Search

Background: The European Organisation for Research and Treatment of Cancer (EORTC) 1325/KEYNOTE-054 trial assessed pembrolizumab versus placebo in patients with resected high-risk stage III melanoma. At 15-month median follow-up, pembrolizumab improved recurrence-free survival (hazard ratio [HR] 0·57 [98·4% CI 0·43–0·74], p1 mm), IIIB, or IIIC (without in-transit metastasis), and with an Eastern Cooperative Oncology Group performance status of 0 or 1 were eligible. Patients were randomly assigned (1:1) via a central interactive voice response system to receive intravenous pembrolizumab 200 mg or placebo every 3 weeks for up to 18 doses or until disease recurrence or unacceptable toxicity. Randomisation was stratified according to disease stage and region, using a minimisation technique, and clinical investigators, patients, and those collecting or analysing the data were masked to treatment assignment. The two coprimary endpoints were recurrence-free survival in the intention-to-treat (ITT) population and in patients with PD-L1-positive tumours. The secondary endpoint reported here was distant metastasis-free survival in the ITT and PD-L1-positive populations. This study is registered with ClinicalTrials.gov, NCT02362594, and EudraCT, 2014-004944-37. Findings: Between Aug 26, 2015, and Nov 14, 2016, 1019 patients were assigned to receive either pembrolizumab (n=514) or placebo (n=505). At an overall median follow-up of 42·3 months (IQR 40·5–45·9), 3·5-year distant metastasis-free survival was higher in the pembrolizumab group than in the placebo group in the ITT population (65·3% [95% CI 60·9–69·5] in the pembrolizumab group vs 49·4% [44·8–53·8] in the placebo group; HR 0·60 [95% CI 0·49–0·73]; p

Ocular immune-related adverse events (IrAEs) associated with use of checkpoint inhibitors (CPIs) in cancer therapeutics are relatively rare, occurring in approximately 1% of treated patients. Recognition and early intervention are essential because the degree of tissue damage may be disproportionate to the symptoms, and lack of appropriate treatment risks permanent loss of vision. International guidelines on managing ocular IrAEs provide limited advice only. Importantly, local interventions can be effective and may avoid the need for systemic corticosteroids, thereby permitting the continuation of CPIs. We present a single institution case series of eight affected patients managed by our multidisciplinary team. Consistent with previously published series and case reports, we identified anterior uveitis as the most common ocular IrAE associated with CPIs requiring intervention. Based on our experience, as well as published guidance, we generated a simple algorithm to assist clinicians efficiently manage patients developing ocular symptoms during treatment with CPIs. In addition, we make recommendations for optimising treatment of uveitis and address implications for ongoing CPI therapy.

Background Beneficial treatment options are limited for advanced melanoma patients who progress after or do not respond to immune checkpoint inhibitors (ICI) and targeted therapies. Lifileucel is an adoptive cell therapy using tumor-infiltrating lymphocytes (TIL), that has shown efficacy in patients with advanced melanoma who have progressed on/after anti-PD-1therapy (Sarnaik et al., ASCO 2020). We present the 28-month (mos) follow-up data here. Methods C-144-01 (NCT02360579) is a global Phase 2 open-label, multicenter study of efficacy and safety of lifileucel in patients with unresectable metastatic melanoma who have progressed on anti-PD-1 therapy and BRAF/MEK inhibitors, if BRAF V600 mutant. We report on Cohort 2 (N = 66) patients who have received lifileucel. Tumors were resected at local institutions, processed in central GMP facilities for TIL production, manufactured, cryopreserved and shipped back to sites in a 22-day process. Therapy consisted of one week of nonmyeloablative lymphodepletion using cyclophosphamide (60 mg/kg on Day -7, -6) and fludarabine (25 mg/m2 on Day -5 through -1), a single infusion of lifileucel, and up to six doses of IL-2 doses (600,000 IU/kg/dose). Objective response rate (ORR) was based on RECIST v1.1 as assessed by investigators. Data cutoff was December 14, 2020. Results Baseline characteristics: 3.3 mean prior therapies (100% anti-PD-1; 80% anti-CTLA-4; 23% BRAF/MEK inhibitor), high baseline tumor burden (106 mm mean target lesion sum of diameters), 42% liver/brain lesions, 40.9% had LDH > ULN. Median time from last therapy to tumor harvest was 2.2 mos and 58% of the tumors resected were extra-nodal or non-skin/subcutaneous. ORR by Investigator was 36.4% (3 CR, 21 PR). One patient converted from a PR to CR at 24 mo after lifileucel therapy. Median time to first response was 1.4 mos (range: 1.3-5.6 mos). Median duration of response (mDOR) was still not reached at median follow-up of 28 mos (DOR range: 2.2-35.2 mos). No new safety risks have been identified for lifileucel during the long-term follow-up. Exploratory analyses of product-specific characteristics, including levels of phenotypic markers of T-cell lineage, memory subset, youth, activation/exhaustion, or trafficking did not demonstrate association with response. Conclusions Lifileucel treatment results in a 36.4% ORR and mDOR was not reached at 28 mo of median study follow up in heavily pretreated advanced melanoma patients with high baseline disease burden who progressed on multiple prior therapies, including anti-PD1 and BRAF/MEK inhibitors, if BRAF V600 mutant. Citation Format: Jason Alan Chesney, James M. Larkin, John M. Kirkwood, Jeffrey S. Weber, Nikhil I. Khushalani, Karl Lewis, Theresa M. Medina, Harriet M. Kluger, Sajeve S. Thomas, Evidio Domingo-Musibay, Judit Oláh, Eric D. Whitman, Salvador Martin-Algarra, Philippa G. Corrie, Jose Lutzky, Omid Hamid, Wen Shi, Xiao Wu, Madan Jagasia, Friedrich Graf Finckenstein, Maria Fardis, Amod A. Sarnaik. Lifileucel (LN-144), a cryopreserved autologous tumor infiltrating lymphocyte (TIL) therapy in patients with advanced (unresectable or metastatic) melanoma: durable duration of response at 28 month follow up [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr CT008.

TPS9587 Background: Circulating tumor DNA (ctDNA; the tumour derived fraction of circulating free DNA in the blood) has been shown to be a biomarker of tumor burden/progression in many cancers. We recently accurately monitored treatment response and resistance in stage IV melanoma by ctDNA analysis in serial peripheral blood samples. Pre-clinical data has previously revealed that BRAF inhibition provokes a micro-environment with increased T cell infiltration, improved T cell recognition of melanoma associated antigens and reduced production of immunosuppressive cytokines that could enhance immune responses. We aimed to test the hypothesis that ctDNA could be implemented as a personalised, real-time liquid biopsy to identify when tumours are responding to targeted therapy in order optimise a switch to immunotherapy. Methods: We validated the ctDNA assays for BRAF mutation calling as a primary trial endpoint. We designed a phase II multicenter, parallel arm study across 6 UK sites, to assess primary objectives of i). Whether a ctDNA result can be turned around within 7 days and actioned in a clinically relevant timeframe ii). to assess whether a decrease in ctDNA levels of mutant BRAF by ≥80% from baseline on targeted therapy is an appropriate ‘cut off’ to instruct switching to immunotherapy. Secondary endpoints include Overall Response Rate (ORR) to immunotherapy, radiological/clinical and ctDNA determined progression free survival (PFS) on each treatment. Forty patients are planned based on inclusion criteria of stage IV or stage III unresectable cutaneous BRAF mutant melanoma, baseline ctDNA BRAF variant allele frequency (VAF) ≥1.5%, ECOG 0/1/2, no symptomatic brain metastases, no prior adjuvant nivolumab plus ipilimumab (N+I). Prior adjuvant dabrafinib + trametinib (D+T) is allowed as long as recurrence is >6 months from completion. Patients are randomised 1:1 to either standard Arm A; investigator choice of either D+T (150mg BD +2mg OD respectively) or N+I (1 mg/kg N +3 mg/kg I q3 wkly, then N 480mg q4 wkly) first line, then switch on progression to the other treatment. In the experimental Arm B; all patients start on D+T and have BRAF ctDNA monitored q2 wkly for 4 wks then q4 wkly. When ≥80% decrease vs. baseline in ctDNA BRAF VAF occurs, patients switch to N+I. If patients subsequently progress on N+I, they will resume D+T. The study is open with 9 patients enrolled at time of submission. Clinical trial information: NCT03808441.

e21546 Background: Four independent groups have demonstrated that the pre-treatment gut microbiome of cancer patients impacts the subsequent response to Immune Checkpoint Inhibitor (ICIs) therapy [1-4]. However, the patient’s outcome was linked to different bacteria in each study, which has limited the development of drug response biomarkers and clinic-first design of novel microbiome-based therapeutics. Methods: The Cambridge (UK) MELRESIST study includes a cohort of advanced melanoma patients receiving approved ICIs. Pretreatment stool samples from MELRESIST were analysed by Microbiotica using shotgun metagenomic sequencing. Microbiotica’s platform comprises the world’s leading Reference Genome Database to give the most comprehensive and precise mapping of the gut microbiome. Results: MELRESIST samples showed an overall difference in the microbiome composition between advanced melanoma patients and healthy donors, but not between patients who did or did not respond to ICIs. However, we did identify a discrete microbiome signature that differentiated responders and non-responders with an accuracy of 93%. We extended this signature by reanalysing three published melanoma cohorts [1-3] using the Microbiotica platform, and a propriety bioinformatic model. The resultant bacterial signature was very accurate at predicting response in all 4 published studies combined (91%), and each cohort individually (82-100%). We validated the model using independent validation cohorts and the signature using lung and renal cancer studies [4]. At the core of our microbiome signature was 9 bacteria most significantly associated with ICI efficacy. All 9 were overrepresented in patients who responded to immunotherapy suggesting high abundance of these bacteria is a central driver of ICI response. A consortium comprised of all 9 strains had very potent anti-tumor efficacy in a cancer syngeneic mouse model. The bacteria also demonstrate multiple interactions with primary human immune cells in vitro leading to dendritic cells activation, Cytotoxic T lymphocyte activation and tumor cell killing. These validate the potential of this consortium as a novel therapy for use in combination with ICIs. Conclusions: We have identified a unique microbiome signature predictive of ICI response in 4 independent melanoma cancer cohorts. This removes a major challenge to the field, and could represent a new highly accurate biomarker with clinical application. Nine core bacteria appear to be driving response, and demonstrate anti-tumor activity in vivo and in vitro. This consortium holds great potential as a co-therapy with ICIs. References:1 Matson V et al, Science (2018) 359:104; 2 Gopalakrishnan V et al, Science (2018) 359:97; 3 Frankel AE et al, Neoplasia (2017) 19:848; 4 Routy B et al, Science (2018) 359:91.

1533 Background: At the outset of the COVID-19 pandemic, concerns for the safety of patients receiving anti-cancer treatment coupled with pressures on healthcare services prompted review of standard clinical care pathways in the UK. Revised consensus treatment guidelines were generated. Individual patient-level data regarding actual treatment modifications implemented in clinical practice are lacking. Methods: All anti-cancer treatment plans of patients with breast, lung, renal, hepatopancreatobiliary, CNS cancers and melanoma attending a single academic cancer centre in the UK between 16 March and 31 May 2020 were reviewed and any modifications to standard practice were documented. The effect of patient (age, ECOG performance status [PS], sex) and cancer (site, stage, treatment intent) characteristics on likelihood of treatment modifications were analysed using univariable and multivariable models. Results: Treatment plans for 925 patients were reviewed: median patient age was 63 (range 19-97); 66% were female; 73% were PS 0-1; 45% were on a curative pathway. Overall, 47% of all patients had one or more modifications made to their treatment plans: 53% of surgeries (primarily being delayed); 41% of radiotherapy (primarily reduced fractions delivered); 39% of systemic therapy prescriptions. 96-100% of all systemic therapy modifications resulted in treatment de-escalation, excluding endocrine therapy used as a bridge to defer primary breast cancer surgery. Biological therapy was predominantly interrupted (49%), immunotherapy was mostly omitted entirely (36%), and chemotherapy varied between interruptions (39%) or omissions (31%). Relative to the likelihood of modification to chemotherapy, surgery was significantly more likely to be modified (OR 1.69 95%CI 1.20-2.38). Chemotherapy, radiotherapy, biological therapy and immunotherapy were all modified to a similar degree. Multivariate analysis identified PS ≥2 (OR 1.79, 95% CI 1.18–2.75), but not patient age, as a predictor of treatment modification. Some tumour types were less likely to undergo any modification: stage 1-3 lung (OR 0.13, 95%CI 0.04-0.37), stage 4 lung (OR 0.26 95%CI 0.24–0.60) and stage 4 renal cancer (OR 0.22 95%CI 0.09-0.52). Conclusions: This single centre analysis demonstrated almost half of cancer patients had their treatment modified, the overwhelming majority resulting in treatment de-escalation. The impact of the treatment modifications on overall cancer patient outcomes remains to be determined.

9505 Background: Immune checkpoint inhibitors (ICI) have become standard of care for treatment of metastatic melanoma. Most patients with advanced melanoma progress on ICI and treatment options are limited for these patients. Progression may be through primary resistance (lack of response) or secondary resistance (initial response then progression). Lifileucel is an adoptive cell therapy using TIL, that has shown efficacy in patients with advanced melanoma who progress on/after an anti-PD-1 (Sarnaik, 2020). We present the 28-month (mos) follow-up data and highlight the impact of prior anti-PD-1 response and duration of exposure on outcome with lifileucel. Methods: C-144-01 is a Phase 2, open-label, multicenter study of efficacy and safety of lifileucel in patients with advanced melanoma who have progressed on anti-PD-1 therapy and BRAFi ± MEKi, if BRAF V600+. We report long-term follow up on Cohort 2 (N = 66). Tumors were resected at local sites, processed in central GMP facilities for TIL production in a 22-day manufacturing process. Therapy consisted of nonmyeloablative lymphodepletion using 2 days of cyclophosphamide and 5 days of fludarabine, a single infusion of lifileucel, and up to six doses of IL-2. Objective response rate (ORR) was assessed by RECIST 1.1. Data cutoff was Dec. 14, 2020. Results: Baseline characteristics: 3.3 mean prior therapies (100% anti-PD-1; 80% anti-CTLA-4; 23% BRAFi/MEKi), high baseline tumor burden (106 mm mean target lesion SOD), 42% liver/brain lesions, 40.9% LDH > ULN. ORR by investigator was 36.4% (3 CR, one new CR developed at 24 mos; 21 PR). Median duration of response (mDOR) was not reached at median follow-up of 28 mos (DOR range: 2.2- 35.2 mos). In responders, the median cumulative duration and median prior lines of anti-PD-1 therapy was 4.4 mos (range: 1.4-22.5 mos), and 1.5 (range: 1-4). Data in Table demonstrates a meaningful increase in DOR to TIL with primary anti-PD-1 resistance and lower duration of time on prior anti-PD-1 therapy. No new safety risks have been identified for lifileucel during long-term follow-up. Conclusions: One-time lifileucel treatment results in a 36.4% ORR, and mDOR was not reached at 28 mos of median study follow up. One PR converted to a new CR at 24 months as responses continue to deepen. DOR is positively associated with primary resistance to prior anti-PD-1 therapy and with shorter cumulative prior duration of anti-PD-1 therapy. Lifileucel may offer a better clinical outcome when used earlier upon detection of progression on prior anti-PD-1 rather than retreatment with anti-PD-1 based regimens. Clinical trial information: NCT02360579. [Table: see text]

10006 Background: Treatment options are limited for patients with advanced melanoma who have progressed on checkpoint inhibitors and targeted therapies. Adoptive cell therapy using tumor-infiltrating lymphocytes (TIL) leverages and enhances the body’s natural defense against cancer and has shown durable responses in heavily pretreated melanoma patients. Methods: C-144-01 is a global Phase 2 open-label, multicenter study of efficacy & safety of lifileucel in patients with unresectable metastatic melanoma who have progressed on checkpoint inhibitors and BRAF/MEK inhibitors, if BRAFv600 mutant. We report on Cohort 2 (N = 66) patients who have received TIL. Tumors were resected at local institutions, processed in central GMP facilities for TIL production, manufactured, cryopreserved & shipped back to sites in a 22-day process. Therapy consisted of one week of lymphodepletion, a single lifileucel infusion, and up to 6 IL-2 doses. ORR was based on RECIST v1.1 by investigator assessment. Data cutoff was Feb 2, 2020. Results: Baseline characteristics: 3.3 mean prior therapies (anti-PD1 100%; anti-CTLA-4 80%; BRAF/MEK inhibitor 23%), high baseline tumor burden (106 mm mean target lesion sum of diameters), 44% liver/brain lesions, 40.9% LDH > ULN. ORR by investigator was 36.4% (2 CR, 22 PR) and DCR was 80.3%. Mean time to response was 1.9 months (range: 1.3-5.6). After a median study follow-up of 17.0 months, median DOR (mDOR) was still not reached. Six responders have progressed, 2 have died and 2 started other anti-cancer therapy without progression. The adverse event profile was consistent with the underlying advanced disease and the lymphodepletion and IL-2 regimens. Additional follow-up data will be available for presentation. Conclusions: Lifileucel treatment results in a 36.4% ORR and mDOR was not reached at 17.0 months of median study follow up in a heavily pretreated metastatic melanoma patients with high baseline disease burden who progressed on multiple prior therapies, including anti-PD1 and BRAF/MEK inhibitors, if BRAFv600 mutant. Clinical trial information: NCT02360579.

55 Background: Population-based research provides real world information about prescribing trends, as well as benefits and harms associated with new interventions first evaluated in randomized clinical trials (RCTs). The national health service in England has comprehensive databases which we used to track prescribing of checkpoint inhibitors (CPIs) approved for 1st line use in metastatic melanoma and determine patient outcomes associated with survival and treatment toxicity. Methods: National Cancer Registration and Analysis Services (NCRAS) datasets were used to identify metastatic melanoma patients (pts) prescribed 1st line CPIs and calculate survival outcomes. The HES (Hospital Episode Statistics) database was used to identify emergency attendances and emergency hospital admissions. Results: 5465 melanoma pts were registered between April ‘14 and March ‘18. 2322 pts received 1st line CPIs: 1174 pembrolizumab (pem), 724 ipilimumab (ipi), 52 nivolumab (nivo), 372 ipinivo. Prescribing reflected national access, with ipi starting in July ‘14 and dropping rapidly on approval of pem in Nov ’15. Ipinivo prescriptions increased annually since July ’16. There was good 3yr OS concordance with RCT outcomes for ipinivo (56%), nivo (51%), pembro (40%) and ipi (32%). Pts treated with ipinivo were significantly younger (88% vs 49%

9523 Background: ImmTAC molecules are unique TCR–anti-CD3 bispecifics that redirect T cells against intracellular antigens. IMCgp100, an ImmTAC targeting melanocyte-expressed gp100 antigen, has demonstrated monotherapy activity in advanced melanoma and can cause rash and cytokine-mediated AEs, hypothesized to be on-target (gp100) or effector (CD3) mediated. A preclinical MoA for T cell bispecifics suggests chemokine CXCL10 redirection of CXCR3+ T cells from blood into antigen-positive tissues; this has not been clinically validated. Methods: 84 HLA-A2+ pts with advanced melanoma (n = 61 cutaneous [CM], n = 19 uveal [UM], n = 4 other) received IMCgp100 (NCT01211262). Serum (n = 40) and PBMC (n = 22) samples were taken pre- and post-infusion to analyze changes in cytokines and circulating T cells. Pre- (n = 16) and post-treatment (n = 11) tumor biopsies were analyzed by IHC for CD3, PD-L1 and gp100 expression; tumor RNA (n = 12) was analyzed for gene expression. Results: IMCgp100 induced a transient increase in IFNg-inducible cytokines, most prominently CXCL10. A greater increase in serum CXCL10 was associated with longer OS (p = 0.0002), tumor shrinkage (p = 0.003), and greater transient reduction in peripheral CXCR3+CD8+ T cells (p = 0.001). Reduction in CXCR3+ CD8+ T cells also trended with longer OS (p = 0.02), and tumor shrinkage (p = 0.03). 3/16 pre-treatment biopsies had < 1% gp100 expression (all progressive disease). 8/11 biopsies post-IMCgp100 had increased CD3+ T cells compared with matched pre-treatment samples (associated with baseline gp100 but not PD-L1 expression). Based on tumor biopsy gene expression analysis, IMCgp100 increased T cell markers, IFNg-inducible and cytotoxicity-related genes. Conclusions: The association of clinical benefit with increased serum CXCL10 and decreased peripheral CXCR3+ T cells supports the MoA of IMCgp100-induced T cell redirection and activation. Tumor biopsy results support IMCgp100 redirection of T cells to antigen-positive tumor. A Phase II trial in CM (NCT02535078), a Phase I/II trial in UM (NCT02570308), and a Pivotal RCT in UM (NCT03070392) are ongoing. Clinical trial information: NCT01211262.

Systemic therapy is one of a number of treatment options routinely used in the management of advanced, unresectable neuroendocrine tumours (NETs). In contrast to many of the other NET treatment modalities, there is at least some evidence base to justify its use. Even so, well-designed clinical trials are limited, since conducting clinical research in this complex group of rare cancers is challenging. The remit of this review article is to summarise the oncology literature and explain the role of systemic therapy in treating NETs of gastroenteropancreatic origin, identifying benefits and limitations. The molecular biology of NETs is now being unravelled, which affords new opportunities for development of mechanism-driven therapies. The rationale for some of the newer systemic targeted therapies that are showing promise in the clinic is discussed.

9501 Background: Bevacizumab (Bev) is a recombinant humanised monoclonal antibody to vascular endothelial growth factor (VEGF) shown to improve survival in several advanced solid tumours. As VEGF is a relevant target in melanoma, AVAST-M aimed to evaluate the role of Bev in melanoma patients at high risk of recurrence. Methods: AVAST-M (ISRCTN81261306) is a randomised phase III trial evaluating single agent Bev (7.5mg/kg IV 3 weekly for 1 year) as adjuvant therapy after resection of AJCC stage IIB, IIC and III cutaneous melanoma compared to standard observation (Obs). 1320 patients were needed to detect 8% differences in 5 year overall survival (OS) rate from 40% to 48%; 85% power, 5% alpha level. Primary endpoint was OS; secondary endpoints included disease free interval (DFI), distant-metastasis free interval (DMFI). BRAF and NRASmutation status were obtained in 682 patients in a translational sub-study. Results: From July 2007 to March 2012, 1343 patients were recruited (671 to Bev; 672 to Obs). 56% were male, median age was 56 years (range 18-88 years), 14% were stage IIIA and 59% were stage IIIB/C. With 6 years median follow-up, 505 (38%) patients had died (251 [37%] on Bev; 254 [38%] on Obs); 699 (52%) patients had recurred (335 [50%] on Bev, 369 [55%] on Obs). OS at 5 years was 64% on Bev versus 63% on Obs (Hazard ratio [HR] 0.99; 95% confidence interval [CI] 0.84-1.18, p=0.96). At 5 years, 51% were disease free on Bev versus 45% on Obs (HR 0.85; 95% CI 0.74-0.99, p=0.04) and 59% were distant metastasis free on Bev versus 54% on Obs (HR 0.91; 95% CI 0.77-1.07, p=0.24). A BRAF V600 mutation was found in 44% of tumours assessed; 20% were NRAS mutant. BRAF mutant patients treated with Bev tended to have better DFI (HR=0.79 95% CI 0.58-1.08, p=0.14) and OS (HR=0.79; 95% CI 0.55-1.13, p=0.20); this was not evident for BRAF WT. NRAS mutant patients tended to have worse DFI (HR=1.39; 95% CI 1.03-1.88; p=0.03) and OS (HR=1.18; 95% CI 0.85-1.62, p=0.20) than NRASWT patients. Conclusions: This large, multi-centre trial of melanoma patients at high risk of recurrence has shown that adjuvant Bev improves DFI, but this does not translate into an overall survival benefit. Funding:Cancer Research UK; drug supplied by Roche Products Ltd. Clinical trial information: ISRCTN81261306.

9583 Background: Patients with high-risk stage II/III resected melanoma commonly develop metastatic disease. Adjuvant high dose interferon and ipilimumab 10mg/kg are associated with survival benefit, but at the expense of toxicity. At present, we cannot differentiate between patients who will progress to stage IV disease or those cured by surgery. Circulating tumor DNA (ctDNA) is a biomarker of disease progression in many cancers including stage IV melanoma, however its utility in the adjuvant setting has not been investigated. Methods: We performed droplet digital polymerase chain reaction to detect BRAF and NRAS mutations in plasma of 161 stage IIB, IIC/III high-risk melanoma patients enrolled in AVAST-M; a randomized study of bevacizumab vs. observation. Baseline plasma samples were collected within 12 weeks of surgery for both arms. Results: Mutant BRAF or NRASctDNA was detectable ( > 1 copy of mutant ctDNA) in 12% (19/161) of patients. Patients with detectable ctDNA had a decreased disease free interval (DFI; hazard ratio [HR] 3.12; 95% confidence interval [CI] 1.79-5.47; p < 0.0001) and distant metastasis-free interval (DMFI; HR 3.22; 95% CI 1.80-5.79; p < 0.0001) vs. those with undetectable ctDNA. One year DFI rate for patients with detectable ctDNA was 26% (95% CI 10-47%) vs. 74% (95% CI 66-81%) for those with undetectable ctDNA. One year DMFI rate for patients with detectable ctDNA was 37% (95% CI 17-57%) vs. 84% (95% CI 77-89%) with undetectable ctDNA. Detectable ctDNA remained a significant predictor after adjustment for N classification and performance status (PS) (DFI HR 3.48; 95% CI 1.95-6.19, p < 0.0001; DMFI HR 3.41, 95% CI 1.88-6.20, p < 0.0001). Five year overall survival (OS) rate for patients with detectable ctDNA was 33% (95% CI 14-55%) vs. 65% (95% CI 56-72%) for those with undetectable ctDNA. OS was significantly worse for patients with detectable ctDNA (HR 2.63; 95% CI 1.4-4.96); p = 0.003) and remained significant after adjustment for PS (HR 2.5, 95% CI 1.32-4.74, p = 0.005). Conclusions: CtDNA is predictive of relapse and survival in high-risk, resected melanoma and can potentially aid stratification of patients for adjuvant therapy.

e21032 Background: Circulating tumour DNA (ctDNA) is released by cancer cells into the bloodstream and can be analysed via liquid biopsy, providing a real-time snapshot of tumour burden. After treatment, ctDNA concentrations may be low, making detection challenging, and collecting larger sample volumes may be impractical. Our study aims to achieve high-sensitivity monitoring of melanoma patients melanoma on therapy, maximising the number of mutations targeted per patient by individualised sequencing. Methods: 72 patients with stage 3 or 4 melanoma have so far been recruited to MelResist, a translational research study. 235 plasma, urine, tumour and buffy coat samples have been analysed for 10 BRAF mutant metastatic melanoma patients receiving systemic therapies, and 30 matched CT scans were analysed for RECIST response. Exome or targeted sequencing were carried out on tumour samples and plasma at baseline and progression and mutations identified were used to design individualised amplicon and hybrid-capture sequencing panels targeting hundreds to thousands of mutations per patient. Results: Baseline ctDNA allele fraction predicted for overall survival ( r = -0.56, p < 0.05). Longitudinal analysis of ctDNA showed that mutant allele fractions significantly correlated with tumour burden from CT imaging ( r = 0.79, p < 0.0001) and ctDNA changes were concordant with 20/22 (90%) of RECIST response events. ctDNA allele fractions increased with a lead time of 70 days relative to serum lactate dehydrogenase, a standard measure of disease burden (IQR = 42-142 days). Targeting multiple mutations per patient enabled detection of less than one mutant genome copy per sample and allowed comprehensive monitoring of clonal evolution on therapy, even using limited sample volumes. Conclusions: Analysis of an initial cohort of BRAF mutant metastatic melanoma patients has confirmed feasibility to apply an individualised targeted sequencing panel on both plasma and urine DNA. For a given sample volume, monitoring multiple mutations improves detection sensitivity compared to targeting individual loci, has predictive value and can be applied to all melanoma patients, irrespective of BRAF mutation status.

4006 Background: Despite improvements in multidisciplinary management, BTC has a poor outcome. Approximately 20% of cases are suitable for surgical resection with a 5 year survival of < 10%. BILCAP aimed to determine whether capecitabine (Cape) improves overall survival (OS) compared to observation (Obs) following radical surgery. Methods: Patients with completely-resected cholangiocarcinoma (CCA) or gallbladder cancer (including liver and pancreatic resection, as appropriate), with adequate biliary drainage, no ongoing infection, adequate renal, haematological and liver function, and ECOG PS ≤2, were randomized 1:1 to Cape (1250 mg/m2 D1-14 every 21 days, for 8 cycles) or Obs. Randomization was minimized on tumor site, resection status, ECOG PS and surgical center. The primary outcome was OS in the intention to treat (ITT) population. 410 patients were needed to detect a hazard ratio (HR) of 0.69 (2-sided α = 0.05 and 80% power). HR was estimated by Cox survival model with adjustment for the minimization factors. Primary analysis performed with at least 24 months (m) follow-up. Results: 447 participants were randomized to Cape (n = 223) or Obs (n = 224) from 44 UK sites between 2006-2014. Median age was 63y (IQR 55, 69) and 201 (45%), 232 (52%), and 14 (3%) patients were ECOG PS 0, 1 and 2 respectively. Primary site: 84 (19%) intrahepatic, 128 (28%) hilar, 156 (35%) extrahepatic CCA and 79 (18%) muscle-invasive gallbladder cancers. Resection margins: R0 in 279 (62%) and R1 in 168 (38%); 207 (46%) were node-negative. Follow up was at least 36m in > 80% of surviving patients. By ITT analysis (n = 447), median OS was 51m (95%CI 35, 59) for Cape and 36m (95%CI 30, 45) for Obs, HR 0.80 (95%CI 0.63, 1.04; p = 0.097). Sensitivity analyses with adjustment for nodal status, grade of disease and gender indicated HR 0.71 (95%CI 0.55, 0.92 p < 0.01). In the per-protocol analysis (Cape n = 210, Obs n = 220) median OS was 53m (95%CI 40, NR) for Cape and 36m (95%CI 30, 44) for Obs, HR 0.75 (95%CI 0.58, 0.97; p = 0.028). Median RFS (ITT) was 25m (95%CI 19, 37) for Cape and 18m (95%CI 13, 28) for Obs. Grade 3-4 toxicity was less than anticipated. Conclusions: Cape improves OS in BTC when used as adjuvant and should become standard of care. Clinical trial information: ISRCTN72785446.

4100 Background: NabP+GEM chemotherapy improves survival compared with GEM monotherapy as treatment for mPDAC. A PDAC mouse model suggested that nabP potentiates GEM activity by reducing cytidine deaminase levels and scheduling may be critical to optimise clinical benefit. Methods: Patients (pts) were randomised to receive standard concomitant (CON) nabP+GEM or sequential (SEQ) administration, with nabP given 24 hours before GEM. After 6 cycles, pts benefiting from treatment could continue the same regimen until disease progression. The primary endpoint was progression-free survival (PFS) by RECIST v1.1; secondary endpoints included safety, objective response rate (ORR), overall survival (OS) and quality of life (QoL). Serial blood and baseline tumour samples were collected for exploratory biomarkers. Results: Between March 2014 and 2016, 146 pts (71 SEQ, 75 CON) were recruited. Median age (range) was 66 (45-82) years; Karnofsky performance status was 70 (in 12% pts), 80 (27%), 90 (38%) or 100 (24%); 47% had pancreatic head primaries; 84% had liver metastases. Median no. cycles received was 4 SEQ, 3 CON; 51 pts (35%) received ≥6 cycles of treatment (42% SEQ, 28% CON). A 24+2hr interval was achieved in > 90% SEQ admin. Grade ≥3 adverse events experienced by ≥10% pts (SEQ, CON) were neutropaenia (54%, 30%; p = 0.003), febrile neutropaenia (12%, 12%), fatigue (22%, 15%), vomiting (7%, 11%) and anaemia (10%, 5%). G-CSF was administered at local investigator's discretion to 35 pts (23 SEQ, 12 CON; p = 0.015). To date, 112 pts have died. 6 month (m) PFS by SEQ and CON arms were 47% and 33%; median PFS were 5.8 and 4.0m; hazard ratio (HR) = 0.66, 95% CI = 0.46-0.95; 12m OS by SEQ and CON arms were 29% and 26%; median OS were 10.1 and 7.9m; HR = 0.88, 95% CI = 0.61-1.29. ORR was 50% SEQ and 33% CON (p = 0.065). Mean baseline QoL Global health status score was 60.6 SEQ and 63.4 CON. The mean change in QoL score from baseline at 24 weeks was -2.1 SEQ and -12.1 CON. Conclusions: Sequential delivery of nabP combined with GEM trended towards improving all clinically relevant efficacy end points: PFS, OS, and ORR. Translational correlates will be reported in due course. Clinical trial information: ISRCTN71070888.

342 Background: NabP+GEM chemotherapy improves survival compared with standard GEM monotherapy as treatment for advanced PDAC. A PDAC mouse model showed that delivery of nabP prior to GEM generated a higher intratumoural dFdC:dFdU ratio, which correlated with improved outcome. Therefore, scheduling may be critical to optimise clinical benefit. Methods: Patients were randomised to receive nabP+GEM, either concomitantly (CON) or sequentially (SEQ), 24 hours apart. After 6 cycles, patients benefiting from treatment could continue the same regimen until disease progression. The primary endpoint was progression-free survival (PFS) assessed by 8-weekly CT scanning; secondary endpoints included safety, response rate, overall survival (OS) and quality of life (QoL). Results: Between March 2014 and 2016, 146 patients (71 SEQ, 75 CON) from 19 UK sites were randomly assigned. Median age (range) was 66 (45-82) years; Karnofsky performance status was 70 (in 12% of patients), 80 (27%), 90 (38%) or 100 (24%); 47% of primary tumours were situated in the pancreatic head; 84% had liver metastases. Median number of cycles was 4 SEQ, 3 CON; 50 patients (34%) received ≥ 6 cycles of treatment (41% SEQ, 28% CON). A 24+2hr interval was achieved in over 90% SEQ administrations. Most patients stopped treatment prior to 6 cycles due to disease progression/death (42%) or toxicity (41%). Grade ≥ 3 adverse events experienced by ≥ 10% patients (SEQ, CON) were neutropaenia (54%, 30%), febrile neutropaenia (12%, 12%), fatigue (22%, 16%), vomiting (7%, 11%) and anaemia (10%, 5%). G-CSF was administered to 35 patients (23 SEQ, 12 CON). To date, 109 patients have died. 6-month PFS by SEQ and CON arms was 46.7% and 33.8%; median PFS was 5.8 and 4.3 months. 12-month OS by SEQ and CON arms was 29.1% and 20.1%; median OS was 10.1 and 7.9 months. The objective response rate was 44% SEQ and 30% CON. Mean baseline QoL Global health status was 60.6 SEQ and 63.4 CON. The change from baseline (mean) at 24 weeks was -2.1 SEQ and -12.1 CON. Conclusions: Sequential delivery of nabP and GEM improves progression-free and overall survival. Although more myelosuppressive, QoL was maintained. Clinical trial information: ISRCTN71070888.

3016Background: T cell-based bispecific agents have shown activity in hematologic cancers, but solid tumor efficacy remains elusive. IMCgp100 is a bispecific biologic comprising an affinityenhanced...

9531Background: IMAGE (ipilimumab: management of advanced melanoma in real practice) is an ongoing multinational, prospective, observational study that assesses the treatment of advanced (unresecta...

e21027Background: Combination BRAF and MEK inhibition in mm has demonstrated improved Overall Survival (OS) and Progression Free Survival (PFS) compared to single agent BRAF inhibition (BRAFi) in p...

4002 Background: ABCs exhibit increased VEGF expression, correlating with metastatic disease. Combining cediranib (which inhibits VEGFR1, 2 and 3 tyrosine kinase and VEGF-induced signaling in endot...

LBA9000 Background: Bevacizumab (Bev) is a recombinant humanized monoclonal antibody to vascular endothelial growth factor (VEGF) shown to improve survival in several advanced solid tumors. Multiple adjuvant trials are underway, but trials that have reported in colon and triple-negative breast cancer did not meet their primary end points. Since VEGF is a relevant target in melanoma, AVAST-M aimed to evaluate the role of Bev in patients (pts) with resected melanoma at high risk of recurrence. Methods: AVAST-M is a randomized phase III trial evaluating single agent Bev (7.5mg/kg IV 3 weekly for 1 year) as adjuvant therapy following resection of AJCC stage IIB, IIC, and III cutaneous melanoma compared to standard observation (Obs). 1,320 pts were required to detect 8% differences in 5-year overall survival (OS) rate from 40% to 48%; 85% power, 5% alpha level. Primary endpoint is OS; secondary endpoints are disease free interval (DFI), distant-metastasis free interval (DMFI), safety, and quality of life (QoL). An associated translational study is ongoing. Results of the first pre-planned interim analysis (agreed by the IDSMC) are reported here. Results: Between July 2007 and March 2012, 1,343 pts were recruited. 56% were male; median age 56 years (range 18-88 years), 16% were stage IIB, 11% IIC, 15% IIIA, 36% IIIB, 20% IIIC, and 2% unknown stage. Ulceration status of the primary melanoma was: 38% present, 45% absent, 17% unknown. At the time of the interim analysis, 286 (21%) patients had died. Median follow-up for survival was 25 months. Median duration of Bev treatment in 671 treated pts was 51 weeks (dose intensity 86%). Main outcomes are shown in the table. Grade 3/4 adverse events were experienced in 101 (15%) Bev pts and 36 (5%) Obs pts. Conclusions: Interim analysis of this large, multicenter trial of melanoma patients at high risk of recurrence has shown that adjuvant Bev monotherapy is well tolerated and improved DFI. Longer follow-up is required to determine an impact on the primary endpoint of 5-year OS. Clinical trial information: 81261306. [Table: see text]

3018 Background: Since publication of the registration trial in 2010 (Hodi et al, NEJM 2010;363:711-23), real world use of ipilimumab (Ipi) in previously treated advanced melanoma patients has extended beyond the specific trial entry criteria of ECOG PS 0-1. We undertook a review of UK patients (pts) treated in the international EAP prior to European licensing of Ipi in August 2011, to compare real world survival outcomes. Methods: UK clinicians registered in the EAP provided anonymised data using pre-specified variable fields for all pts. The EAP stipulated pts should have previously treated, unresectable stage III or IV metastatic melanoma and receive Ipi 3 mg/kg, 3 weekly IV, for up to 4 cycles. Response using RECIST criteria was assessed 12 weekly. Grade ≥3 adverse events (AEs) using CTCAE v3.0 were collected. Results: To date, information on 162 pts has been received from 16 UK sites. Primary sites were: 78% cutaneous, 4% ocular, 1% mucosal, 17% unknown. 78% pts had M1c disease, 14% had brain metastases. No prior therapies ranged from 0-4, 72% pts received 1 prior therapy. Median age was 60 years, men>women (1.6:1). ECOG PS was: 38% 0, 47% 1, 14% 2, 1% 3. BRAF status was known in 38% cases and WT in 75% of these. 19% pts were on steroids at baseline. No cycles delivered was 4 in 52%, 3 in 13%, 2 in 16%, 1 in 17% pts. Most frequent reason for stopping early was clinical evidence of disease progression (71%), death (16%) or unacceptable AE (12%). 32% pts experienced a grade ≥3 AE, the most common being diarrhoea (13%) and fatigue (8%). Complete and partial responses were reported in 1% and 21% of treated pts. At median follow-up of 17 months, median progression free survival and overall survival (OS) were 2.8 and 5.7 months, 1 year OS was 30%. Comparing outcomes of various pt subgroups, the strongest prognostic factor for OS was ECOG PS at the start of treatment (p

LBA9000 The full, final text of this abstract will be available at abstract.asco.org at 7:30 AM (EDT) on Saturday, June, 1, 2013, and in the Annual Meeting Proceedings online supplement to the June 20, 2013, issue of Journal of Clinical Oncology. Onsite at the Meeting, this abstract will be printed in the Saturday edition of ASCO Daily News.

LBA4004 Background: GV1001, a promiscuous class II epitope encompassing aa 611-626 of hTERT led to the development of CD4+ clones recognizing hTERT in patients with advanced pancreatic cancer (APC). Preclinically gemcitabine increases antigen cross-presentation, enhances T cell trafficking/activation, and reduces MDSCs and Tregs. Methods: Patients with APC were randomized 1:1:1 to: Arm 1 GemCap; 2 GemCap for 8/52 followed by GV1001 followed by further GemCap if no PD at week 8; 3 concurrent administration of GemCap and GV1001. 735 (69.2%) had metastatic disease and 948 (89.3%) had ECOG PS=0 or 1. Randomization was stratified by stage and PS. Primary endpoint was overall survival (OS); secondary endpoints included ORR, TTP, and AEs. Recruitment target was 1,110 patients (780 deaths) to permit detection of a hazard ratio of 0.748 between either GV1001 arm and Arm 1 using a 2-sided α=0.025 level of significance with at least 80% power. Results: 1,062 pts from 51 centers were randomized. Trial maturity was high (72.7% patients died): median follow-up was 6.11 months. The overall response rates were Arm 1=17.6%; Arm 2=8.9% (p=0.001); Arm 3: 15.5% (p=0.460 compared with Arm 1). Conclusions: OS with concurrent GemCap/GV1001 was not different to that with GemCap alone. OS with sequential GV1001 was not statistically different to GemCap alone as it did not meet the criterion for statistical significance (p

Background: ABC-03 was a randomized phase II trial of cediranib (C; AZD2171) [inhibitor of VEGFR-1, -2 and -3 tyrosine kinase and VEGF-induced signaling] or placebo (P) in combination with cisplatin/gemcitabine for pts with ABC. Changes in plasma biomarkers including angiogenic proteins and a marker of cell death (CK18) have not been reported in ABC.Methods: ELISA was performed for 15 angiogenesis or inflammatory-related proteins and CK18 in plasma of pts with repeated sampling during treatment and at disease progression. Tumor markers were also evaluated. Overall survival (OS) was analyzed using Cox model for baseline (BL) biomarkers and time-varying covariate Cox model (TVC) for biomarkers with serial measurements.Results: Plasma samples were available from 117 of 124 pts (59 in C arm, 58 in P). After adjusting for treatment and pt characteristics, higher BL VEGFR2 (p = 0.02), CK18 (p = 0.005), CEA (p < 0.001), CA19-9 (p = 0.03) and CA125 (p = 0.001) correlated with shorter OS, as did incremental increases in CK18 (HR 1.08, 95%CI 1.05-1.11, p < 0.001) and IL-6 (HR 1.02, 95%CI 1.00-1.03, p = 0.008) using TVC. There was evidence for interaction between treatment and BL CA19-9 and PDGFbb for OS: incremental increases in CA19-9 were associated with an increased risk of death (C: HR 1.44, 95%CI 1.16-1.80; P: HR 1.03, 95%CI 1.00-1.05, p = 0.001) and higher values of PDGFbb were associated with a decreased risk of death with C (HR 0.82, 95%CI 0.68-1.00) and an increased risk of death with P (HR 1.15, 95%CI 0.99-1.34, p = 0.01). In TVC, there was evidence of interaction between treatment and VEGFA (p = 0.03), PDGFbb (p = 0.08), VEGFC (p = 0.06) and SDF1b (p = 0.06). Higher values of VEGFA were associated with increased risk of death with P (HR 1.31, 95%CI 1.06-1.62) but not with C (HR 0.99, 95%CI 0.86-1.15) (p = 0.03).Conclusions: High BL VEGFR2, CK18 and tumor markers are associated with a poor outcome in pts with ABC. Increasing levels of CA19-9 and PDGFbb from BL were associated with increased and decreased risk of death, respectively, in C-treated pts. Changes in VEGFA correlated with OS for pts on P but not C. These findings may inform future clinical trial stratification and design. Clinical trial information: NCT00939848