Search

PURPOSE The therapeutic management of pancreatic neuroendocrine tumors (PanNETs) is based on pathological tumor grade assessment. A noninvasive imaging method to grade tumors would facilitate treatment selection. This study evaluated the ability of quantitative image analysis derived from computed tomography (CT) images to predict PanNET grade. METHODS Institutional database was queried for resected PanNET (2000-2017) with a preoperative arterial phase CT scan. Radiomic features were extracted from the primary tumor on the CT scan using quantitative image analysis, and qualitative radiographic descriptors were assessed by two radiologists. Significant features were identified by univariable analysis and used to build multivariable models to predict PanNET grade. RESULTS Overall, 150 patients were included. The performance of models based on qualitative radiographic descriptors varied between the two radiologists (reader 1: sensitivity, 33%; specificity, 66%; negative predictive value [NPV], 63%; and positive predictive value [PPV], 37%; reader 2: sensitivity, 45%; specificity, 70%; NPV, 72%; and PPV, 47%). The model based on radiomics had a better performance predicting the tumor grade with a sensitivity of 54%, a specificity of 80%, an NPV of 81%, and a PPV of 54%. The inclusion of radiomics in the radiographic descriptor models improved both the radiologists' performance. CONCLUSION CT quantitative image analysis of PanNETs helps predict tumor grade from routinely acquired scans and should be investigated in future prospective studies.

Background Grade 3 gastroenteropancreatic neuroendocrine neoplasms (G3 GEPNENs) are often aggressive, and the optimal treatment is unclear for this subgroup of neuroendocrine neoplasms (NENs). Temozolomide (TEM)-based regimens have been increasingly used to treat grade 1–2 NENs, but their efficacy in G3 NENs remains undetermined. We aimed to assess the clinical efficacy of TEM-containing regimens in advanced grade 3 GEPNENs. Materials and Methods A multicenter retrospective review (2008–2018) of patients with metastatic/unresectable G3 GEPNENs who received a TEM-containing regimen was undertaken within a North American partnership to pool data. The primary endpoint was time to treatment failure (TTF). Radiologic response was extracted from local reports. Results One hundred and thirty patients in six high-volume NEN centers were included (median age 55, 64% male, 18% functional, 67% pancreatic NEN). Forty-nine percent were well-differentiated, 35% poorly differentiated, and 15% unknown based on local pathology reports. The regimen used was capecitabine and temozolomide (CAPTEM) in 92% and TEM alone in 8%. Radiological response by local assessment was seen in 36% of patients. Median TTF was 3.6 months and median overall survival (OS) 19.2 months. Six percent of patients required discontinuation of therapy due to adverse events. TTF was longer in first-line treatment (7.8 months vs. 2.9 months; hazard ratio, 1.62; 95% confidence interval, 1.11–2.36; p = .015) and in patients with pancreatic NENs (panNENs) compared with gastrointestinal NENs (5.8 months vs 1.8 months; p = .04). The overall response rate was higher in the first-line setting (51% vs 29%; p = .02) and in panNEN (41% vs 23%; p = .04). Conclusion This is the largest TEM treatment series in G3 NEN, involving collaboration of several major North American NEN centers as a partnership. Thirty-six percent of patients showed some degree of radiographic response, and treatment was generally well tolerated, although the median duration of response was short. Response rates and time to treatment failure were superior in the first-line setting. CAPTEM should be considered a viable treatment option in this setting. Further randomized trials are warranted. Implications for Practice Neuroendocrine neoplasms (NENs) are heterogeneous, and optimal treatment for aggressive grade 3 (G3) NENs remains undetermined. The capecitabine and temozolomide (CAPTEM) regimen has been used in low-grade pancreas NENs but there are few data for its safety and efficacy in the G3 setting. This article reports on the efficacy of temozolomide-containing regimens, particularly CAPTEM, in management of G3 NENs. The good tolerance and response rate show that CAPTEM should be considered a viable regimen in treatment of G3 NENs pending confirmatory prospective studies.

Paired imaging/therapy with radiolabeled somatostatin receptor (SSTR) antagonists is a novel approach in neuroendocrine tumors (NETs). The aim of this study was to compare tumor uptake of 68Ga-DOTA-JR11 and 177Lu-satoreotide tetraxetan (177Lu-DOTA-JR11) in patients with NETs. As part of a prospective clinical trial, 20 patients with metastatic NETs underwent 68Ga-DOTA-JR11 PET/CT and serial imaging with 177Lu-satoreotide tetraxetan. PET/CT and SPECT/CT parameters for lesion uptake and absorbed dose of 177Lu-satoreotide tetraxetan in lesions were compared using linear regression analysis and Pearson correlation. A total of 95 lesions were analyzed on 68Ga-DOTA-JR11 PET/CT and 177Lu-satoreotide tetraxetan SPECT/CT. SUVs and tumor-to-normal-tissue ratios on PET/CT and SPECT/CT were significantly correlated (p

1669 Introduction: Microcalcification in the arterial wall traced by 18F-sodium fluoride (NaF) positron emission tomography (PET) is considered an indicator of early phase atherosclerosis and a potential precursor of macrocalcification detectable by computed tomography (CT). Ongoing carotid artery disease is a major cause of stroke that typically occurs without preceding symptoms. Thus, methods that can detect and monitor early phase carotid atherosclerosis may be of clinical interest. We aimed to examine 2-year changes in carotid NaF uptake by means of repeat NaF-PET/CT. Methods: Twenty-nine healthy subjects and 20 angina pectoris patients underwent 90-min NaF-PET/CT twice two years apart using the same scanner in each patient on both occasions. The right and left main carotid arteries were manually segmented. NaF uptake was expressed as the mean and total standardized uptake values without and with partial volume correction (SUVmean, SUVtotal and cSUVmean, cSUVtotal). Results: Recorded NaF uptake values are shown in Table 1. Insignificant tendencies were similar or higher NaF uptake in angina patients at both time points with less uptake in healthy subjects and generally higher uptake in angina patients after 2 years. Thus, mean (± standard deviation) right carotid cSUVmean of angina patients was 1.84±0.56 and 1.88±0.97 at baseline and after 2 years (p=0.15) vs. 1.62±0.35 and 1.44±0.36 in healthy subjects (p=0.001). Similarly, left carotid cSUVmean of angina patients was 1.5±0.44 and 1.53±0.51 at baseline and after 2 years (p=0.87) vs. 1.43±0.23 and 1.34±0.27 in healthy subjects (p=0.08). NaF uptake at baseline could not predict a change in CT-calcification after 2 years. Conclusions: Slightly higher NaF uptake in the angina group indicated more ongoing carotid microcalcification at both time points in patients than in healthy subjects. The 2-year changes were very small in both groups, albeit with a tendency of slight decreases among healthy controls, suggesting a more dynamic, waxing and waning, process in these compared to a slow, but more consistent, progression in angina patients despite statin therapy in half of these.

The NCCN Clinical Practice Guidelines in Oncology (NCCN Guidelines) for Neuroendocrine and Adrenal Gland Tumors focus on the diagnosis, treatment, and management of patients with neuroendocrine tumors (NETs), adrenal tumors, pheochromocytomas, paragangliomas, and multiple endocrine neoplasia. NETs are generally subclassified by site of origin, stage, and histologic characteristics. Appropriate diagnosis and treatment of NETs often involves collaboration between specialists in multiple disciplines, using specific biochemical, radiologic, and surgical methods. Specialists include pathologists, endocrinologists, radiologists (including nuclear medicine specialists), and medical, radiation, and surgical oncologists. These guidelines discuss the diagnosis and management of both sporadic and hereditary neuroendocrine and adrenal tumors and are intended to assist with clinical decision-making. This article is focused on the 2021 NCCN Guidelines principles of genetic risk assessment and counseling and recommendations for well-differentiated grade 3 NETs, poorly differentiated neuroendocrine carcinomas, adrenal tumors, pheochromocytomas, and paragangliomas.

10505 Background: Lynch syndrome (LS) is caused by germline mutations in DNA mismatch repair genes and is characterized by microsatellite instability (MSI) status of associated tumors. MSI sensitizes tumors to immune checkpoint inhibition (ICI). The incidence of second primary malignancies and colonic polyps in LS patients with MSI tumors following ICI has not been evaluated. Methods: The Memorial Sloan Kettering LS database was queried for all patients with cancer who received ≥1 cycle of ICI using an IRB approved protocol. LS was confirmed by presence of a germline pathogenic/likely pathogenic alteration in a DNA mismatch repair gene ( MLH1, MSH2, MSH6, PMS2, EPCAM). Tumor and matched normal DNA sequencing was performed via MSK-IMPACT, an IRB approved protocol (NCT01775072). MSI status was assessed using MSIsensor. Results: At our center, 131 patients with LS received ICI mostly (73.3%) in the context of metastatic cancer. While 108 patients received ICI for an MSI tumor, 23 LS patients received ICI for a MSS (Microsatellite Stable) tumor (4 hepatobiliary, 3 colorectal, 3 CNS, 2 endometrial, 2 gastroesophageal, 2 renal, 2 sarcoma, 5 other). Five patients (3.8%) developed ≥1 second primary malignancy while on or following ICI comprised of an MSI small bowel adenocarcinoma, MSI upper tract urothelial cancer, 2 MSS prostate cancers, MSS HCC and MSI sebaceous neoplasm; 1 patient developed both MSS prostate cancer and MSI urothelial cancer on ICI. All five patients’ primary malignancy was MSI. Only a minority of patients underwent surveillance colonoscopies post completion of ICI (31.3%, (41/131)). Of these 41, nine (22%, (9/41)) were identified to have polyps including 8 tubular adenomas and 1 tubulo-villous adenoma. Median time to development of a polyp was 22 months (95% CI 16.82-27.18) from last colonoscopy and 13 months (95% CI 4-25 months) from last ICI. Notably amongst the 23 LS patients whose tumors were MSS, 14 had progression on ICI. Conclusions: Herein, we demonstrate that in LS-patients in receipt of ICI for cancer treatment, the risk of a second primary cancer and polyps remain high following treatment with ICI. Biological mechanisms underlying immune escape warrant further investigation. Surveillance strategies should be continued for LS patients post ICI.

4004 Background: Patients with advanced pancreatic neuroendocrine tumors (NETs) have few treatment options that yield objective tumor response. Retrospective and small, prospective studies suggest that the combination of capecitabine and temozolomide is associated with high response rates (RR) and relative long progression-free survival (PFS). This trial was conducted to establish a role for the combination of capecitabine and temozolomide. Methods: E2211 was a multicenter, randomized, phase II trial comparing temozolomide (200 mg/m2 PO QD days 1-5) vs. capecitabine/temozolomide (capecitabine 750 mg/m2 PO BID days 1-14; temozolomide 200 mg/m2 PO QD days 10-14) in patients with advanced pancreatic NETs. Eligibility criteria included: metastatic or unresectable, low or intermediate grade pancreatic NETs, progression within preceding 12 months, and no prior temozolomide, DTIC, capecitabine or 5-fluorouracil. The primary endpoint was PFS; secondary endpoints were Overall Survival (OS), RR, safety, and MGMT as evaluated by immunohistochemistry (IHC) and promoter methylation. Allowing for 5% ineligibility, 145 randomized patients were required to obtain 138 eligible patients to detect a difference in median PFS of 9 versus 14 months (hazard ratio of 0.64) using a two-sided log-rank test at the overall 0.20 significance level with 81% power. Results: 144 patients were enrolled between 4/2013 to 3/2016 to temozolomide (n = 72) or capecitabine/temozolomide (n = 72); the efficacy analysis population included 133 eligible patients. At the scheduled interim analysis in January 2018, median PFS was 14.4 months for temozolomide vs. 22.7 months for capecitabine/temozolomide (HR = 0.58), which was sufficient to reject the null hypothesis for this final primary endpoint (stratified log rank p = 0.022. In the final analysis (5/2021), median OS was 53.8 months for temozolomide and 58.7 months for capecitabine/temozolomide (HR = 0.82, p = 0.42) and RR was 34% for temozolomide and 40% for capecitabine/temozolomide (p = 0.59). Capecitabine/temozolomide was associated with higher rates of grade 3-4 AEs (45% vs. 22%, p = 0.005). MGMT deficiency, defined as either low IHC or positive promoter methylation, was associated with greater odds of response (OR [95% CI] = 6.38 [2.19, 18.60] and 9.79 [1.09, 87.71], respectively). Conclusions: E2211 is the first prospective randomized trial of capecitabine/temozolomide and shows the longest PFS and highest RR reported for patients with pancreatic NETs in a prospective randomized study. MGMT deficiency was associated with greater odds of objective response. Clinical trial information: NCT01824875.

Background /Objectives: Pancreatic adenocarcinoma (PDAC) metastatic to the leptomeninges is a rare and lethal event. Leptomeningeal disease (LMD) research is limited in PDAC, and insights into clinical descriptors, possible disease predictors, and treatment strategies is necessitated. Methods Memorial Sloan Kettering databases were queried with Institutional Review Board approval to identify patients with LMD and PDAC treated between January 2000 and June 2020. Medical record review was used to abstract clinical, genomic, pathologic, and radiographic data. Overall survival was calculated from date of PDAC diagnosis to date of death. Previously published literature on LMD from PDAC was reviewed. Results Four patients with LMD from PDAC were identified, two males and two females. Age at diagnosis ranged from 57 to 68 years. All four patients had predominant lung metastasis and a relatively low burden of intra-abdominal disease. Somatic testing indicated alterations typical of PDAC and no PDAC defining pathogenic germline mutations were identified. An extended clinical course prior to LMD diagnosis was observed in all patients, ranging from 16 to 148 months. Upon diagnosis of LMD, three patients elected for supportive care and one patient received a limited course of craniospinal radiation. The median survival following diagnosis of LMD was 1.6 months (range 0.5–2.8 months). Conclusions LMD from PDAC is a rare occurrence that may be more frequent in patients with lung metastasis and/or a more indolent clinical course. Following diagnosis of LMD, prognosis is poor, and survival is short. New treatment strategies for this manifestation of PDAC are needed.

507 Background: Young onset of several gastrointestinal cancers is associated with more advanced disease and poor oncological outcomes, however, this association has not been fully investigated for well differentiated pancreatic neuroendocrine tumors (PanNET). Our study aimed to evaluate clinical and pathological differences as well as disease outcomes between young-onset PanNET (YO-PanNET) and late-onset PanNET (LO-PanNET). Methods: Patients with localized PanNET who underwent surgery at Memorial Sloan Kettering Cancer Center between 2000 to 2017 were identified. Those with hereditary syndromes, metastatic disease, and postoperative mortality were excluded. YO-PanNET was defined as 50 years of age at time of diagnosis. Family history, clinical and pathology characteristics, were recorded. Fisher’s exact test was used to detect difference between the age groups. Kaplan-Meier method was used to investigate age prognostic significance. Results: A total of 366 patients were identified, 84 (23%) with YO-PanNET. Compared with LO-PanNET, YO-PanNET were less likely to have a personal history of another cancer (p

506 Background: Immune checkpoint inhibitors (ICIs) have antitumor activity in some solid tumors, however a role for ICI’s in the treatment of neuroendocrine neoplasms (NENs) is uncertain. We analyzed response and outcomes of NENs, both well differentiated neuroendocrine tumors (WDNET) and poorly differentiated neuroendocrine carcinomas (PDNEC), treated with ICIs at Memorial Sloan Kettering Cancer Center during the past decade. Methods: Patients with NENs who received ICIs between 2010-2021 were identified. Demographics, pathology characteristics, treatment response and outcomes were recorded. Progression-free survival (PFS) and overall survival (OS) were estimated using Kaplan-Meier methods. Analysis of next-generation sequencing (NGS) results from archival tumor tissue was performed. Results: 57 patients with NENs treated with ICIs were identified (WDNET-16; PDNEC-41). All patients had heavily-treated metastatic disease (median number of prior systemic treatments, WDNET-4, PDNEC-2). Median time from diagnosis to ICI treatment was 41.5 months in WDNET, 8 months in PDNEC. WDNET response: 1 (6%) partial response (PR), 4 (25%) stable disease (SD), 10 (63%) progressive disease (PD). PDNEC response: 6 (15%) PR (5 microsatellite stable, 1 unknown), 3 (7%) SD, 32 (78%) PD. Median duration of ICI treatment: 2 months (range 0-27) in WDNET, 1 month (range 0-16) in PDNEC; median duration of treatment for the 7 responders was 11 months (range 2-27). While receiving ICIs, 3 PDNEC responders developed PD in an escape lesion (2 brain, 1 nodal) after response in all other disease sites and successfully continued ICI treatment post-RT of escape lesion. Six-month PFS was 20% [95%CI: 7%-55%] and 9.8% [95%CI: 4%-25%], OS 60% [95%CI: 40%-91%] and 37% [95%CI: 25%-56%], respectively in WDNET and PDNEC. NGS results were available in archival tumor tissue of 35 patients (61%); most common alterations were TP53 (20/26, 77%) and RB1 (14/26, 54%) in PDNEC versus ATM, SMAD4, CREBBP, and IRS2 (all 1/9, 11%) in WDNET. Conclusions: We observed limited activity for ICI’s in NEN treatment, primarily restricted to a subset of PDNEC. PFS/OS findings were consistent with historical data, reflecting the pathogenesis of distinct NEN subtypes.

TPS514 Background: Gastroenteropancreatic neuroendocrine tumors (GEP-NETs), which represent approximately 70% of NETs, frequently develop metastatic disease with limited treatment options. Current standard therapies for well-differentiated high grade 2 and grade 3 GEP-NETs include cytoreductive procedures, somatostatin analogues, molecular targeted therapies (everolimus or sunitinib), chemotherapy and peptide receptor radionuclide therapy (PRRT), with no specified sequence of use. PRRT may stabilize disease and induce objective tumor responses. This treatment uses radiolabeled somatostatin analogues to selectively target somatostatin receptor expressing (SSTR+) tumor cells. 177Lu-edotreotide is an innovative radiolabeled somatostatin analogue with a favorable safety profile and promising efficacy. Retrospective data in metastatic GEP-NETs treated with two or more 177Lu-edotreotide cycles demonstrated a progression free survival (PFS) of at least 30 months. The currently recruiting Phase III COMPETE trial compares the efficacy and safety of 177Lu-edotreotide, versus everolimus, in grade 1 and grade 2 GEP-NETs. Methods: COMPOSE (NCT04919226) is a prospective, randomized, controlled, open-label, multi-center Phase III study, in patients with well-differentiated high grade 2 and grade 3 (Ki-67 index 15−55%), SSTR+, GEP-NETs. This trial is to evaluate the efficacy, safety and patient-reported outcomes of first- or second-line treatment with 177Lu-edotreotide PRRT compared to best standard of care. It aims to randomize 202 patients 1:1 to a defined number of cycles of 177Lu-edotreotide or an active comparator (either chemotherapy [CAPTEM or FOLFOX] or everolimus, according to investigator´s choice). The primary endpoint is PFS, assessed every 12 weeks until disease progression (RECIST v1.1), or death, whichever occurs earlier. Secondary outcomes include overall survival, assessed up to 2 years after disease progression. Study recruitment for COMPOSE commenced in September 2021. It is expected that COMPOSE will inform optimal treatment options for patients with well-differentiated high grade 2 and grade 3 SSTR+ GEP-NETs, including for first-line therapy. Clinical trial information: NCT04919226.

Somatostatin receptor antagonists have shown promise for imaging neuroendocrine tumors (NETs) in preclinical studies, but clinical data is still very limited. In this study, we assess the feasibility of using the novel somatostatin antagonist 68Ga-DOTA-JR11 for PET imaging of NETs. Twenty patients with advanced NETs underwent whole-body PET/CT imaging 60 min after injection of 169 MBq (median) 68Ga-DOTA-JR11 as part of a prospective study. Volumes of interest were drawn around up to four 68Ga-DOTA-JR11-avid lesions per patient (with uptake greater than liver) and standardized uptake values were estimated. Additionally, target-to-normal tissue ratios were calculated. A subset of six patients had additional imaging (25-min dynamic scan of the upper abdomen including, at least partly, cardiac left ventricle, liver, spleen, and kidney, and a whole-body PET/CT scan at 30 min post-injection) to determine the time course of tracer distribution and facilitate radiation dose estimates. Absorbed doses were calculated using OLINDA/EXM 1.0. In contrast to the known biodistribution of somatostatin receptor agonists, little or no uptake above background was seen in the pituitary gland, spleen, adrenals, and uninvolved liver; e.g., median spleen SUVmean 1.4 (range: 0.7–1.8), liver SUVmean 1.1 (0.7–1.9). A total of 42 tumor lesions were analyzed with median SUVmax 13.0 (range: 2.9–94), TNR blood 9.3 (1.8–87), TNR spleen 4.9 (1.9–48), TNR kidney 2.2 (0.52–28), and TNR liver 10.5 (2.3–107). Tumor uptake reached plateau levels by 20-30 min post-injection. The highest absorbed dose estimates (mGy/MBq) to normal tissues were: urinary bladder wall (0.30; SD 0.06) and kidneys (0.050; SD 0.013). The effective dose (ICRP 103) was 0.022 (SD 0.003) mSv/MBq. 68Ga-DOTA-JR11 demonstrated rapid tumor uptake, high tumor/background ratios, and rapid clearance from blood. The low liver background is advantageous and may facilitate detection of liver metastases. Dosimetric data compare favorably with published data for 68Ga-DOTATATE and 68Ga-DOTATOC.

Aim:\ud Significant recent changes in management of locally advanced rectal cancer include preoperative staging, use of extended neoadjuvant therapies, and minimally invasive surgery (MIS). This study was aimed at characterizing those changes and associated short‐term outcomes.\ud \ud Method:\ud We retrospectively analysed treatment and outcome data from patients with T3/4 or N+ locally advanced rectal cancer ≤15 cm from the anal verge who were evaluated at a comprehensive cancer center in 2009–2015.\ud \ud Results:\ud In total, 798 patients were identified and grouped into five cohorts based on treatment year: 2009‐2010, 2011, 2012, 2013, and 2014‐2015. Temporal changes included increased reliance on MRI staging, from 57% in 2009‐2010 to 98% in 2014‐2015 (p < 0.001); increased use of total neoadjuvant therapy, from 17% to 76% (p < 0.001); and increased use of MIS, from 33% to 70% (p < 0.001). Concurrently, median hospital stay decreased (from 7 to 5 days; p < 0.001), as did the rates of grade III‐V complications (from 13% to 7%; p < 0.05), surgical site infections (from 24% to 8%; p < 0.001), anastomotic leak (from 11% to 3%; p < 0.05), and positive circumferential resection margin (from 9% to 4%; p < 0.05). TNM downstaging increased from 62% to 74% (p = 0.002).\ud \ud Conclusion:\ud Shifts toward MRI‐based staging, total neoadjuvant therapy, and MIS occurred between 2009 and 2015. Over the same period, treatment responses improved, and lengths of stay and the incidence of complications decreased.

Context Optimal advance care planning allows patients to articulate their values as a touchstone for medical decision making. Ideally, this occurs when patients are clinically stable, and with opportunities for iteration as the clinical situation unfolds. Objectives Testing feasibility and acceptability in busy outpatient oncology clinics of a novel program of systematic, oncology nurse-led values discussions with all new cancer patients. Methods Within an institutional initiative integrating primary and specialist palliative care from diagnosis for all cancer patients, oncology nurses were trained to use specific questions and an empathic communication framework to discuss health-related values during outpatient clinic visits. Nurses summarized discussions on a template for patient verification, oncologist review, and electronic medical record documentation. Summaries were reviewed with the patient at least quarterly. Feasibility and acceptability were evaluated in three clinics for patients with hematologic or gastrointestinal malignancies. Results Oncology nurses conducted 177 total discussions with 67 newly diagnosed cancer patients (17 with hematologic and 50 with gastrointestinal malignancies) over two years. No patient declined participation. Discussions averaged eight minutes, and all patients verified values summaries. Clinic patient volume was maintained. Of 31 patients surveyed, 30 (97%) reported feeling comfortable with the process, considered it helpful, and would recommend it to others. Clinicians strongly endorsed the values discussion process. Conclusion Nurse-led discussions of patient values soon after diagnosis are feasible and acceptable in busy oncology clinics. Further research will evaluate the impact of this novel approach on additional patient-oriented outcomes after broader dissemination of this initiative throughout our institution.

1504 Background: Oncology patients are particularly vulnerable to adverse outcomes from COVID-19 and require careful monitoring to identify early deterioration and render higher level care when indicated. Several institutions launched remote patient monitoring programs (RPMPs) to care for patients with COVID-19. We describe patients’ perspectives on a COVID-19 RPMP at a National Comprehensive Cancer Center. Methods: Adult patients who had either tested positive for COVID-19 on an outpatient microbiology test or were discharged after hospitalization for the virus were eligible. Patients enrolled in the RPMP received a daily 10-question electronic patient-reported outcome assessment of COVID-19 symptoms and their responses generated alerts to a centralized monitoring team for new or worsening symptoms. A subset of high-risk patients also received a pulse oximeter which alerted when blood oxygen levels dropped below 93%. RPM was discontinued 14 days after a patient’s positive test result and following 3 days without worsening symptoms or fever. Patients who exited the program and had completed at least one assessment were sent a patient engagement survey. The objective of the survey was to evaluate the patient’s experience with digital monitoring and symptom management for COVID-19. The assessment was structured with objective response questions, including a net promoter score, and free text questions to elicit patient perspectives on RPM value. Free text responses were analyzed using grounded theory to identify primary themes regarding perceived value. Results: The survey was distributed to 452 patients; 241 responded as of June 10, 2020 (53% completion rate). The net promoter score was 91%. The table provides responses to objective questions. Qualitative analysis of free text responses identified the primary themes regarding patient perceived value which included: 1) Security: patients appreciated that the RPMP provided a clinical safety net; 2) Connection: patients appreciated the link to their clinical team during a period of isolation; 3) Empowerment: patients appreciated that the RPMP provided education on the virus and symptom management. Conclusions: RPMPs are perceived to be of value to oncology patients with COVID-19. A key barrier to maintaining these programs is cost. Policymakers should consider how these programs can be reimbursed in the future so that they can continue to provide care to vulnerable patients and keep them at home out of the acute care setting.[Table: see text]

e16204 Background: The somatostatin analogs (SSAs) octreotide LAR and lanreotide are equally acceptable in the NCCN guidelines to treat WDNETs. Average Sales Price for 1 year of lanreotide at 120mg is $106,802 versus $53,471 for 1 year of octreotide LAR 20mg and $80,206 for 1 year of 30mg. Lanreotide is given by “deep subcutaneous injection” while octreotide LAR is given intramuscularly. We conducted a randomized, blinded trial evaluating patient (pt) experience, as measured by injection site pain, with octreotide LAR and lanreotide, during the treatment of advanced, nonfunctional, WDNETs. We also investigated drug preference and financial toxicity in this pt population. Methods: This randomized single-blinded pilot study enrolled 51 pts recommended to begin SSA therapy. Pts received injections q 4 weeks and received 6 injections on study; Arm 1: octreotide LAR for 3 injections then lanreotide for 3 injections; Arm 2: lanreotide for 3 injections then octreotide LAR for 3 injections. Pts were blinded as to which agent they received throughout the study. Self-reported injection site pain scores were obtained after each of the first 3 injections using a 0 to 10 scale (0: “I didn’t feel it”; 10: “worst pain ever”). Primary endpoint was comparison of mean pain scores over the first 3 injections of octreotide LAR (Arm 1) or lanreotide (Arm 2). Secondary endpoints, evaluated with descriptive statistics, included pt-reported preference of octreotide LAR versus lanreotide, and willingness to pay for the preferred therapy, both assessed after 6 months of therapy by questionnaire. Results: 51 pts were enrolled (Arm 1: N = 26, Arm 2: N = 25). All pts were evaluable for the study primary endpoint. All pts received lanreotide at a dose of 120mg monthly; among those pts (49) receiving octreotide LAR, 30 (61%) received 20mg, 18 (37%) received 30mg, 1 (2%) received 10mg. No significant difference was identified in mean pain scores over the first 3 SSA injections; Arm 1: mean 2.4, standard deviation 1.9 versus Arm 2: mean 1.9, standard deviation 1.5 (p = 0.5). 34/51 (67%) pts (15 pts in Arm 1; 19 pts in Arm 2) were evaluated for secondary endpoints and completed post-therapy questionnaires. 7 (47%) in Arm 1 and 8 (42%) in Arm 2 indicated no drug preference at the end of the 6 months. There was a trend towards preference for octreotide LAR versus lanreotide in both arms, with more pts indicating mild or strong preference for octreotide LAR. 7 (50%) and 10 (56%) of pts in Arms 1 and 2, respectively, were unwilling to pay more for their preferred SSA; the rest of the cohort was willing to experience increased financial toxicity to receive their preferred SSA. Conclusions: This randomized, blinded study evaluating pt comfort with SSAs found minimal pain with both agents and no significant differences in pain scores between octreotide LAR versus lanreotide. Clinical trial information: NCT03289741.

4148 Background: The efficacy of immune checkpoint inhibitor (CPI) therapy has not been established in extrapulmonary poorly differentiated neuroendocrine carcinomas (EP-PDNECs). In small cell lung cancer, CPI therapy is approved for use in the first-line and salvage settings. We investigated the efficacy and safety of pembrolizumab (PEM)-based therapy in biomarker-unselected patients (pts) with EP-PDNECs. PEM alone (Part A, N=14) was inactive (ASCO GI 2019; Abstr#363). We now report the results of Part B (PEM plus chemotherapy). Methods: We conducted an open label, multicenter, phase 2 study of PEM-based therapy in pts with EP-PDNECs, excluding Merkel cell carcinoma and well differentiated grade 3 neuroendocrine tumors (NET), with disease progression on first-line systemic therapy. In Part B of this trial, patients were treated with PEM 200 mg IV every 3 week cycle plus dealers’ choice chemotherapy (chemo): weekly irinotecan (IRI, 125 mg/m2 day 1,8 of every 21 day cycle) or weekly paclitaxel (PAC, 80 mg/m2). After PEM/IRI safety lead-in (N=6), 16 additional pts (total N=22) were enrolled. This was based on a primary endpoint of objective response rate (ORR) by RECIST 1.1 and a plan to test Ha ORR 31% vs H0 ORR 10% with 80% power at a type I error rate of 0.05. Secondary endpoints include safety, overall survival (OS), and progression-free survival (PFS). Serial blood samples and baseline tumor biopsies were required in all pts. Results: Preliminary data from Part B are available. Of 22 pts enrolled, male/female 15/7; median age 57 years (range 34-75); ECOG PS 0/1: 10/12; 6 large cell, 8 small cell, 8 NOS. Primary sites of disease: GI 73%, GYN 5%, unknown 23%. Ki67 index (available for 18 pts) median 68% (range 30 to >95%). Chemo choice: 17 IRI (77%) and 5 PAC (23%). PEM/IRI was safe based on lead-in. Median number of cycles of therapy administered was 3 (range 0-13). Treatment-related Gr 3 or 4 AE occurred in 7 (32%) of 22 pts overall: 4 (18%) had at least one Gr 3 AE attributed to PEM (1 pt each with pain, ALT increase, or nausea; 2 with fatigue); 7 (32%) had at least one Gr 3/4 AE attributed to chemo (2 with fatigue, 2 with neutropenia; 1 each with pain, ALT increase, hyponatremia, diarrhea, nausea, and/or acute kidney injury). No grade 5 AE. ORR was 9%: PR in 2 pts (9%), SD 3 pts (14%), PD 13 pts (60%); 4 pts (18%) unevaluable (off study before first scheduled scan). Median PFS 2 mo. At last follow-up, 5 pts (23%) were alive with 1 pt still on treatment. Median OS 4 mo. Of 21 pts off treatment, 76% off for PD, 10% off for AE, 14% off for withdrawal of consent/other therapy. Conclusions: PEM + chemotherapy was not effective in this pretreated, biomarker-unselected population of EP-PDNECs arising in different organs. Biomarker studies are planned (Parts A/B). Clinical trial information: NCT03136055.

1554 Background: Cancer patients face an increased risk of developing acute complications from COVID-19. Remote monitoring can help with the critical need for early detection of symptoms among those diagnosed with COVID-19, enabling timely symptom management that can mitigate clinical deterioration. In response to this need, Memorial Sloan Kettering Cancer Center fast-tracked a program to monitor patients with COVID-19 from home, using an electronic symptom-tracking questionnaire and digital pulse oximeter to track patients’ status and alert care teams to intervene if symptoms worsened. A multi-disciplinary group composed of Oncology providers, advanced practice providers, nursing, nursing informatics and biomedical informatics formed to manage the program. Methods: Memorial Sloan Kettering launched a remote monitoring program for patients diagnosed with COVID-19 on March 25, 2020. All patients testing positive for COVID-19 were enrolled in the program and asked to complete a daily symptom tracking questionnaire accessed through their patient portal or administered verbally over the phone. A subset of high risk patients were also provided with a digital pulse oximeter linked to their patient portal and capable of transmitting readings directly to the care team. Clinicians received alerts for patients reporting symptoms or an oxygen saturation below 92%. Alerts resulted in an immediate response from the care team to determine if the patient needed additional care. We retrospectively evaluated the program usage, outcomes and learnings from March 25, 2020 to December 22, 2020. Results: In total, 1,721 patients were enrolled in the program from March 25, 2020 to December 22, 2020. Among these, 210 were deemed high risk patients who received a pulse oximeter in addition the daily symptom questionnaire. Over this period, 27% of patients triggered an alert from an electronic symptom questionnaire, and 63% of patients with a pulse oximeter triggered an alert from their device. Among patients who triggered an alert of any kind, 3% were triaged to a higher level of care. Patients reported that the program was highly valued and alleviated anxiety about their care. Iterative improvements were made to the program over time in response to the evolving knowledge about care for patients with COVID-19. Conclusions: Memorial Sloan Kettering was able to quickly implement a program to detect and triage symptoms among patients with COVID-19 and cancer. Refinements were made over time to many aspects of the program in response to learnings about care related to COVID-19, including to clinical eligibility, alert criteria, monitoring duration and workflows. The program also demonstrated value for patients who felt more comfortable with their care while being monitored remotely. This program established a successful model for remote monitoring of patients with COVID-19 with the potential to be scaled to other institutions or clinical areas.

BACKGROUND Ovarian metastases from colorectal cancer (OM-CRC) often are unresponsive to chemotherapy and are associated with poor survival. To the authors' knowledge, the clinicopathologic and genomic predictors of OM-CRC are poorly characterized and optimal clinical management remains unclear. METHODS Women with a histopathological diagnosis of OM-CRC who were treated at Memorial Sloan Kettering Cancer Center from 1999 to 2015 were identified. Next-generation somatic mutation profiling (Memorial Sloan Kettering-Integrated Mutation Profiling of Actionable Cancer Targets [MSK-IMPACT]) was performed on 38 OM-CRC cases, including 21 matched tumor pairs/trios. Regression models were used to analyze variables associated with progression-free survival and overall survival (OS). RESULTS Kirsten Rat Sarcoma Viral Oncogene Homolog (KRAS), SMAD family member 4 (SMAD4), and neurotrophic receptor tyrosine kinase 1 (NTRK1) mutations were more frequent in cases of OM-CRC than in instances of CRC occurring without OM. SMAD4 and lysine methyltransferase 2D (KMT2D) mutations were associated with reduced OS. Matched multisite tumor sequencing did not identify OM-specific genomic alterations. Of the 195 patients who underwent oophorectomy for OM-CRC (median age, 49 years with a progression-free survival of 9.4 months and an OS of 23 months from oophorectomy), 76% had extraovarian metastasis (EOM). In multivariable analysis, residual disease after surgery (R2 resection) was associated with worse survival. Patients with EOM were less likely to achieve R0/R1 surgical resection status (complete macroscopic resection without clinical/radiological evidence of disease) (48% vs 94%). However, if R0/R1 resection status was achieved, both patients with (35.9 months vs 12 months) and without (43.2 months vs 14.5 months) EOM were found to have better OS. Among 114 patients with R0/R1 resection status, 23 (20%) had no disease recurrence, including 10 patients (9%) with > 3 years of follow-up. CONCLUSIONS Loss-of-function alterations in SMAD4 are frequent and predictive of worse survival in patients with OM-CRC. Similar to oligometastatic CRC to the lung or liver, surgical resection of OM-CRC is associated with a better outcome only if all macroscopic metastatic disease is resected. Cancer 2016. © 2016 American Cancer Society.

2027 Background: Early detection and management of symptoms in patients with cancer improves outcomes, however, the optimal approach to symptom monitoring and management is unknown. This pilot program uses a mobile health intervention to capture and make accessible symptom data for high-risk patients to mitigate symptom escalation. Methods: Patients initiating antineoplastic treatment at a Memorial Sloan Kettering regional location were eligible. A dedicated staff of RNs and nurse practitioners managed the patients remotely. The technology supporting the program included: 1) a predictive model that identified patients at high risk for a potentially preventable acute care visit; 2) a patient portal enabling daily ecological momentary assessments (EMA); 3) alerts for concerning symptoms; 4) an application that allowed staff to review and trend symptom data; and 5) a secure messaging platform to support communications and televisits between staff and patients. Feasibility and acceptability were evaluated through enrollment (goal ≥25% of new treatment starts) and response rates (completion of > 50% of daily symptom assessments); symptom alerts; perceived value based on qualitative interviews with patients and providers; and acute care usage. Results: Between October 15, 2018 and July 10, 2019, the pilot enrolled 100 high-risk patients with solid tumors and lymphoma initiating antineoplastic treatment (median age: 66 years, 45% female). This represented 29% of patients starting antineoplastics. Over six months of follow-up, the response rate to the daily assessments was 56% and 93% of patients generated a severe symptom alert (Table). Both patients and providers perceived value in the program and 5,010 symptom-related secure messages were shared between staff and enrolled patients during the follow-up period. There was a preliminary signal in acute care usage with a 17% decrease in ED visits compared to a cohort of high-risk unenrolled patients. Conclusions: This pilot program of intensive monitoring of high-risk patients is feasible and holds significant potential to improve patient care and decrease hospital resources. Future work should focus on the optimal cadence of EMAs, the workforce to support remote symptom management, and how best to return symptom data to patients and clinical teams. [Table: see text]

708 Background: Given encouraging responses of platinum agents and poly-ADP ribose polymerase inhibitors (PARPi) in BRCA mutated (mut) PDAC, we sought to identify patients (pts) with BRCA mut PDAC treated at MSKCC and to evaluate outcome. Methods: Institutional database at MSK with IRB approval was queried for PDAC germline (g) or somatic (s) BRCA1/2 mut. Genomic profiling, clinicopathologic characteristics and outcomes were collected. Overall survival (OS) from diagnosis was estimated using Kaplan-Meier method. Results: n = 126 with BRCA1/2 mut PDAC were identified between 1/2011-12/2018. n = 77 (61%) male and median age of 62 (range 24-85) at diagnosis. n = 78 (62%) had g BRCA mut (n = 21 BRCA1; n = 57 BRCA2). n = 54 (43%) had a family history of BRCA-related malignancies; 35pts (28%) with a personal history of other BRCA-associated malignancy. n = 66 (52%) AJCC stage IV; of these 43pts (65%) received platinum-based therapy with a partial response (PR) in 35pts (81%); median duration 7 months (m) (range 0.5-39m). n = 40 (32%) received ≥ 4 lines of therapy (range 1-6 lines). n = 44 (35%) received PARPi and 11% (n = 14) received immunotherapy. Median OS for the entire cohort 32.1 m (95% CI 23.9, 42.6). Median OS for stage I-II 49.9m (95% CI 38.5,-); stage III 43m (95% CI 33.9,-) and stage IV 19.1m (95% CI 19.1 16.1,25.8). We did not observe a statistically significant difference in OS between BRCA1 vs BRCA2 pts. Conclusions: BRCA mut PDAC constitutes a small but likely distinct biologic subgroup. Improved OS was notable relative to historical data, possibly due to the integration of platinum and PARPi therapy and possibly due to contribution from disease biology. [Table: see text]

Solid pseudopapillary neoplasms (SPNs) are rare and relatively indolent tumors of the pancreas. While primary SPNs can be surgically resected, there are currently no therapies available for patients with advanced stage disease. Given that these tumors frequently carry CTNNB1 hotspot (recurrently mutated loci in a gene) mutations resulting in β-catenin nuclear accumulation, it has been speculated that the Wnt pathway may be a driver in this disease. Here, we present a comprehensive "multi-omics" study where the genome, transcriptome, and methylome of SPNs were analyzed. We found that SPNs are characterized by a low-complexity genome where somatic mutations in CTNNB1, present in 100% of the cases, are the only actionable genomic lesions. Compared to more common subtypes of pancreatic tumors (adenocarcinomas and pancreatic neuroendocrine tumors), SPNs show high expression levels of genes belonging to the Wnt pathway. Their methylome was consistent with an epithelial cell origin and a general upregulation of Wnt pathway genes. Clinical studies to evaluate the exquisite sensitivity of SPNs to inhibitors of the Wnt pathway are warranted.

Purpose: The Centers for Medicare & Medicaid Services (CMS) identifies suboptimal management of treatment toxicities as a care gap and proposes the measurement of hospital performance on the basis of emergency department visits for 10 common symptoms. Current management strategies do not address symptom co-occurrence. Methods: We evaluated symptom co-occurrence in three patient cohorts that presented to a cancer hospital urgent care center in 2016. We examined both the CMS-identified symptoms and an expanded clinician-identified set defined as symptoms that could be safely managed in the outpatient setting if identified early and managed proactively. The cohorts included patients who presented with a CMS-defined symptom within 30 days of treatment, patients who presented within 30 days of treatment with a symptom from the expanded set, and patients who presented with a symptom from the expanded set within 30 days of treatment start. Symptom co-occurrence was measured by Jaccard index. A community detection algorithm was used to identify symptom clusters on the basis of a random walk process, and network visualizations were used to illustrate symptom dynamics. Results: There were 6,429 presentations in the CMS symptom-defined cohort. The network analysis identified two distinct symptom clusters centered around pain and fever. In the expanded symptom cohort, there were 5,731 visits and six symptom clusters centered around fever, emesis/nausea, fatigue, deep vein thrombosis, pain, and ascites. For patients who newly initiated treatment, there were 1,154 visits and four symptom clusters centered around fever, nausea/emesis, fatigue, and deep vein thrombosis. Conclusion: Uncontrolled symptoms are associated with unplanned acute care. Recognition of the complexity of symptom co-occurrence can drive improved management strategies.

Importance Treatment of locally advanced rectal (LARC) cancer involves chemoradiation, surgery, and chemotherapy. The concept of total neoadjuvant therapy (TNT), in which chemoradiation and chemotherapy are administered prior to surgery, has been developed to optimize delivery of effective systemic therapy aimed at micrometastases. Objective To compare the traditional approach of preoperative chemoradiation (chemoRT) followed by postoperative adjuvant chemotherapy with the more recent TNT approach for LARC. Design, Setting, and Participants A retrospective cohort analysis using Memorial Sloan Kettering Cancer Center (MSK) records from 2009 to 2015 was carried out. A total of 811 patients who presented with LARC (T3/4 or node-positive) were identified. Exposures Of the 811 patients, 320 received chemoRT with planned adjuvant chemotherapy and 308 received TNT (induction fluorouracil- and oxaliplatin-based chemotherapy followed by chemoRT). Main Outcomes and Measures Treatment and outcome data for the 2 cohorts were compared. Dosing and completion of prescribed chemotherapy were assessed on the subset of patients who received all therapy at MSK. Results Of the 628 patients overall, 373 (59%) were men and 255 (41%) were women, with a mean (SD) age of 56.7 (12.9) years. Of the 308 patients in the TNT cohort, 181 (49%) were men and 127 (49%) were women. Of the 320 patients in the chemoRT with planned adjuvant chemotherapy cohort, 192 (60%) were men and 128 (40%) were women. Patients in the TNT cohort received greater percentages of the planned oxaliplatin and fluorouracil prescribed dose than those in the chemoRT with planned adjuvant chemotherapy cohort. The complete response (CR) rate, including both pathologic CR (pCR) in those who underwent surgery and sustained clinical CR (cCR) for at least 12 months posttreatment in those who did not undergo surgery, was 36% in the TNT cohort compared with 21% in the chemoRT with planned adjuvant chemotherapy cohort. Conclusions and Relevance Our findings provide additional support for the National Comprehensive Cancer Network (NCCN) guidelines that categorize TNT as a viable treatment strategy for rectal cancer. Our data suggest that TNT facilitates delivery of planned systemic therapy. Long-term follow-up will determine if this finding translates into improved survival. In addition, given its high CR rate, TNT may facilitate nonoperative treatment strategies aimed at organ preservation.

BACKGROUND Pancreatic adenocarcinoma (PAC) is part of several cancer predisposition syndromes; however, indications for genetic counseling/testing are not well-defined. In the current study, the authors sought to determine mutation prevalence and characteristics that are predictive of an inherited predisposition for PAC. METHODS A total of 175 consecutive patients with PAC who underwent clinical genetics assessment at Memorial Sloan Kettering Cancer Center between 2011 and 2014 were identified. Clinical data, family history, and germline results were evaluated. RESULTS Among 159 patients with PAC who pursued genetic testing, 24 pathogenic mutations were identified (15.1%; 95% confidence interval, 9.5%-20.7%), including BRCA2 (13 mutations), BRCA1 (4 mutations), p16 (2 mutations), PALB2 (1 mutation), and Lynch syndrome (4 mutations). BRCA1/BRCA2 prevalence was 13.7% in Ashkenazi Jewish (AJ) patients (95 patients) and 7.1% in non-AJ patients (56 patients). In AJ patients with a strong, weak, or absent family history of BRCA-associated cancers, the mutation prevalence was 16.7%, 15.8%, and 7.4%, respectively. The mean age at the time of diagnosis in all mutation carriers was 58.5 years (range, 45-75 years) compared with 64 years (range, 27-87 years) in those not carrying a mutation (P = .02). Although BRCA2 was the most common mutation identified, no patients with early-onset PAC (diagnosed at age ≤ 50 years) harbored a BRCA2 mutation and the mean age at diagnosis in BRCA2 carriers was equivalent to that of individuals who were not mutation carriers (P = .34). Mutation prevalence in patients with early-onset disease (21 patients) was 28.6%, including BRCA1 (2 mutations), p16 (2 mutations), MSH2 (1 mutation), and MLH1 (1 mutation). CONCLUSIONS Mutations in BRCA2 account for > 50% of patients with PAC with an identified susceptibility syndrome. AJ patients were found to have high BRCA1/BRCA2 prevalence regardless of personal/family history, suggesting that ancestry alone indicates a need for genetic evaluation. With the exception of BRCA2-associated PAC, an inherited predisposition for PAC is associated with an earlier age at PAC diagnosis, suggesting that this subset of patients may also represent a population warranting further evaluation. Cancer 2015;121:4382–8. © 2015 American Cancer Society.

Purpose: Apparent diffusion coefficient (ADC), derived from diffusion-weighted magnetic resonance images (DW-MRI), measures the motion of water molecules in vivo and can be used to quantify tumor response to therapy. The accurate measurement of ADC can be adversely affected by organ motion and imaging artifacts. In this paper, the authors’ goal was to develop an automated method for reducing artifacts and thereby improve the accuracy of ADC measurements in moving organs such as liver. Methods: The authors developed a novel method of computing ADC with fewer artifacts, through simultaneous image segmentation and iterative registration (SSIR) of multiple b-value DW-MRI. The authors’ approach reduces artifacts by automatically finding the best possible alignment between the individual b-value images and a reference DW image using a sequence of transformations. It selects such a sequence by an iterative choice of b-value DW images based on the accuracy of their alignment with the reference DW image. The authors’ approach quantifies the accuracy of alignment between a pair of images using modified Hausdroff distance computed between the structures of interest. The structures of interest are identified by a user through strokes drawn in one or more slices in the reference DW image, which are then volumetrically segmented using GrowCut. The same structures are segmented in the remaining b-value images by transforming the user-drawn strokes through registration. The ADC values are computed from all the aligned b-value images. The images are aligned by using affine registration followed by deformable B-spline registration with cubic B-spline resampling. Results: The authors compared the results of ADC computed using their approach with ADC computed (a) without registration and (b) with basic affine registration of all b-value images to a chosen reference. The authors’ approach was the most effective in reducing artifacts compared to the other two methods. It resulted in a mean artifact ratio (fraction of voxels in a structure with negative ADC over total number of voxels in the structure) of 2.7% versus 5.4% for affine registration and 32% for no registration for >200 tumors. The authors’ approach also resulted in the lowest median standard deviation in the computed mean ADC for all tumors [0.05,0.09,0.07,0.58] compared to those from affine image registration [0.02, 0.14, 0.58, 0.79] and no image registration [0.64, 0.83, 0.83, 1.09] on tests where random displacement [8,10,12,16] pixels were introduced in multiple trials in the b-value images. Conclusions: The authors developed a novel approach for reducing artifacts in ADC maps through simultaneous registration and segmentation of multiple b-value DW images. The authors’ method explicitly employs a registration quality metric to align images. When compared to basic affine and no image registrations, the authors’ approach produces registrations of greater accuracy with lowest artifact ratio and median standard deviation of the computed mean ADC values for a wide range of displacements.

Brain metastases (BM) from pancreatic ductal adenocarcinoma (PDAC) are an infrequent event. We identified n=25 pts from MSKCC with BM and PDAC; median age 58years. Median time to the development of BM was 17 months (range 0 to 79). Median overall survival from the time of BM development was 1.5 months (1 to 31). Six patients had germline testing; with BRCA1 (n=1) or BRCA2 (n=2) alterations detected. Seven patients had molecular profiling with KRAS, TP53 and MYC amplification most frequent. BACKGROUND: The purpose of this study was to assess clinical characteristics of patients with metastatic pancreas ductal adenocarcinoma (PDAC) and brain metastases (BM) as well as the somatic and germline molecular profiles where performed. PATIENTS AND METHODS: Patients with PDAC and BM between January 1990 and January 2016 were identified. Molecular characteristics of somatic and germline testing where performed in the subset of patients that had provided informed consent. Somatic alterations were assessed by either MSK-IMPACT testing (>340 key cancer genes) or Sequenom testing (8 gene panel). Overall survival (OS) was calculated from date of diagnosis to either date of last follow up or death. Survival post brain metastasis was calculated from date of diagnosis of brain metastases by radiology or pathology to either date of last follow up or death. RESULTS: From a total of 5,824 patients with PDAC identified from January 2000 to January 2016, twenty-five (0.4%) patients had brain metastases. Median age of PDAC diagnosis was 58 years. Median time to the development of BM from initial PDAC diagnosis was 17 months (0 to 79). Median OS following BM diagnosis was 1.5 months (range 1 to 31). OS for patients that had craniotomy (n=4) was 11 months (1 to 31 months) with two long term survivors; 21 and 31 months respectively. Four patients had leptomeningeal disease. 6/25 pts had germline testing, 3 had BRCA mutations; BRCA1 (n=2) and BRCA2 (n=1). Somatic profiling identified KRAS mutations in 100%; G12D (n=4), G12V (n=2) and Q61K (n=1). CONCLUSION: BM from PDAC is a rare event. We identified a speculitive association of germline BRCA1/2 alterations with BM in PDAC, which requires corroboration. Survival post BM development is poor with prolonged survival in selected patients via a multidisciplinary approach.

The role for neoadjuvant systemic therapy in resectable pancreas adenocarcinoma remains undefined.We evaluated the efficacy of gemcitabine and oxaliplatin administered as preoperative therapy in patients with resectable pancreas adenocarcinoma.Eligible patients were screened using computed tomography-pancreas angiography, laparoscopy, endoscopic ultrasonography, and fine-needle aspiration cytology to identify 38 patients who received 4 cycles of neoadjuvant gemcitabine 1000 mg/m intravenously over 100 minutes and oxaliplatin 80 mg/m intravenously over 2 hours, every 2 weeks. Patients whose tumors remained resectable at restaging proceeded to operation and subsequently received 5 cycles of adjuvant gemcitabine (1000 mg/m intravenously over 30 minutes days 1, 8, and 15 every 4 weeks). The primary endpoint was 18-month overall survival and secondary endpoints included radiological, tumor marker and pathological response to neoadjuvant therapy, time to recurrence, patterns of failure, and feasibility of obtaining preoperative core biopsies.Thirty-five of 38 patients (92%) completed neoadjuvant therapy. Twenty-seven patients underwent tumor resection (resectability rate 71%), of which 26 initiated adjuvant therapy for a total of 23 patients (60.5%) who completed all planned therapy. The 18-month survival was 63% (24 patients alive). The median overall survival for all 38 patients was 27.2 months (95% confidence interval: 17-NA) and the median disease-specific survival was 30.6 months (95% confidence interval: 19-NA).This study met its endpoint and provided a signal suggesting that exploration of neoadjuvant systemic therapy is worthy of further investigation in resectable pancreas adenocarcinoma. Improved patient selection and more active systemic regimens are key. Clinical trials identification: NCT00536874.

No systemic agents that are known to be effective for the treatment of solid-pseudopapillary neoplasm (SPN) are available. We report the prolonged and sustained control of metastatic pancreatic SPN to the liver using hepatic arterial embolization (HAE), where a total of 13 HAE sessions were performed over a 6-year period.

High-grade neuroendocrine carcinomas (HGNECs) of the colon and rectum are rare, constituting less than 1 % of colorectal cancers. The purpose of this study was to identify the natural history and oncologic outcomes of this disease, describe the use of surgery, and determine the clinical and pathological factors associated with outcomes.Following Institutional Review Board approval, patients with HGNEC were identified from our institutional database. Patient charts and pathology reports were analyzed retrospectively for clinical and pathological factors.A total of 126 patients with a median follow-up of 9 months were identified. Median survival was 13.2 months, and 85 (67 %) patients had metastatic disease at diagnosis. Three-year overall survival (OS) was 5 and 18 % for patients with and without metastatic disease, respectively. Factors associated with improved OS on multivariable analysis were absence of metastatic disease and presence of an adenocarcinoma component within the tumor. In patients with metastatic disease, response to chemotherapy was the only factor associated with survival. In patients with localized disease, an adenocarcinoma component within the tumor was the only factor associated with survival. Resection of tumor was not associated with survival in either localized or metastatic disease.High-grade colorectal NECs are extremely aggressive tumors with poor prognosis. Patients appear to have a marginally better prognosis if they present without metastatic disease, have an adenocarcinoma component within their tumor, or respond to chemotherapy. Surgery, particularly in the presence of metastatic disease, may not offer a survival benefit for the majority of patients.

6535 Background: Monitoring and managing patient reported outcomes (PROs) has been recommended for oncology patients on active treatment but can be time and resource intensive. Identifying patients likely to benefit and the optimal frequency of PRO capture is still under investigation. We tested the feasibility of monitoring patients who are high-risk risk for acute care with daily PROs. Methods: Using data from our institution, we developed a model that employs over 400 clinical variables to calculate a patient’s risk of an emergency room visit within 6 months following the onset of treatment. From October 15, 2018 to January 23, 2019, we enrolled patients identified as high risk through a technology-enabled program to monitor and manage those patients’ symptoms. Enrolled patients entered PRO assessments daily via an online portal. Symptoms were monitored and managed by a centralized clinical team. Tiered notifications informed the team of concerning or escalating symptoms. We assessed how frequently patients completed symptom assessments and the frequency of symptom notifications. Results: During the pilot, 28 patients were identified as high risk and enrolled in the program (median age 65; 64% percent female). Disease types were: 15 (54%) thoracic, 7 (25%) gynecologic, 6 (21%) gastrointestinal. Median time in the program was 50 (6-98) days. Patients completed 840 of 1,350 assessments (62%). There were 328 assessments that triggered moderate alerts (39%) and 220 that triggered severe alerts (26%). The table describes the prevalence of symptoms at the patient-level. Conclusions: A model can be employed to identify high-risk patients in collaboration with clinicians. Our adherence rate with a daily symptom assessment was similar to those found in studies of less frequent PRO capture. Future work will expand to a larger patient population with other cancer types, evaluate impact on outcomes, and assess optimal frequency for PRO collection and alert thresholds. [Table: see text]

6554 Background: Acute care accounts for half of cancer expenditures and is a measure of poor quality care. Identifying patients at high risk for emergency department (ED) visits enables institutions to target resources to those most likely to benefit. Risk stratification models developed to date have not been meaningfully employed in oncology, and there is a need for clinically relevant models to improve patient care. Methods: We established and applied a predictive framework for clinical use with attention to modeling technique, clinician feedback, and application metrics. The model employs electronic health record data from initial visit to first antineoplastic administration for patients at our institution from January 2014 to June 2017. The binary dependent variable is occurrence of an ED visit within the first 6 months of treatment. The final regularized multivariable logistic regression model was chosen based on clinical and statistical significance. In order to accommodate for the needs to the program, parameter selection and model calibration were optimized to suit the positive predictive value of the top 25% of observations as ranked by model-determined risk. Results: There are 5,752 antineoplastic administration starts in our training set, and 1,457 in our test set. The positive predictive value of this model for the top 25% riskiest new start antineoplastic patients is 0.53. From over 1,400 data features, the model was refined to include 400 clinically relevant ones spanning demographics, pathology, clinician notes, labs, medications, and psychosocial information. At the patient level, specific features determining risk are surfaced in a web application, RiskExplorer, to enable clinician review of individual patient risk. This physician facing application provides the individual risk score for the patient as well as their quartile of risk when compared to the population of new start antineoplastic patients. For the top quartile of patients, the risk for an ED visit within the first 6 months of treatment is greater than or equal to 49%. Conclusions: We have constructed a framework to build a clinically relevant risk model. We are now piloting it to identify those likely to benefit from a home-based, digital symptom management intervention.

4112 Background: Adrenocortical carcinomas (ACC) are rare and aggressive. Treatment options are limited and marked by poor efficacy and substantial toxicity. In this phase II single-center study, the efficacy and safety of pembrolizumab was assessed in patients (pts) with advanced ACC. Methods: Enrolled pts were aged ≥18 y with advanced ACC, ECOG ≤ 1, available tumor samples for biomarker analysis. Pts received pembrolizumab 200 mg Q3W for 2 y or until disease progression, intolerable toxicity, physician/patient decision to stop treatment. Imaging was performed every 9 wks. Tumor PD-L1 positivity (modified proportion score ≥ 1% or presence of stromal interface) was evaluated. Primary endpoint was ORR (by RECIST v1.1). Secondary endpoints included DOR, PFS, OS, safety. Somatic and germline next-generation sequencing was performed. Results: 39 pts were treated. Median age 62 (range, 19-87), 28% ECOG 0, 72% received ≥ 1 therapy. In available samples to date, 7/31 (23%) PD-L1+. At time of analysis, median follow-up among survivors was 17.8 mo (range, 5.4-34.7). ORR was 23.1% (95% CI, 11.1-39.3); 0 CR, 9PR. Seven pts (17.9%) had SD as best response. Among the 9 PRs, median time to PR was 4.1 mo (range, 1.7-10.5) and median DOR was not reached (95% CI, 4.1-not reached). Three pts achieving PR have completed 2 y of treatment with ongoing response noted. Tumor PD-L1 status is currently available in 6 pts with PR, 2/6 (33%) PD-L1+. Median PFS was 2.1 mo (95% CI, 2.0-10.7). Median OS was 24.9 mo (95% CI, 4.2-not reached); 2-year OS rate was 50% (95% CI, 36-69%). In the 34 tested tumors, germline testing identified 2 PR pts with Lynch syndrome; the remaining 7 PRs were MSS. Median tumor mutation burden for all PRs was 4.1 mutations/megabase (range, 0-31.5). There was no significant relationship between somatic alterations and response to treatment. Grade 3/4 treatment-related AEs occurred in 7/39 (17.9%) pts. Two pts discontinued therapy due to AEs; both pts achieved PRs and continue to respond. All pts with PRs had LFT elevation ≥ grade 2. Conclusions: Pembrolizumab demonstrated antitumor activity and was well tolerated in advanced ACC. Durable responses were noted. Complete evaluation of tumor PD-L1 and microsatellite status will be reported at the meeting. Clinical trial information: NCT02673333.

321 Background: G3 NENs are aggressive, and optimal systemic treatment is unclear. Temozolomide (TEM)-based regimens have been used to treat grade 1-2 NETs, but their efficacy in G3 NENs (Ki-67 > 20%) remains undetermined. Aims: To assess the clinical efficacy of TEM-containing regimens in advanced grade III gastroenteropancreatic NENs (GEPNENs). Methods: A multicentre retrospective review (2008-2017) of patients with metastatic/unresectable G3 GEPNENs who received a TEM-containing regimen. The primary endpoint was time to treatment failure (TTF). Radiologic response was extracted from local reports without formal RECIST criteria. Results: 118 patients in six centers were included (median age 55, 65% male, 15% functional, 75% pancreatic NEN). 57% were well-differentiated, 35% poorly-differentiated, and 18% unknown based on local pathology reports. The regimen used was CAPTEM in 93% and TEM in 7%. Best radiological responses were: complete response (1%), partial response (39%), stable disease (22%), progressive disease (31%), unknown (7%) not by RECIST. Median TTF was 150 days and median overall survival (OS) 18.0 months. Fifteen patients (14%) required dose reductions/discontinuation due to adverse events. TTF was shorter for patients on TEM alone (p = 0.02, Table 1). Well-differentiated NENs had better response rate (52% vs 26%, p = 0.02) and overall survival (30.1 vs 12.0 mo, p = 0.008) compared to poorly-differentiated NEN. Conclusions: This is the largest TEM treatment series in G3 NEN, involving collaboration of several major North American NET centers. 40% of patients showed some degree of response, and treatment was generally well-tolerated. TEM-based regimens should be considered a viable treatment option in this setting. Prospective confirmatory trials (such as EA2142) may face difficulties in accrual due to disease rarity. [Table: see text]

363 Background: Immune checkpoint inhibitor (CPI) efficacy has not been established in extrapulmonary poorly-differentiated neuroendocrine carcinomas (EP-PDNECs). In small cell lung cancer, promising antitumor activity of CPI led to accelerated approval of nivolumab in 8/2018. We investigated the efficacy and safety of pembrolizumab (PEM)-based therapy in biomarker-unselected EP-PDNECs. Methods: Open label, multicenter, phase 2 study of PEM-based therapy in patients (pts) with EP-PDNECs, excluding Merkel cell carcinoma and well differentiated grade 3 NET, with progression on first-line systemic therapy, ECOG 0-1, and adequate hepatic and renal function. Enrollment via an adaptive Simon’s 2-stage design. Plan for 14 pts treated with PEM alone (Part A Stage 1) 200 mg IV every 3 weeks. If > 2 of 14 pts respond by week 18, then 21 additional pts enroll in Part A Stage 2, corresponding to H0 10% vs. H1 26% response rate (RR) at type I error 0.05 with power 80%. Otherwise study proceeds to Part B: PEM plus chemotherapy (dealer's choice of weekly irinotecan or paclitaxel). Primary endpoint is objective RR (ORR) by RECIST 1.1. Secondary endpoints include safety, overall survival, and progression-free survival (PFS). Serial blood samples and baseline tumor biopsies required in all pts for future biomarker studies. Results: Preliminary data from Part A Stage 1 are available. Of 14 pts enrolled, male/female 9/5; median age 63; 1 large cell, 11 small cell, 2 NOS. Primary site of disease: GI 43%, GU 29%, and other 29%. Median Ki67 80% (available for 9 pts). Best response: CR (1), PR (0), SD (2), PD (10), unevaluable (1; early death from sepsis) for ORR 7%. Median PFS was 58 days. Six (43%) pts went off study for early PD or clinical deterioration before first scheduled scan at 9 weeks. PEM was well tolerated with no grade 3-5 AEs attributed to therapy. At last follow up, 9 (64%) pts were alive with 1 pt still on treatment after 19 cycles. Conclusions: PEM monotherapy was not effective in this pretreated, biomarker-unselected population of EP-PDNECs arising in different organs. Part B (PEM plus chemotherapy) enrollment is ongoing. Biomarker studies are planned (Parts A/B). Clinical trial information: NCT03136055.

394 Background: There are several therapies now available for panNETs, raising concern that treatment-related toxicity, namely myelosuppression, may preclude use of all treatments. AA are an effective therapy in panNETs, but can cause irreversible bone marrow suppression. For this reason, our institutional approach is to treat patients (pts) with AA for a defined time period of up to 12 months prior to a treatment holiday; anecdotally, we observed continued disease shrinkage or stability after drug cessation. We aimed to identify and characterize these exceptional responders (ER). Methods: We retrospectively identified pts with well differentiated panNETs treated at MSKCC with AA from 2007 to present, and identified those that were placed on a treatment holiday after response to therapy. Pts with continued disease response by radiographic report (tumor shrinkage or stable disease) for greater than 9 months while on a treatment holiday were considered exceptional responders (ER). Results: 124 patients were evaluable, with 104 pts (84%) receiving temozolomide and 18 pts (15%) receiving dacarbazine. 40 (32%), 57 (46%), and 22 (18%) pts had low, intermediate and high-grade tumors, respectively. 80 pts (65%) achieved a response to AA and 21 pts (17%) responded to therapy initially with subsequent disease progression. In total, 58 pts (47%) entered a treatment holiday; 39/58 pts (67%) were ER and 16/58 pts (28%) had disease growth in < 9 months. No significant differences in tumor grade ( P = 0.85) were seen between ER and those who had disease growth in < 9 months. Next generation sequencing results were available in 17 tumors of pts classified as ER; 14/17 (82%) with alterations in chromatin remodeling genes ( MEN1/DAXX/ATRX). Upon disease progression, 11 ER were reintroduced to alkylating agents; 9/11 (82%) had progressive disease or death after reintroduction of drug, and 2/11 (18%) responded and went on a second treatment holiday. Conclusions: We observed durable responses to AA after cessation of therapy. Many of the ER tumors harbored alterations in chromatin remodeling genes. Reintroduction of alkylating agents upon disease progression was generally ineffective.

690 Background: Colorectal high-grade neuroendocrine carcinomas (HGNEC) are aggressive tumors; treatment options consist of cis/carboplatin and 5FU-based chemotherapy. To date, choice of systemic therapy and sequencing of these drugs remains poorly understood. We examined clinical and molecular characteristics of colorectal HGNEC to better define predictors of response to cis/carboplatin and 5FU-based treatment. Methods: Patients (pts) with colorectal HGNEC treated at MSKCC from 1990-2018 were identified. MANEC (mixed adeno-neuroendocrine carcinoma) were excluded. Demographics, response to first- and second-line chemotherapy (by radiology report), outcomes, and molecular data (next-generation sequencing of tumor tissue), were collected. Results: 65 pts (mean age 58, 52% male) were identified, 13 (20%) with small cell carcinomas. 52 (79%) metastatic, 13 (20%) locally advanced. 27 (42%) received surgery and 11 (17%) received radiation. 56 pts received cis/carboplatin-based therapy, partial response (PR) in 18 (32%), stable disease (SD) in 4 (7%), and progressive disease (PD) in 31 (55%); 3 (5%) did not tolerate therapy. 28 pts received 5FU-based therapy, 13 PR in (46%), SD in 6 (21%), and PD in 7 (25%); 2 (7%) did not tolerate therapy. Median overall survival was 11.4 months. 21/65 (32%) pts underwent molecular sequencing of tumor; the most common alterations were KRAS 11 (52%), TP53 13 (62%), BRAF 7 (33%), APC 8 (38%), RB1 7 (33%). Most tumors (13/21, 62%) harbored alterations in genes traditionally altered in colorectal adenocarcinoma (KRAS/BRAF/APC) and in HGNEC (TP53/RB1). There was no significant difference in response to cis/carboplatin or 5FU-based chemotherapy based on location of the primary tumor (right vs. left) (p = 0.69/0.85), histologic features of the disease (p = 0.71/0.87), and for response to cis/carboplatin by molecular alterations in KRAS (p = 0.94), BRAF (p = 0.24), APC (p = 0.28), TP53 (p = 0.58), or RB1 (p = 0.28). Conclusions: Colorectal HGNEC are highly aggressive and more effective therapies are desperately needed. In this series, OS was poor. Clinical and molecular characteristics failed to predict response to cis/carboplatin and 5FU-based chemotherapy.

144 Background: Acute care accounts for half of cancer expenditures and is a measure of poor quality care. Identifying patients at high risk for ED visits enables institutions to target symptom management resources to those most likely to benefit. Risk stratification models developed to date have not been meaningfully employed in oncology, and there is a need for clinically relevant models to improve patient care. Methods: We established a predictive analytics framework for clinical use with attention to the modeling technique, clinician feedback, and application metrics. The model employs EHR data from initial visit to first antineoplastic administration for new patients at our institution from January 2014 to June 2017. The binary dependent variable is occurrence of an ED visit within the first 6 months of treatment. From over 1,400 data features, the model was refined to include 400 clinically relevant ones spanning demographics, pathology, clinician notes, labs, medications, and psychosocial information. Clinician review was performed to confirm EHR data input validity. The final regularized multivariate logistic regression model was chosen based on clinical and statistical significance. Parameter selection and model evaluation utilized the positive predictive value for the top 25% of observations ranked by model-determined risk. The final model was evaluated using a test set containing 20% of randomly held out data. The model was calibrated based on a 5-fold cross-validation scheme over the training set. Results: There are 5,752 antineoplastic starts in our training set, and 1,457 in our test set. The positive predictive value of this model for the top 25% riskiest new start antineoplastic patients is 0.53. The 400 clinically relevant features draw from multiple areas in the EHR. For example, those features found to increase risk include: combination chemotherapy, low albumin, social work needs, and opioid use, whereas those found to decrease risk include stage 1 disease, never smoker status, and oral antineoplastic therapy. Conclusions: We have constructed a framework to build a clinically relevant model. We are now piloting it to identify those likely to benefit from a home-based, digital symptom management intervention.

Cardiopulmonary resuscitation (CPR) is an important advance directive (AD) topic in patients with progressive cancer; however such discussions are challenging.This study investigates whether video educational information about CPR engenders broader advance care planning (ACP) discourse.Patients with progressive pancreas or hepatobiliary cancer were randomized to an educational CPR video or a similar CPR narrative. The primary end-point was the difference in ACP documentation one month posttest between arms. Secondary end-points included study impressions; pre- and post-intervention knowledge of and preferences for CPR and mechanical ventilation; and longitudinal patient outcomes.Fifty-six subjects were consented and analyzed. Rates of ACP documentation (either formal ADs or documented discussions) were 40% in the video arm (12/30) compared to 15% in the narrative arm (4/26), OR=3.6 [95% CI: 0.9-18.0], p=0.07. Post-intervention knowledge was higher in both arms. Posttest, preferences for CPR had changed in the video arm but not in the narrative arm. Preferences regarding mechanical ventilation did not change in either arm. The majority of subjects in both arms reported the information as helpful and comfortable to discuss, and they recommended it to others. More deaths occurred in the video arm compared to the narrative arm, and more subjects died in hospice settings in the video arm.This pilot randomized trial addressing downstream ACP effects of video versus narrative decision tools demonstrated a trend towards more ACP documentation in video subjects. This trend, as well as other video effects, is the subject of ongoing study.

A 68-year-old woman presented with bilateral visual loss as the only clinical manifestation of an occult pancreatic nonsecretory neuroendocrine tumor (NET). The suspected diagnosis of paraneoplastic optic neuropathy was confirmed using immunofluorescence assays to demonstrate the presence of antibodies in the patient's serum that reacted with antigen(s) in the optic nerve and in the pancreatic NET hepatic metastasis. Treatment of the underlying cancer was followed by marked improvement in visual function.

Purpose To describe the clinical and CT imaging features of goblet cell carcinoid (GCC) neoplasm of the appendix. Methods and materials A computer search of pathology and radiology records over a 19-year period at our two institutions was performed using the search string “goblet”. In the patients with appendiceal GCC neoplasms who had abdominopelvic CT, imaging findings were categorized, blinded to gross and surgical description, as: “Appendicitis”, “Prominent appendix without peri-appendiceal infiltration”, “Mass” or “Normal appendix”. The CT appearance was correlated with an accepted pathological classification of: low grade GCC, signet ring cell adenocarcinoma ex, and poorly differentiated adenocarcinoma ex GCC group. Results Twenty-seven patients (age range, 28–80 years; mean age, 52 years; 15 female, 12 male) with pathology-proven appendiceal GCC neoplasm had CT scans that were reviewed. Patients presented with acute appendicitis ( n = 12), abdominal pain not typical for appendicitis ( n = 14) and incidental finding ( n = 1). CT imaging showed 9 Appendicitis, 9 Prominent appendices without peri-appendiceal infiltration, 7 Masses and 2 Normal appendices. Appendicitis (8/9) usually correlated with typical low grade GCC on pathology. In contrast, the majority of Masses and Prominent Appendices without peri-appendiceal infiltration were pathologically confirmed to be signet ring cell adenocarcinoma ex GCC. Poorly differentiated adenocarcinoma ex GCC was seen in only a small minority of patients. Hyperattenuation of the appendiceal neoplasm was seen in a majority of cases. Conclusions GCC neoplasm of the appendix should be considered in the differential diagnosis in patients with primary appendiceal malignancy. Our cases demonstrated close correlation between our predefined CT pattern and the pathological classification.

To assess local control, survival and conversion to resectability among locally advanced pancreatic cancer (LAPC) patients treated with induction chemotherapy (ICT) followed by chemoradiotherapy treatment using intensity-modulated radiation therapy (IMRT).Between 2007 and 2012, 134 LAPC patients were treated with ICT followed by IMRT. After chemoradiotherapy, 40 patients received maintenance chemotherapy.With a median follow-up of 20 months, median overall survival (OS) was 23 months. One- and two-year OS was 85% and 47%, respectively. On multivariate analysis, progression of disease after IMRT was associated with worse OS. Cumulative incidence of local failure was 10% at one year and 36% at two years. Twenty-six patients (19%) underwent resection after chemoradiotherapy including 22 patients (85%) with negative margins. On multivariate analysis, response to IMRT was associated with surgery (p = .01). Acute grade 3-4 hematologic and non-hematologic toxicity rates were 26% and 4.5%, respectively.IMRT is safe in patients with LAPC. Patients with non-progressive LAPC after ICT and who received IMRT had high rates of local control and prolonged survival.

Purpose To assess the impact on acute toxicity of replacing 5-fluorouracil (5-FU) with capecitabine in definitive chemoradiation for patients with anal squamous cell carcinoma (ASCC). Methods and Materials We retrospectively reviewed the records of 107 consecutive patients with nonmetastatic ASCC treated with definitive chemoradiation from January 2009 to May 2014. In 2011, based on the noninferiority of capecitabine versus 5-FU, our institutional practice shifted to use capecitabine instead of 5-FU for ASCC. Of 107 patients, 63 were treated with infusional 5-FU (1000 mg/m 2 /day for 4 days) and mitomycin C (MMC) (10 mg/m 2 ) during weeks 1 and 5, and 44 patients were treated with capecitabine (825 mg/m 2 twice daily) Monday through Friday throughout radiation therapy (RT) and MMC (10 mg/m 2 ) during weeks 1 and 5. The incidence of grade 3 to 4 acute toxicity was compared between the 2 groups. Results The median age at diagnosis was 59 years, and 78 patients (73%) were female. The patient characteristics were similar between the 2 treatment groups. All patients in both groups were treated with intensity modulated RT (median dose, 56 Gy). In the 5-FU group, 52% experienced grade 3 to 4 neutropenia compared with 20% in the capecitabine group ( P =.001). Treatment breaks resulting from toxicity, primarily related to grade 3+ hematologic toxicity, were necessary for 42% of patients treated with 5-FU versus 16% of those treated with capecitabine ( P =.006). Conclusions Pelvic radiation therapy with MMC plus capecitabine was well tolerated and appeared to have less grade 3+ acute hematologic toxicity and fewer treatment interruptions than in a population of ASCC patients undergoing definitive chemoradiation with MMC and 5-FU.

Cetuximab (C), alone or with irinotecan, demonstrates activity in irinotecan-refractory colorectal cancer (CRC). Activity of 5-fluorouracil (5-FU), leucovorin (L), and bevacizumab (B), and preliminary data of cetuximab + bevacizumab, and toxicity profiles suggests that FOLF-CB (5-FU, L, C+B) may have activity with a favorable toxicity profile as first-line therapy.Eligible patients were randomized at registration to either arm A (mFOLFOX6-B) (modified, 5-FU. L (folinic acid), oxaliplatin (O) + bevacizumab), administered days 1 and 15 of each 28-day cycle as bevacizumab 5 mg/kg, oxaliplatin 85 mg/m(2), leucovorin 400 mg/m(2), and 5-FU 400 mg/m(2) then 1200 mg/m(2)/day for 48 hours, or arm B (FOLF-CB), which included bevacizumab, leucovorin, and 5-FU as in arm A and cetuximab 400 mg/m(2) day 1 cycle 1; all other weekly cetuximab doses were 250 mg/m(2).Two hundred forty-seven patients (arm A/arm B 124/123) were enrolled, and 239 were treated (118/121). Twelve-month progression-free survival (PFS) was 45%/32%, objective response rates (ORR) (complete response [CR] + partial response [PR]) were 52%/41%, disease control rates (CR+PR+stable disease [SD]) were 87%/83%, and median overall survival (OS) was 21/19.5 months, respectively. Grade 3-4 neutropenia was higher in arm A (28%/7%), as was grade 3 fatigue (12%/3%), and grade 3 neuropathy (11%/1%), whereas acneiform rash was confined to arm B. Retrospective analysis of KRAS mutational status did not demonstrate KRAS as a meaningful determinant of activity, except in arm B patients with KRAS-mutated tumors, which resulted in inferior PFS. Patient satisfaction favored the control (mFOLFOX6-B).FOLF-CB was not superior to mFOLFOX6-B in terms of 12-month PFS and ORR, and was not more acceptable to patients. This trial supports the conclusion of other recently reported trials that concurrent cetuximab+bevacizumab should not be routinely used in metastatic CRC.

178 Background: In 2017, 5,939 outpatients (5%) and 1,065 inpatients (9%) treated at MSK were non-English-speaking. Given the new and varying treatment paradigms combined with the need to consider clinical trial enrollment, there are increasingly complex communication needs. We sought to address the needs of this vulnerable patient population. Methods: Patient Family Advisory Council in collaboration with the Immigrant Health & Cancer Disparities Service conducted focus groups to determine needs of non-English speaking patients treated at MSKCC. Mass emails were sent out to the MSK attendings and nurses to recommend appropriate patients to participate. A coordinator called all those patients and invited them to the focus groups. A trained bilingual moderator led each 120-minute group discussion using an interview guide. Results: Three focus groups were conducted : 2 in Spanish and 1 in Russian. Questions related to communication, patient experience and support services were asked to the patient volunteers. In total, 10 patients and 4 caregivers participated. Across all 3 focus groups, three major themes emerged: (1) inadequate communication, participants brought up examples of communication breakdowns (2) access to resources and information, participants expressed that there was a substantial need for accessible, competent, and social support/navigation services and (3) MSK patient experience, participants discussed interactions with doctors, nurses, and staff. Conclusions: Non-English-speaking patients and family members not only deal have to deal with the fear of a cancer diagnosis but also lack the ability to comfortably communicate their concerns. Programs to empower patients and their families to ask for help and advocate for themselves are desperately needed. Additionally, navigation services are needed to provide assistance in making the appropriate linkages for the services required. This qualitative research at MSK is ongoing.

314 Background: Acute care accounts for half of cancer expenditures and is a measure of poor quality care. Identifying patients at high risk for ED visits enables institutions to target symptom management resources to those most likely to benefit. Risk stratification models developed to date have not been meaningfully employed in oncology, and there is a need for clinically relevant models to improve patient care. Methods: We established a predictive analytics framework for clinical use with attention to the modeling technique, clinician feedback, and application metrics. The model employs EHR data from initial visit to first antineoplastic administration for new patients at our institution from January 2014 to June 2017. The binary dependent variable is occurrence of an ED visit within the first 6 months of treatment. From over 1,400 data features, the model was refined to include 400 clinically relevant ones spanning demographics, pathology, clinician notes, labs, medications, and psychosocial information. Clinician review was performed to confirm EHR data input validity. The final regularized multivariate logistic regression model was chosen based on clinical and statistical significance. Parameter selection and model evaluation utilized the positive predictive value for the top 25% of observations ranked by model-determined risk. The final model was evaluated using a test set containing 20% of randomly held out data. The model was calibrated based on a 5-fold cross-validation scheme over the training set. Results: There are 5,752 antineoplastic starts in our training set, and 1,457 in our test set. The positive predictive value of this model for the top 25% riskiest new start antineoplastic patients is 0.53. The 400 clinically relevant features draw from multiple areas in the EHR. For example, those features found to increase risk include: combination chemotherapy, low albumin, social work needs, and opioid use, whereas those found to decrease risk include stage 1 disease, never smoker status, and oral antineoplastic therapy. Conclusions: We have constructed a framework to build a clinically relevant model. We are now piloting it to identify those likely to benefit from a home-based, digital symptom management intervention.

Lynch Syndrome (LS) is characterized by germline mutations in mismatch repair pathway genes. The increased cancer susceptibility risk in these patients and the association of these tumors with microsatellite instability (MSI) provides a transformative opportunity for early detection of cancer in these patients. The presence of microsatellite unstable DNA fragments or zygosity changes of the MMR mutations in the plasma of LS patients is predictive of the presence of cancer. Here we describe an approach to detect MSI from plasma cell-free DNA (cfDNA) of LS associated cancer patients. First, to determine the background rate of MSI in cfDNA of microsatellite stable patients, we screened plasma samples collected from a large number of advanced cancer patients using MSK-IMPACT, a custom sequencing assay targeting 468 cancer genes encompassing ~1.5 megabases. MSK-IMPACT is approved by the NYS Department of Health and authorized by the FDA for clinical testing, including MSI assessment based on more than 1,000 microsatellite regions covered by the assay, and has been used to profile more than 20,000 patients at our institution. We were also able to confirm the ability to detect MSI in cfDNA in a small cohort of plasma samples collected from patients with MSI High tumors, and determine thresholds to delineate MSI from MSS cases from plasma. Using MSK-IMPACT, we detected MSI in the plasma of four patients with advanced MSI-H cancer, including one with confirmed LS. Three of these were patients with confirmed MSI signature from tumor tissue (one prostate cancer and two colorectal cancers. The fourth patient with prostate cancer had no tissue available for sequencing, though MSI was independently confirmed. To improve the sensitivity to detect MSI in patients with earlier stage disease where the fraction of tumor-derived cfDNA is lower, we have developed a novel targeted panel with optimized sequencing process and informatics, incorporating a set of highly informative microsatellite regions as well as SNPs to assess tumor-specific zygosity changes in mismatch repairs pathway genes. Through this work, we demonstrate the ability of our assay to detect MSI in plasma cfDNA with high sensitivity. Citation Format: Preethi Srinivasan, Christina Tran, Jonathan Reichel, Juber Ahamad Patel, Maysun Hasan, Fanli Meng, Xiaohong Jing, Sumit Middha, Ahmet Zehir, Rona D. Yaeger, Diane Reidy, Dana Tsui, Zsofia Stadler, Michael F. Berger. Detecting MSI in plasma: Implications for early detection of lynch associated tumors [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 3656.