Citations

7516 Background: PCNSL is a diffuse large B cell lymphoma confined to the CNS. Pomalidomide (POM) is a novel immunomodulatory agent with excellent CNS penetration (~40%) based on CNS PK analysis in rats and pre-clinical therapeutic activity against CNS lymphoma. Methods: A Phase I clinical trial was undertaken to determine the maximal tolerated dose (MTD) of POM, safety profile and overall response rate (ORR). Treatment consists of POM daily for 21 days every 28 days in combination with dexamethasone 40 mg PO weekly for two cycles followed by POM alone in subsequent cycles until progression or intolerance. 4 dose escalation levels of POM (3mg, 5mg, 7mg, and 10 mg) were planned. Thromboprophylaxis with oral anticoagulant or aspirin was required. MTD determination has been completed and expansion of the MTD cohort is ongoing. Therapeutic responses were evaluated per the international PCNSL collaborative group (IPCG) criteria after 2 cycles of treatment. The trial is registered with ClinicalTrials.gov (#NCT01722305). Results: 21 of 25 patients accrued were eligible for assessment. The MTD was determined to be 5 mg qd for 21 days every 28 days. Two DLTs were seen at dose level 3 (Grade 3 dyspnea and grade 4 thrombocytopenia). One DLT was seen in the expanded MTD cohort (Grade 4 neutropenia and lymphocytopenia). ORR for the study (9/21) was 43% (95% CI- 22%, 66%) with 4 CR, 1 CRu and 4 PR. 3 responders completed 2, 4, and 6 cycles before progression. 6 responders have completed 4, 5, 6, 10, 12, and 32 cycles and remain on treatment. ORR for the MTD dose level was (5/12) 42% (95% CI- 15%, 72%) with 3 CR and 2 PR. 2 patients had stable disease (SD). Pseudo-progression was seen in 1 patient. Overall, grade 3/4 toxicity was hematologic (neutropenia, anemia, and thrombocytopenia) in 38.1% and non-hematologic in 33.3% (fatigue, pneumonia, sepsis, syncope, dyspnea, hypoxia, respiratory failure and maculo-papular rash). Percent CSF/blood ratio of POM was determined to be 19% in 1 patient. Conclusions: Pomalidomide treatment is feasible with therapeutic activity against relapsed/refractory PCNSL and should be further developed. Clinical trial information: NCT01722305.

7512 Background: Outcomes in activated B cell subtype diffuse large B cell lymphoma (ABC-DLBCL) and r-NHL are poor (Sehn Blood 2015, Crump ASCO 2016). ZUMA-1 is the first, multicenter trial of anti-CD19 chimeric antigen receptor (CAR) T cells, axi-cel, in r-NHL. Methods: Dosing and eligibility were per Neelapu ASH 2016. The primary endpoint was objective response rate (ORR); secondary endpoints were duration of response (DOR), overall survival (OS), and safety. Cell of origin (COO) and CD19 status were assessed centrally using Lymphoma Subtyping Test NanoString (Wallden JCO 2015) and a validated immunohistochemistry assay, respectively. Results: As of Jan 27, 2017, 111 patients (pts) were enrolled; the manufacturing success rate was 99% with an average 17-d turnaround time; 101 pts (modified intent-to-treat [mITT] population) received axi-cel. In the mITT population, the ORR was 82% (complete response [CR], 54%). With 8.7 m median follow-up, 44% remain in response and 39% in CR. The median DOR was 8.1 m and not reached (NR) for pts with CR. Median OS was NR. Results for clinical and biologic covariates are listed in the table. In pts who received tocilizumab (n = 43) and/or steroids (n = 27) for cytokine release syndrome (CRS) and/or neurologic events (NE), ORR was 84% and 78%, respectively. Most common grade ≥3 adverse events (AEs) were neutropenia (66%), leukopenia (44%), anemia (43%), febrile neutropenia (31%), thrombocytopenia (24%), and encephalopathy (21%). Rates of grade ≥3 CRS and NE were 13% and 28%, respectively. There were 3 grade 5 AEs (Neelapu ASH 2016). Conclusions: Axi-cel induced an ORR of 82% in pts with r-NHL, response is ongoing in 44% of pts at 8.7 m. Similar clinical responses were observed in pts with r-ABC-DLBCL. AEs were manageable and the use of tocilizumab/steroids did not appear to impact ORR. Drs Locke and Neelapu contributed equally. Funding source: Kite Pharma and Leukemia & Lymphoma Society Therapy Acceleration Program Clinical trial information: NCT02348216. [Table: see text]

7056 Background: No therapies are approved for MDS after HMA failure. 04-24 was a single-arm study to evaluate best BMBL response as a potential surrogate for OS in higher-risk (HR) MDS pts who progressed on or after an HMA. Rigo is a Ras-mimetic that inhibits the RAS-RAF-MEK pathway, which is frequently activated in HR MDS (Athuluri-Divakar Cell 2016; Gil-Bazo Cancer Biol Ther 2016). Methods: Eligible MDS pts had 5-30% BMBL confirmed within 6 wks pre-study and progression per International Working Group (IWG) 2006 criteria on or after HMAs within 2 yrs. Rigo 1800 mg/24 hrs was continuously infused over 72 hrs q 2 wks × 8 cycles, then q 4 wks until progression or unacceptable toxicity. Primary endpoint was relationship of best BMBL IWG response to OS by Kaplan Meier method. Results: 64 pts were treated (median 5 cycles, range 1-32+), with 61% male, median age 73 (range 47-87), median prior HMA duration 10.8 mos (range 1.2-70.2). Revised International Prognostic Scoring System scores were low 2%, intermediate 11%, high 27%, very high 53%, and unknown 8%. ≥Grade 3 adverse events in ≥10% of pts were anemia 19%, thrombocytopenia 19%, and febrile neutropenia 16%. At the analysis time 40 pts (63%) had died. Best BMBL IWG response was marrow complete response (mCR) 14 pts (22%), stable disease (SD) 30 (47%), progressive disease (PD) 15 (23%), and failure (early death/withdrawal) 5 (8%); 2 mCR pts had transplant. Median OS was 7.0 mos (95% confidence interval 4.8-10.8). Landmark median OS (from day of best BMBL response) was mCR not reached; SD 6.3 mos; PD 3.3 mos. Median OS of mCR+SD was 8.5 mos, with log-rank p = 0.011 (mCR+SD OS to PD OS). Conclusions: BMBL response is a predictor of survival for MDS pts receiving Rigo after HMA failure, confirming findings in earlier Phase 1/2 studies (Silverman ASCO 2015 Abstr 7017). Based on earlier results identifying an MDS subset benefitting from Rigo (Garcia-Manero Lancet Oncol 2016; ASCO 2016 Abstr 165681), a randomized Phase 3 trial of Rigo vs physician’s choice (INSPIRE) is ongoing to determine if Rigo improves survival after HMA failure within 9 cycles. Clinical trial information: NCTO 1928537.

3541 Background: Pembrolizumab (PEM) has activity in patients with deficient mismatch repair (dMMR) colorectal cancer (CRC). Oxaliplatin (OX) and 5FU lead to immunogenic cell death and increased antigen presentation. We hypothesized that combining mFOLFOX6 and PEM may enhance immunogenic cell death and improve outcome in patients with CRC irrespective of MMR status. Methods: Subjects ≥18 years old with untreated, unresectable CRC were assigned to a single arm study. The study had a safety run in cohort of six patients (OX 85 mg/m2, leucovorin 400 mg/m2, 5FU 400 mg/m2, 5FU infusion 2400 mg/m2over 46 hours) and PEM 200 mg Q 3 weeks, followed by a phase II cohort. The primary objective was median progression free survival (mPFS), with secondary objectives: safety and toxicity per CTCAE V4.03, median overall survival, response rate, immune related response, disease control rate, and molecular correlates. Results: Between 4/2015 and 9/2016, 30 subjects were enrolled with following characteristics: 11 female, 26 Caucasian, median age: 45 years (25-75), 3 with dMMR, 22 MMR-proficient, and 5 with no available data. During the safety run in, 2 patients had G3 febrile neutropenia (FN) and 1 G4 neutropenia. The data safety monitoring committee recommended dose reduction of mFOLFOX6 to OX 68 mg/m2, leucovorin 400 mg/m2, 5FU of 320 mg/m2, 5FU infusion of 1920 mg/m2over 46 hours and PEM 200 mg Q 3 weeks. At the data cut off (12/29/16), median follow up was 24 weeks (10-66) and 27 patients remained on study. Rate of G3/4 toxicity associated with FOLFOX/PEM and PEM alone was 36.7% and 13.2%, respectively. No further FN was observed. No grade 5 toxicity was seen on study. Best response was recorded as: 1 complete response, 15 partial response (CR +PR = 53%), and 14 stable disease, with 100% DCR at 8 weeks. One patient with dMMR had resection after 2 months of therapy with complete pathologic response. MPFS has not been reached (95% CI: 5.5 months, NR). Conclusions: Based on these preliminary results, PEM/mFOLFOX6 has acceptable toxicity though demonstrated a suggestion of increased neutropenia in the initial cohort. Clinical activity was seen in patients with untreated advanced CRC including those with proficient MMR. Clinical trial information: NCT02375672.

3025 Background: Axi-cel is an autologous anti-CD19 CAR T cell therapy. ZUMA-1 is a multicenter, registrational trial of axi-cel in patients (pts) with refractory/aggressive B-cell non-Hodgkin lymphoma (NHL). In a pre-specified interim analysis, ZUMA-1 met its primary endpoint with 76% objective response rate and 47% complete response (Blood 2016;128:LBA-6). We describe, for the first time, a tumor microenvironment immune gene signature associated with CAR T cell treatment (tx) of NHL pts. Methods: Paired biopsies, pre- and within 3 weeks post-axi-cel tx, were analyzed by digital gene expression followed by a pre-specified bioinformatics algorithm applied to IGES15 and IGES21 genes involved in immune-mediated tumor regression (Immunosign; Galon Immunity 2013). Immunosign profiles expression of a pre-defined set of effector T cell, Th1, chemokine, and cytokine genes. Expression analysis and hierarchical clustering were used to define an axi-cel-related tumor immune gene signature. Wilcoxon signed rank test with multiple test correction by FDR (Benjamini-Yekutieli) was used. Results: Gene expression profile comparisons of pre- and post-axi-cel tx biopsies from 14 pts showed profound changes in gene expression within the tumor environment after infusion. The most upregulated genes post-axi-cel tx were CCL5 (RANTES), CTLA4, and GZMA (log2 fold change > 2, P< 0.05, FDR < 0.050). Immune checkpoints PD-L1 and LAG3 were also upregulated post-axi-cel (log2 fold change > 1.6, P< 0.05, FDR < 0.055). Other genes associated with T cell proliferation, homing, and effector function were also upregulated: IL-15, GZMK, CXC3CL1 (Fractalkine), CD8A, and STAT4 (log2 fold change > 1.6; P< 0.05, FDR < 0.074). Additional baseline tumor characteristics and associative analysis will be presented. Conclusions: We define a mechanistic tumor immune gene signature in NHL pts associated with axi-cel tx. This signature comprises upregulation of T cell activation, effector, chemokine, and immune checkpoint genes. These data will potentially lead to rational optimization of T cell interventions in cancer Clinical trial information: NCT02348216.

3023 Background: Axi-cel (formerly KTE-C19) is an autologous anti-CD19 chimeric antigen receptor (CAR) T cell therapy. ZUMA-1 is a multicenter, registrational trial of axi-cel in patients (pts) with refractory aggressive non-Hodgkin lymphoma. In a prespecified interim analysis, ZUMA-1 met its primary endpoint, with a 76% objective response rate and a 47% complete response rate ( Blood 2016;128:LBA-6). Post-treatment CAR T cell blood levels were associated with objective response. Here, we describe novel associations between product characteristics and CAR T cell levels in pts. Methods: CAR T cell characteristics in axi-cel produced from 62 pts were analyzed by flow cytometry and modeled against CAR T cell levels. In vivo CAR T cell levels were measured by qPCR. T cell expansion during production (fold expansion/total days in culture) was compared with CAR T cell blood levels, using a partition analysis with expansion rates of ≥1 vs < 1. Wilcoxon 2-sample test and linear regression were used. Results: Axi-cel contained CCR7+ T cells (median, 42%; range, 15–73%), with naïve (CD45RA+/CCR7+; median, 12%; range, 1–57%), central memory (CD45RA−/CCR7+; median, 29%; range, 12–49%) phenotypes, and more differentiated CCR7− effector memory and effector T cells. On infusion, CAR T cells expanded rapidly, reaching peak levels within 2 weeks (median, 43 cells/μL; range, 1–1513), and were also measurable in all pts at 1 month (median, 2 cells/μL; range, 0.03–89). The CCR7+/CCR7− T cell ratio in axi-cel associated positively with peak ( P =0.001) and cumulative ( P =0.003) CAR T cell levels through 1 month. Axi-cel lots that expanded more rapidly during production (≥1.0-fold/d; n = 18/62) associated with higher cumulative levels of CAR T cells ( P =0.03). Other product characteristics, eg, CD4/CD8 ratio or number of infused T cells, were not significantly associated with CAR T cell blood levels. Conclusions: An association was observed between CAR T cell expansion in vivo and both the T cell growth rate during production and product cell phenotype pretreatment. A key attribute of axi-cel product was the presence of CCR7+ naïve/central memory T cells, without upfront T cell subset selection. Clinical trial information: NCT02348216.

TPS2079 Background: Patient with glioblastoma (GBM) with unmethylated MGMT promoters derive limited benefit from temozolomide (TMZ) and have dismal outcome. Prioritizing the numerous available therapies and biomarkers for late stage testing requires an efficient clinical testing platform. INSIGhT (NCT02977780) is a biomarker-based, Bayesian adaptively randomized, multi-arm phase II platform screening trial for patients with newly diagnosed GBM and unmethylated MGMT promoters. Methods: INSIGhT compares experimental arms to a common control of standard concurrent TMZ and radiation therapy (RT) followed by adjuvant TMZ. The primary endpoint is overall survival (OS). Patients with newly diagnosed, unmethylated GBM that are IDH R132H mutation negative, and with genomic data available or who consent to whole exome sequencing through the ALLELE companion study for biomarker grouping are eligible. Two experimental arms consist of concurrent RT/TMZ followed by adjuvant neratinib (EGFR, HER2, and HER4 inhibitor) or abemaciclib CDK 4/6 inhibitor), respectively, in place of TMZ. The other experimental arm is CC-115 (TORC1/2 and DNA PK inhibitor), which replaces TMZ in both the concurrent and adjuvant phases. Biomarker groups include: EGFR + patients with EGFR amplification/mutation; PI3K + patients with PIK3CA mutation/amplification, PIK3R1 mutation, AKT3amplification, PIK3C2B > 1 copy gain, or PTEN dual loss; CDK: + patients with wild-type RB1 and CDK4 amplification, CDK6 amplification, or CDKN2A > 1 copy loss. Given the lack of pretrial biomarker data and the anticipated overlap of the groups, randomization will initially be equal. As the trial progresses, randomization probabilities will be adapted based on the Bayesian estimation of the probability of treatment impact on progression-free survival (PFS). These randomization probabilities can vary among the biomarker groups so predictive biomarkers will be identified and utilized if present. Treatment arms may drop due to low probability of treatment impact on OS, and new arms may be added. Experimental arms are compared only with control and should be thought of as discrete experimental questions, with INSIGhT being open to new investigators with proposed therapeutic hypotheses. Clinical trial information: NCT02977780.

2023 Background: TRC105 is a humanized antibody targeting CD105 (endoglin), a member of the TGFβ receptor superfamily. CD105 is expressed in glioma stem cells and circulating endothelial cells (CECs) in GBM patients (pts), increasing following VEGF inhibition. N1174 was conducted in recurrent bevacizumab (bev) naïve GBM pts to investigate the hypothesis that CD105 blockade+bev can delay development of bev resistance. Methods: After a ph I cohort (15 pts) established the ph II dose of TRC105 as 10 mg/kg IV over 4 hours every 7 days in combination with standard dose bev, the ph II trial used 1:1 randomization with 90% power and 0.10 type I error to detect a 3 month (mo) difference in progression-free survival (PFS) between the two arms. CECs including CD105 positive subsets were measured by flow cytometry at baseline and multiple time points. Results: Based on 101 evaluable ph II pts, there was no significant difference in PFS between TRC105+bev and bev only (2.9 vs 3.2 mo, respectively; HR=1.14, 95% CI 0.73-1.77, p=0.57), or overall survival (10.0 vs 7.4 mo; HR 1.02 95% CI 0.63-1.65, p=0.93). Response rate (complete or partial) was 12.2% (6/49) for TRC105+bev and 11.6% (5/43) for bev only. Overall incidence of grade (gr) 3+ toxicity was higher for TRC105+bev (67.3% vs 32.7%, p<0.001) mainly due to 14 pts with gr 3 anemia in the TRC105+bev arm vs 0 in the bev only pts. Gr 3+ non-heme toxicities were higher in pts receiving TRC105+bev vs bev only (50% vs 30.6%, p=0.07), mainly due to headache (6 vs 1 pts) and hyperglycemia (3 vs 0 pts). CECs remained stable in combination treated pts, suggesting a possible impact of the anti-CD105 blockade, but increased at progression in the bev only arm. There was no association between PFS and initial changes in CEC values, independent of treatment received. Conclusions: TRC105 plus bev did not prolong median PFS vs single agent bev in recurrent GBM pts, although it was associated with a non-significant prolongation of OS. These data and associated correlative analysis of CECs from study pts point against endoglin being a clinically significant factor for the development of bev resistance. Support: U10CA180821, U10CA180882. Clinical trial information: NCT01648348.

e13083 Background: CNS involvement from low grade B-cell NHL is rare and has only been reported as case series. The distribution, demographics and outcomes of patients with low grade B-cell CNS NHLs have not been well characterized. Methods: The NCDB represents ~70% of cancer cases in the United States. Using the 2004-2013 NCDB extranodal NHL database, we identified all CNS B-cell NHLs based on ICD-O-3 site and histology codes. Primary or secondary CNS involvement could not be determined. Results: Out of 9435 CNS NHL cases, 475 [5.03%] had low grade histologies. In this group, the median age at diagnosis was 58 years [range 19-89]. Majority of the cases were female [56%], White, non-Hispanic [72%], privately insured [53%], with no comorbidities [74%] and treated in academic/research programs [38%]. Site of CNS disease was not specified in 22%. HIV status was known in 318 cases (6.3% positive). The brain [44%] was the most common site of involvement followed by spinal cord [19%] and meninges [15%]. Follicular lymphoma (FL) [48%] was the most common histology overall followed by marginal zone (MZL) [37%], small lymphocytic (SLL) [8%] and lymphoplasmacytic lymphomas (LPL) [7%]. MZL was the most common histology in the brain [44%] and meninges [61%] while FL was most common in the spinal cord [77%] and nervous system, NOS [69%]. Cranial nerves and eye (retina/optic nerve) involvement was very rare [2 and 1 case each- both MZL]. The overall survival (OS) of CNS B-cell NHL was significantly better if histology was low grade vs other [5-year OS 74% vs 32%, P < 0.0001]. Among CNS low grade B-cell NHLs, 5-year OS varied by histology [MZL 83%, FL 75%, LPL 56% and SLL 50%, P = 0.0003] and site of disease [spinal cord 89%, meninges 78% and brain 63%, P = 0.03] in addition to age at diagnosis and co-morbidities on both uni- and multivariate analysis. Survival was not influenced by sex, race, insurance, year of diagnosis, facility type or location. Conclusions: CNS involvement with low grade B-cell NHL is rare but has a relatively good outcome with most patients surviving beyond 5 years. FL and MZL are the more common low grade histologies. Both histology and disease site are important factors affecting survival.