Yao, James C.

142 Background: Fruquintinib (F) is a highly selective, oral tyrosine kinase inhibitor of all 3 vascular endothelial growth factor receptors (VEGFRs -1, -2, -3) that is approved by the US FDA and in the EU, UK, and Japan for previously treated metastatic colorectal cancer (mCRC), regardless of biomarker status. FRESCO-2 (NCT04322539) met its primary endpoint demonstrating significantly improved overall survival (OS) with F + best supportive care (BSC) vs placebo (P) + BSC. This subgroup analysis assessed landmark survival of all FRESCO-2 patients (pts), and baseline (BL) characteristics and safety data of pts treated with F who had OS ≥10 months (mos) compared with the intent-to-treat (ITT) population (pop). Methods: Pts were randomized 2:1 to receive F 5 mg or matching P, by mouth, once daily, 3 weeks on, 1 week off, + BSC. Pts had received prior chemotherapy, anti-VEGF therapy and, if RAS wild type, anti-EGFR therapy; and had prior exposure to trifluridine/tipiracil (TAS-102) and/or regorafenib. The hazard ratio (HR) between the two treatment arms was calculated from a stratified Cox proportional hazard model with treatment group as the only covariate in the model. Results: In the ITT pop, OS was improved with F vs P (HR 0.66); OS rates (95% confidence intervals [CIs]) at 6 and 9 mos were 60.4% (55.9–64.9) vs 41.5% (35.0–48.0) and 41.1% (36.4–45.8) vs 28.2% (22.1–34.3), respectively. Progression-free survival (PFS) rates (95% CIs) at 6 and 9 mos were 23.8% (19.7–28.0) vs 1.1% (0–2.6) and 11.3% (8.1–14.6) vs 0.5% (0–1.6), respectively. Of 461 pts who received F, 113 (24.5%) had OS ≥10 mos (range 10.0–18.9) with a median duration of F treatment of 6.3 mos (range 0.7–19.1). Among pts receiving F, a higher proportion (≥10% difference) of pts with OS ≥10 mos had no liver metastases (mets) and an Eastern Cooperative Oncology Group performance status (ECOG PS) of 0 at BL compared with the ITT pop (Table). In pts with OS ≥10 mos, 61.9% experienced a grade ≥3 treatment-emergent adverse event compared with 62.7% in the ITT pop. Conclusions: Longer-term OS and PFS rates at 6 and 9 mos were higher with F vs P in the FRESCO-2 ITT pop. The data show a higher proportion of pts with OS ≥10 mos had an absence of liver mets and ECOG PS 0 at BL than the ITT pop. The safety profile of F in pts with OS ≥10 mos was consistent with the ITT pop. Clinical trial information: NCT04322539 . BL characteristics of pts treated with F with OS ≥10 mos vs ITT pop. BL characteristic Pts with OS ≥10 mos (n=113) ITT pop(n=461) ECOG PS (0 / 1), % 54.0 / 46.0 42.5 / 57.5 Median time since 1 st CRC diagnosis, mos 52.0 47.2 Primary tumor location at 1 st diagnosis (colon / rectum / both), % 51.3 / 35.4 / 13.3 60.5 / 31.0 / 8.5 Primary site at 1 st diagnosis (colon left / colon right), % 42.5 / 15.9 41.6 / 21.0 Liver mets, % 58.4 73.5 Median duration of mCRC, mos 42.7 37.9 Prior lines of therapy for mCRC (≤3 / >3), % 26.5 / 73.5 27.1 / 72.9 Prior TAS-102, % 46.9 52.1 Prior regorafenib, % 12.4 8.7

3579 Background: Fruquintinib (F) is a highly selective, oral tyrosine kinase inhibitor of all 3 VEGF receptors. Based on FRESCO (NCT02314819) and FRESCO-2 (NCT04322539) data, F was approved by the US FDA for previously treatedmetastatic colorectal cancer (mCRC). Efficacy and safety of F + best supportive care (BSC) vs placebo (P) + BSC were analyzed according to prior treatment (tx) sequence using FRESCO and FRESCO-2 data. Methods: In FRESCO and FRESCO-2, patients (pts) were randomized 2:1 to receive F 5 mg or matching P, PO daily x 21 days every 28 days. In FRESCO, eligible pts had received ≥2 prior chemotherapy regimens and were permitted to have received prior anti-VEGF and/or anti-EGFR therapies, whereas in FRESCO-2, in addition to standard chemotherapies, anti-VEGF, and anti-EGFR therapies if indicated, prior exposure to trifluridine/tipiracil (TAS) and/or regorafenib (rego) was required. FRESCO-2 pts must have also received a BRAF inhibitor and/or immunotherapy, if indicated and available in their country. Results: In FRESCO (F n=278 vs P n=138), no pts had received rego and 16/416 pts (4%) had received TAS. InFRESCO-2, prior therapies were balanced between F vs P arms; pts had received prior TAS or rego in these sequences: prior TAS only (n=240 vs n=121); prior rego only (n=40 vs n=18); prior rego then TAS (n=84 vs n=45); prior TAS then rego (n=95 vs n=44). Improvements in OS and PFS were seen with F vs P regardless of prior tx sequence (Table), with similar magnitudes of benefit for OS (HRs 0.53–0.77) and PFS (HRs 0.25–0.37). F safety profile was consistent regardless of prior tx sequence (Table). Conclusions: F is effective in mCRC with a manageable safety profile regardless of prior TAS or rego tx and sequence. Improvement with F was similar for OS and PFS despite the number and sequence of prior regimens. F provides a new tx option for pts with previously treated mCRC regardless of prior TAS or rego exposure. Clinical trial information: NCT04322539 . [Table: see text]

480 Background: Currently, there is a lack of imaging biomarkers of immunotherapy outcome in hepatocellular carcinoma (HCC). The study aim was to determine if HCC enhancement on MRI and stiffness change measured by magnetic resonance elastography (MRE) can predict immunotherapy response. Methods: This was a prospective, Institutional Review Board approved study of 38 patients with HCC treated with immune checkpoint blockade (ICB) therapy. All patients had liver MRI/MRE and HCC biopsy at baseline, and MRI/MRE with biopsy or resection after 6 weeks therapy. HCC stiffness (kPa) was measured on MRE elastograms (liver stiffness maps). HCC enhancement and change in stiffness were compared with treatment response to ICB in 1) non-surgical patients (pembrolizumab), and 2) surgical patients (nivolumab +/- ipilimumab). For non-surgical patients, treatment response was defined as overall survival >1 year. For surgical patients, treatment response was defined as <50% viable tumor at time of resection. Analysis was performed using descriptive statistics and Spearman correlation; p-value <0.05 was considered statistically significant. Results: Twenty-five patients were evaluable. Median age was 67 years (32, 78). Etiology of liver disease was NASH (n=8), HCV (n=8), HBV (n=2) and unknown (n=7). Treatment response occurred in 11/25 (44%) patients. Median HCC size and change in size were 4.7 cm (1.2, 14.0) and –0.32 cm, respectively. Median baseline HCC stiffness and change in stiffness were 5 kPa (2.2, 12.4) and –0.1 kPa (–2.2, 1.5), respectively. Median change in HCC size for responders and non-responders was –1.2 cm (–4.8, 0.4) and 0 cm (–1.5, 1.1), respectively (p = 0.02). Treatment response was associated with absence of portal venous phase capsular enhancement and increase in HCC stiffness, (p<0.001). Conclusions: Capsular enhancement and MRE stiffness change may be useful biomarkers of immune cell activated response to ICB therapy.

Purpose Treatment options for neuroendocrine tumors (NETs) remain limited. This trial assessed the progression-free survival (PFS) of bevacizumab or interferon alfa-2b (IFN-α-2b) added to octreotide among patients with advanced NETs. Patients and Methods Southwest Oncology Group (SWOG) S0518, a phase III study conducted in a US cooperative group system, enrolled patients with advanced grades 1 and 2 NETs with progressive disease or other poor prognostic features. Patients were randomly assigned to treatment with octreotide LAR 20 mg every 21 days with either bevacizumab 15 mg/kg every 21 days or 5 million units of IFN-α-2b three times per week. The primary end point was centrally assessed PFS. This trial is registered with ClinicalTrials.gov as NCT00569127. Results A total of 427 patients was enrolled, of whom 214 were allocated to bevacizumab and 213 to IFN-α-2b. The median PFS by central review was 16.6 months (95% CI, 12.9 to 19.6 months) in the bevacizumab arm and was 15.4 months (95% CI, 9.6 to 18.6 months) in the IFN arm (hazard ratio [HR], 0.93; 95% CI, 0.73 to 1.18; P = .55). By site review, the median PFS times were 15.4 months (95% CI, 12.6 to 17.2 months) for bevacizumab and 10.6 months (95% CI, 8.5 to 14.4 months) for interferon (HR, 0.90; 95% CI, 0.72 to 1.12; P = .33). Time to treatment failure was longer with bevacizumab than with IFN (HR, 0.72; 95% CI, 0.58 to 0.89; P = .003). Confirmed radiologic response rates were 12% (95% CI, 8% to 18%) for bevacizumab and 4% (95% CI, 2% to 8%) for IFN. Common adverse events with bevacizumab and octreotide included hypertension (32%), proteinuria (9%), and fatigue (7%); with IFN and octreotide, they included fatigue (27%), neutropenia (12%), and nausea (6%). Conclusion No significant differences in PFS were observed between the bevacizumab and IFN arms, which suggests that these agents have similar antitumor activity among patients with advanced NETs.

609 Background: Targeting MEK is of interest in the development of novel agents for treatment of many malignancies. However, better strategies are needed to overcome acquired resistance to MEK inhibitors. Preclinical studies have shown Wnt pathway overexpression in KRAS mutant cell lines resistant to the MEK inhibitor, Selumetinib. The combination of selumetinib and cyclosporin A (CsA), a non-canonical WnT pathway modulator, demonstrated antitumor activity in patient-derived xenograft (PDX) models. We are conducting an NCI CTEP-approved Phase I/IB trial (NCI # 9571/COMIRB # 13-2628) of selumetinib and CsA combination. Biomarkers of response to therapy are being co-developed. We hypothesize that this combination will be safe and potentially effective in patients with mCRC and that upregulation of FZD2 may predict for sensitivity. Methods: Phase I trial with initial dose escalation investigating the combination of selumetinib and CsA in patients with advanced solid tumors (n = 18) followed by an expansion cohort in patients with irinotecan and oxaliplatin-refractory mCRC (n = 20). The expansion cohort utilizes a selumetinib “run-in” to evaluate efficacy in RAS-WT and RAS-MT mCRC to identify those patients most likely to respond to the combination. Results: 18 patients were enrolled in the dose escalation phase and 10 patients have been enrolled in the dose expansion phase as of September 2016. Grade 1 or 2 nausea and rash were reported as the most common AEs. Most commonly reported Grade 3 or 4 toxicities were hypertension, elevated liver enzymes and rash. Three DLTs were reported with Grade 3 hypertension noted at dose level 1 and 2 and grade 3 rash reported at dose level 2. The maximum tolerated dose was defined as Selumetinib 75 mg BID and CsA 2 mg/kg BID on a 28-day cycle. Two partial responses and sixteen stable disease responses have been observed. Six patients have exhibited progressive disease. Conclusions: Selumetinib in combination with cyclosporin A appears to be well-tolerated with evidence of activity in solid tumors. Expansion cohort will complete enrollment this month. Clinical trial information: NCT02188264.

224 Background: In the phase 3 RADIANT-3 trial, everolimus (EVE) demonstrated significantly improved median progression-free survival (PFS) versus placebo (PBO) (11.0 vs 4.6 months; HR=0.35, p < 0.0001) in patients (pts) with pancreatic neuroendocrine tumors (pNET) (Yao et al, NEJM, 2011). Here we present a planned exploratory analysis by prior chemotherapy (chemo) use. Methods: Pts with progressive low- or intermediate-grade pNET were prospectively stratified by prior chemo use and randomized (1:1) to EVE 10 mg/d (n = 207) or PBO (n = 203) plus best supportive care. Results: Of the 410 pts, 206 (50%) had received prior chemo and 204 (50%) were chemo naive. Baseline characteristics (age, sex, race, tumor type, histologic grade) and baseline tumor biomarker levels were similar for pts with/without prior chemo. More chemo-naive pts were newly diagnosed (time since initial diagnosis ≤ 6 mo: 50 [25%] vs. 7 [3%] pts with prior chemo). A lower proportion of chemo-naive pts received prior somatostatin therapy (45% vs. 54% prior chemo). EVE significantly prolonged median PFS regardless of prior chemo use (with prior chemo: 11.0 vs. 3.2 months; HR = 0.35, 95% CI 0.25-0.49, p < 0.001; chemo naive: 11.4 vs. 5.4 months; HR = 0.42, 95% CI 0.29-0.60; p < 0.001). Stable disease was the best overall response in 73% of EVE-treated pts irrespective of prior chemo use. The benefit from EVE irrespective of prior chemo was further demonstrated by disease stabilization or minor tumor shrinkage and in the lower incidence of progressive disease in waterfall plots. Safety of EVE was consistent regardless of prior chemo use. The most common drug-related adverse events (chemo vs chemo naive) included rash (53% vs. 52%), stomatitis (52% vs. 56%), diarrhea (41% vs. 52%), and fatigue (37% vs. 51%). Conclusions: This planned subgroup analysis demonstrated the beneficial effects of EVE in pts with pNET regardless of prior chemo use. These findings support the possible first-line use of EVE for pts with pNET.

158 Background: There is an unmet medical need for effective treatments for patients with advanced pNET. Systemic therapies for advanced pNET are limited both by toxicity and efficacy. Everolimus, an oral mTOR inhibitor, has shown promising antitumor activity in 2 phase II studies, leading to further investigation in the largest phase III randomized controlled trial completed in pNET patients. Methods: Patients with advanced low- or intermediate-grade pNET were randomly assigned to everolimus 10 mg/d orally + best supportive care (BSC; n = 207) or placebo + BSC (n = 203). Long-acting somatostatin analogs (SSAs) were permitted as BSC during the study. The primary endpoint was progression free survival (PFS). At progression (RECIST), patients could be unblinded and those randomly assigned to placebo were offered open-label everolimus. Results: Compared with placebo, everolimus reduced the risk of progression by 65% and increased median PFS by more than 6 months, from 4.6 to 11.0 months (HR = 0.35; 95% CI: 0.27-0.45; p < 0.0001), by investigator review (primary endpoint). Median PFS by central review was consistent (HR = 0.34; 95% CI: 0.26 to 0.44; p < 0.001] in favor of everolimus. Eighteen-month PFS estimates were 34% for everolimus (95% CI: 26-43) vs 9% (95% CI: 4-16) for placebo. Everolimus demonstrated a significant PFS benefit across all patient subgroups according to baseline characteristics and prior SSA use. Prior SSA use was 49% in the everolimus arm and 50% in the placebo arm. Updated analyses of the impact of concomitant SSA will be reported. The most common drug-related adverse events were stomatitis, rash, diarrhea, fatigue, and infections (primarily upper respiratory); most were grade 1 or 2. Stomatitis (6.9% vs 0%), anemia (6% vs 0%), and hyperglycemia (5% vs 2%) were the most common grade 3-4 events. Conclusions: Everolimus significantly prolonged PFS compared with placebo in patients with advanced pNET in this large phase III clinical trial. This benefit was seen across all patient subgroups. Treatment resulted in a significant 6.4-month prolongation in median PFS. Everolimus had an acceptable and predictable safety profile. [Table: see text]

Purpose Everolimus improved median progression-free survival by 6.4 months in patients with advanced pancreatic neuroendocrine tumors (NET) compared with placebo in the RADIANT-3 study. Here, we present the final overall survival (OS) data and data on the impact of biomarkers on OS from the RADIANT-3 study. Methods Patients with advanced, progressive, low- or intermediate-grade pancreatic NET were randomly assigned to everolimus 10 mg/day (n = 207) or placebo (n = 203). Crossover from placebo to open-label everolimus was allowed on disease progression. Ongoing patients were unblinded after final progression-free survival analysis and could transition to open-label everolimus at the investigator’s discretion (extension phase). OS analysis was performed using a stratified log-rank test in the intent-to-treat population. The baseline levels of chromogranin A, neuron-specific enolase, and multiple soluble angiogenic biomarkers were determined and their impact on OS was explored. Results Of 410 patients who were enrolled between July 2007 and March 2014, 225 received open-label everolimus, including 172 patients (85%) randomly assigned initially to the placebo arm. Median OS was 44.0 months (95% CI, 35.6 to 51.8 months) for those randomly assigned to everolimus and 37.7 months (95% CI, 29.1 to 45.8 months) for those randomly assigned to placebo (hazard ratio, 0.94; 95% CI, 0.73 to 1.20; P = .30). Elevated baseline chromogranin A, neuron-specific enolase, placental growth factor, and soluble vascular endothelial growth factor receptor 1 levels were poor prognostic factors for OS. The most common adverse events included stomatitis, rash, and diarrhea. Conclusion Everolimus was associated with a median OS of 44 months in patients with advanced, progressive pancreatic NET, the longest OS reported in a phase III study for this population. Everolimus was associated with a survival benefit of 6.3 months, although this finding was not statistically significant. Crossover of patients likely confounded the OS results.