the median follow-up in surviving pts of 40 months. On multivariate analysis, significant predictors of lower rate of ELR included TBI- or melphalan-based ablative conditioning (HR=0.5, p 0.003) (compared to RI or busulfan-based), and a haploidentical donor (HR=0.4, p 0.05). Preliminary analyses assessing CMV reactivation and grade II-IV aGVHD as time dependent variables showed these factors to be associated with higher and lower ELR, respectively. MV analyses incorporating these factors in prognostic model are ongoing and final results will be presented at the meeting. Among the 246 pts eligible for the outcomes assessment, ELR (HR=0.2, P

Description: Background: Noncoding RNAs play an important role in the pathogenesis of CLL. Recent publications suggested that higher miR-155 plasma levels (above 56th percentile) were associated with a significantly lower survival rate (P= 0.0122) in CLL patients treated with conventional treatments (Ferrajoli et al. Blood 2013;122:1891). Other studies revealed that expression of miR-29b or miR-181b significantly inhibits T cell leukemia/lymphoma 1(Tcl1) oncogene (Pekarsky et al. Cancer Res 2006;66:11590) whereas high Tcl1 expression correlates with aggressive CLL phenotype showing unmutated immunoglobulin variable region genes and ZAP70 positivity. The impact of these miR expressions in CLL patients undertaking allogeneic stem cell transplantation (alloSCT) is unknown. Methods and Patients: We identified 41 patients with relapsed/refractory CLL who had received a non-myeloablative alloSCT at our center and in whom pre-transplant serum samples collected before initiating the patients’ allogeneic conditioning were available. We measured the serum expression of miR-155 as well as miR-15a/16-1 cluster, miR-29b and miR-181b. Total RNA was isolated from 100 µL of serum using the Total RNA Purification Kit. Serum miRNA levels were measured by qRT-PCR (TaqMan MicroRNA assays). The relative serum levels of miR-155, miR -15, miR-29b, miR-181b, and miR-16-1 were calculated using the equation 2−ΔCt, where ΔCt = mean CtmiRNA – mean Ctcel-miR-39, and Ct = threshold cycle. Twenty fmol of synthetic C. elegans miRNA (cel-miR-39) was spiked into each serum samples to normalize the experimental qRT-PCR data (cel-miR-39 Ct mean ± SD = 17.777± 0.628). Forty one patients were initially evaluated. Thirteen of these 41 patients were later excluded as they received alemtuzumab for graft-versus-host disease (GVHD) prophylaxis. Therefore, our analysis was limited to 28 patients who received their alloSCT between 2000-2010. Median age (range) was 59 (45-70) years. Median number of prior therapies was 3 (range, 2-8). Eleven (39%) had a beta-2 microglobulin level of 〉3 mg/L. Ten of 12 (83%) patients with data that could be evaluated had unmutated immunoglobulin variable-region heavy-chain gene, and 4/19 (21%) had 17p13.1 deletion. 46% of patients had refractory disease at transplantation. The proportion of patients who received a matched related and a matched unrelated donor was 68% and 32%, respectively. All patients received non-myeloablative conditioning with fludarabine, cyclophosphamide, and rituximab as previously published (Khouri et al. Cancer 2011;117:4679). GVHD prophylaxis consisted of tacrolimus and methotrexate. Results: Median (range) follow-up months was 68 (43-141). OS from the time of alloSCT was studied according to the relative expressions of miR under investigation (low, below median; high, above median). The 5-year OS rates of patients with low and high mir-155 were 63% and 50%, respectively (HR=1.6, P=0.4; Figure). No statistically significant differences were found in the other miRs studied. The 5-year OS was 57% for both the low and high miR-15a expression. The 5-year OS in low and high miR-29b and miR-181b were 52% and 61% (HR=0.8; P=0.7), and 60% and 53% (HR=1.1, P=0.9), respectively. The 5-year OS in low and high miR-16 were 43% and 71% (HR=0.4, P=0.2). Conclusions: Our preliminary results suggest that serum levels of miR-155, miR -15, miR-29b, miR-181b, and miR-16 are not a statistically significant prognostic biomarker for survival in relapsed/refractory CLL undertaking alloSCT. Figure 1 Figure 1. Disclosures No relevant conflicts of interest to declare.

Description: Chronic graft-versus-host disease (cGVHD) is the most important long-term complication of allogeneic hematopoietic cell transplantation (HCT). As cGVHD is a fibroproliferative disorder, assessment of plasma biomarkers known to be associated with fibrosis and ECM turnover may provide a novel platform for risk stratification and prognostication. Among ECM biomarkers, we focused this analysis on TGFβ due to its significant profibrotic and immunomodulating properties. Plasma samples were prospectively obtained at the time of cGVHD onset, and ECM biomarkers were assayed simultaneously using a lab-derived (non-commercial) Luminex bead-based assay. TGFβ values at cGVHD diagnosis were available for 112 of 158 consecutive adult patients (pts) who had received HCT at the University of Michigan from 2007 to 2010. All characteristics (except for age) were assessed at the time of cGVHD diagnosis. TGFβ was evaluated in tertiles. The first (N=38), second (N=37), and third (N=37) tertiles are described as low, intermediate, and high TGFβ groups. Median time from cGVHD diagnosis to NRM was 8 (range 4-14), 25 (range 24-37), and 12 months (range 7-17) in the 3 TGFβ groups, respectively. The 2-years cumulative incidence of NRM was equally high in the low (25%) and high (23%) TGFβ groups compared with the intermediate group (0%, Figure 1). cGVHD was the predominant cause of NRM in the 3 groups. The 2-years actuarial progression-free survival was higher in the intermediate group (83%) compared with the low (55%, HR=4.6, P 0.005) or high (66%, HR=2.9, P 0.06) TGFβ groups. The median time from transplant to cGVHD diagnosis was comparable across the 3 groups (Table). Patients in the low TGFβ group were more likely to exhibit common adverse prognostic factors in cGVHD including higher prevalence of poor performance status (P 0.02), low platelet count (P 0.02), low eosinophil count (P 0.02), and moderate to severe lung involvement (P 0.01). These pts also tended to be more likely to have a history of grade III-IV acute GVHD and to present with quiescent cGVHD. The severity of cGVHD skin, liver, or gastrointestinal involvement did not differ across the 3 groups. Remarkably, the prevalence of being on steroids was comparable in pts with low (21%) and intermediate (19%) TGFβ values despite significant differences in the rate of NRM between these 2 groups. In contrast, 8% of pts in the high TGFβ group were on steroids. We could not identify any clinical characteristics distinguishing pts in the intermediate TGFβ group. Age, gender, disease status at transplant, type of conditioning regimen, type of GVHD prophylaxis, donor type, and stem cell source did not differ across the 3 TGFβ groups. Among the ECM markers evaluated, low TGFβ levels were correlated with low levels of surfactant protein-D, chemokine (C-C motif) ligand 2, and detectable IL-13Rα ; and with high levels of collagen, Tenascin-C, and metalloproteinase-2 and -9. Collectively these data indicate that extreme (low and high) TGFβ values at cGVHD diagnosis are associated with significantly faster NRM rate. Different pathologic mechanisms may account for the association with NRM in these 2 groups as suggested by the distinct clinical and biomarkers profiles of the low TGFβ group. The independent prognostic value of TGFβ in cGVHD is analyzed through multivariate analysis. Table TGFβ Group Low Intermediate High P Characteristics (assessed at cGVHD diagnosis, except age) N=38 N=37 N=37 Low vs. Other TGFβ values, median (range) 6K (800-11K) 22K (12K-34K) 46K (35K-105K) Age at transplant, yrs median (range) 52 (19-67) 52 (22-66) 54 (18-68) 0.1 Days from HCT to cGVHD diagnosis, median (range) 162 (37-485) 176 (52-547) 155 (63-336) Concomitant infections 7% 8% 8% 0.6 Karnofsky performance status ≤ 70% 66% 81% 89% 0.02 Platelet count ≤100,000/mm3 37% 16% 19% 0.02 Eosinophil count ≤0.1 K/µL 26% 14% 5% 0.02 NIH -CC lung involvement None-Mild 75% 95% 92% 0.01 Moderate-Severe 25% 5% 8% On steroids 21% 19% 8% 0.09 History of grade II-IV acute GVHD 39% 24% 32% 0.2 cGVHD Presentation Progressive 24% 27% 27% Quiescent 42% 27% 27% 0.1 de novo 34% 46% 46% Organ Involvement according to NIH-CC Skin None or Mild 82% 57% 78% Moderate 5% 24% 11% Severe 13% 19% 11% Liver None 39% 54% 35% Mild 21% 24% 30% Moderate 18% 14% 11% Severe 21% 8% 24% Gastrointestinal None 74% 76% 86% Mild 24% 19% 11% Moderate 3% 5% 3% Number of Organs Involved 1 21% 19% 8% 2 26% 16% 47% 3 36% 46% 30% 4 10% 16% 8% 5 or 6 5% 5% 6% Figure 1 Figure 1. Disclosures No relevant conflicts of interest to declare.