Randomized phase II study of neoadjuvant checkpoint blockade for surgically resectable undifferentiated pleomorphic sarcoma (UPS) and dedifferentiated liposarcoma (DDLPS): Survival results after 2 years of follow-up and intratumoral B-cell receptor (BCR) correlates

LBA11501

Background: We conducted a randomized, phase II non-comparative trial evaluating the efficacy of neoadjuvant ICB [nivolumab or ipilimumab/nivolumab] in patients (pts) with surgically resectable retroperitoneal DDLPS or extremity/truncal UPS treated with concurrent neoadjuvant radiation therapy (XRT, UPS only). Methods: As of February 28 2022, all pts have a minimum follow-up of 2 years from the start of ICB treatment. Progression-free survival (PFS) and overall survival (OS) were calculated using Kaplan Meier method. The association of pathologic response (percent hyalinization and viable tumor at surgery) with PFS and OS was assessed using Cox univariate models. Comparison of survival curves was done by log-rank method. The intratumoral BCR repertoire was evaluated by bulk tumor RNA sequencing with TRUST4 algorithm, on biopsy specimens collected at baseline. Description of the intratumoral BCR repertoire included diversity by inverse Simpson index, and clonal distribution by Gini coefficient. High and low categories were defined by median values. Results: At a median follow-up of 31 months (interquartile range [IQR]=27-43) since start of ICB treatment, the median PFS was not reached (NR) in UPS (IQR=19-NR) and 18 months for DDLPS (IQR=8-NR), with 13 pts experiencing relapse (2 UPS, 11 DDLPS) and 2 pts who had progressive metastatic disease on treatment (1 UPS, 1 DDLPS). Five pts died of disease relapse (1 UPS, 4 DDLPS) and the median OS was NR. There was no association between percent hyalinization at surgery and PFS (Hazard Ratio [HR]=0.98, p=0.12) or OS (HR=0.99, p=0.60) nor between percent viable tumor at surgery and PFS (HR=1.00, p=0.62) or OS (HR=1.00, p=0.67). There was no association between RECIST response and PFS (p=0.67) or OS (p=0.67). The median BCR heavy chain (IgH) clonal counts detected at baseline was 2,536 per sample (IQR=82-7,680), and the median BCR light chain (IgL) clonal count was 8,870 per sample (IQR=306-30,214). Pts with higher intratumoral BCR clonality and diversity at baseline tended to have longer PFS (Table). High BCR IgH clonality was significantly associated with OS (p=0.02) with consistent trends in each histotype (DDLPS: p=0.06; UPS: p=0.25). Conclusions: Survival results demonstrate efficacy of ICB with XRT in UPS but there is a crucial need to define better predictive markers of survival after neoadjuvant therapy. Further characterization of the BCR repertoire is ongoing and will be presented at the meeting. Clinical trial information: NCT03307616. [Table: see text]

A single-center experience of 98 patients (pts) with regionally metastatic Merkel cell carcinoma (MCC) of known (MCCKP) and unknown (MCCUP) primary at presentation

9576

Background: MCC is a rare skin cancer historically associated with poor survival rates and which is increasing in incidence. A small number of retrospective series suggest that MCCUP may be associated with better prognosis than MCCKP while others report worse outcomes. Recent advances in immunotherapy have changed the multimodal treatment landscape and outcomes of advanced MCC pts. We describe our experience with the management and outcomes of pts presenting with regional MCC metastasis of known and unknown primary origin. Methods: A retrospective review of pts with clinical regional disease at MCC diagnosis treated at our institution from 3/2003-3/2021 was performed. Clinicopathologic variables and outcomes were assessed. Overall survival (OS), recurrence-free survival (RFS) and progression-free survival (PFS) were estimated by the Kaplan Meier method. Results: Of 98 pts with regional disease on exam at presentation, 56 (57%) had MCCUP and 42 (43%) had MCCKP. Median follow-up from diagnosis to last follow-up or death was 33 months. Pts were generally older (MCCUP vs MCCKP: 68.7 vs 73.1 years), male (MCCUP vs MCCKP: 82% vs 74%) and Caucasian (MCCUP vs MCCKP: 84% vs 83%). Over half the pts had a history of another malignancy (MCCUP vs MCCKP: 52% vs 60%) with 9% and 14% being immunocompromised at diagnosis, respectively. After completion of staging workup, MCCUP pts had earlier stage disease at presentation compared with MCCKP pts (stage IIIA: 80% vs 55%, IIIB: 5% vs 31%, IV: 15% vs 14%, respectively). The cervical nodal basin was most commonly involved in MCCUP pts while regional disease was more varied in MCCKP pts (MCCUP vs MCCKP: cervical 54% vs 28%, axillary 15% vs 33%, inguinal 33% vs 3%, inguinal and pelvic 0% vs 11%, in transit 0% vs 14%). Formal lymphadenectomy (LND) was performed in 27 (48%) and 18 (43%) of MCCUP and MCCKP pts, respectively. Of these pts, 33% and 50% received neoadjuvant systemic therapy, most commonly immunotherapy; 70% and 55% received adjuvant radiotherapy. MCCUP pts had better outcomes compared to MCCKP pts (Table), with longer RFS in pts who underwent LND (not reached [NR] vs 13.1 months) as well as longer PFS in pts who did not undergo LND (17 vs 9 months) with longer OS in both subgroups (LND: NR vs 102.7 months; no LND: 74.4 vs 48.7 months). Conclusions: MCCUP patients with regional disease on exam at presentation have improved survival compared to MCCKP. Current stage III survival estimates may underestimate survival in patients with resectable disease. [Table: see text]

Health-related quality of life in patients with resectable undifferentiated pleomorphic sarcoma treated with neoadjuvant checkpoint blockade in a single institution randomized phase II clinical trial

11544

Background: In SARC028, patients with undifferentiated pleomorphic sarcoma (UPS) had a 40% overall response rate to pembrolizumab. Based on this, we conducted a randomized, phase II, non-comparative trial of combination nivolumab (nivo/RT) and ipilimumab (ipi/nivo/RT) and demonstrated an 89% major pathologic response. Here, we report the health-related quality of life (HRQoL) metrics. Methods: In this study (NCT03307616), patients with resectable UPS were randomized (1:1) to receive one dose of nivo (3mg/kg) or one dose of combination nivo (3 mg/kg) and ipi (1 mg/kg), followed by combination of nivo (3 doses, 3mg/kg every 2 weeks) plus 50 Gy in 25 fractions (both arms). HRQoL was assessed using the MD Anderson Symptom Inventory (MDASI), European Organisation for Research and Treatment of Cancer (EORTC) Quality of Life Questionnaire (QLC) Core 30, and the Functional Assessment of Cancer Therapy – General (FACT-G) questionnaires. Questionnaire completion rates were calculated using the number of patients with at least one baseline and post-baseline assessment. Analyses included mean change from baseline scores to week 11 (preoperative) and week 54 (postoperative). Results: Ten patients were randomized from October 2017 to February 2020. HRQoL were collected at baseline (n = 10) and at week 11 for five (83%) nivo/RT and three (60%) ipi/nivo/RT patients. Three (60%) nivo/RT and three (100%) ipi/nivo/RT patients had week 54 assessments. MDASI scores indicative of symptom severity and interference of daily life both decreased for patients undergoing ipi/nivo/RT (0.8→0.67 and 0.9→0.46, respectively) while both increased in the nivo/RT group (1.6→2.22 and 1.9→3.33, respectively). Both arms had similar increases at 54 weeks. Patients undergoing ipi/nivo/RT experienced a greater decline in EORTC-QLC global health status at 54 weeks than those undergoing single agent therapy (-22.92 vs -8.33). The mean change in total FACT-G score did not differ between the two arms at 11 weeks (-4.0 vs -4.5), however, there was a significant decline for patients undergoing ipi/nivo/RT at 54 weeks (-20.3 vs -5.7). Conclusions: For patients with resectable UPS, combination immune checkpoint blockade with ipi/nivo/RT is associated with an improvement in short term HRQoL compared to single-agent nivo/RT. This finding warrants further study with more patients, controlling for baseline symptom scores. Combination therapy was associated with a slower recovery to baseline function, with ongoing decline in HRQoL at 54 weeks post treatment. Our study demonstrates the feasibility of collecting HRQoL metrics, which can be key factors in guiding patient management decisions. Clinical trial information: NCT03307616.

Identifying gut microbial signatures associated with B cells and tertiary lymphoid structures (TLS) in the tumor microenvironment (TME) in response to immune checkpoint blockade (ICB)

2511

Background: While ICB has significantly improved clinical outcomes across several cancer types, only 15-20% of patients develop a durable response. Thus, novel and targetable biomarkers are needed. There is increased appreciation of the role of the gut microbiome, and TLS and B-cells in the TME in response to ICB. Here, we investigate the association between these two determinants of response in patient specimens from three randomized phase 2 neoadjuvant ICB trials of nivolumab +/- ipilimumab (melanoma (MEL; NCT02519322; n=23), non-small-cell lung cancer (NSCLC; NCT03158129; n=31), sarcoma (SARC; NCT02301039; n=17). Methods: Patients were categorized as responders (R) or non-responders (NR) based on major pathologic response, as defined in each histotype (MEL and NSCLC viable tumor ≤10%; SARC hyalinization>30%). Baseline fecal samples were profiled via 16S rRNA gene sequencing from all three cohorts to assess the composition of patient gut microbiomes. Transcriptional profiles of biopsies collected pre-ICB for MEL and SARC, and post-ICB for MEL, SARC, and NSCLC were used to assess TLS (CXCL13, CCL18, CCL19, CCL21) and B-cell (PAX5, CD79B, CR2, MS4A1) signatures in the TME, by calculated mean values of normalized gene expressions. Comparison between samples were carried out using the Wilcoxon signed-rank test. Results: There were 21 R overall (NSCLC n=9; MEL n=9; SARC n=3). Despite significant differences in alpha and beta diversity across cohorts, relative abundance of Ruminococcus was significantly higher in R (p=0.003; NSCLC p<0.001; MEL p=0.049; SARC p=0.7). B-cell signature was significantly higher post-ICB in R (R vs NR, post, TLS p=0.13; B-cell p=0.003), with consistent trends in each cohort. Longitudinal evaluation of transcriptional profiles showed that expression of TLS and B-cell signatures increased with treatment in R (pre vs post, MEL and SARC; TLS p=0.0098; B-cell p<0.001) but not NR (pre vs post; TLS p= 0.87; B-cell p= 0.15), with consistent trends in sarcoma and melanoma subgroups. Combined correlative analysis with matched specimen showed that patients with higher pre-ICB relative abundance of Ruminococcus (above median) had significant increase in B-cell signatures (pre vs post, MEL and SARC; TLS p=0.052; B-cell p=0.002) which was not seen in patients with low abundance (below median) of Ruminococcus (pre vs post, MEL and SARC; TLS p=0.56; B-cell p=0.69). Conclusions: Unifying signatures in the gut microbiome are associated with response to ICB and increased B-cell infiltration and TLS formation in the TME. We expect these findings to energize mechanistic studies and new microbiome-based interventional approaches to improve clinical outcomes with ICB. Clinical trial information: NCT02519322, NCT03158129, NCT02301039.

Evaluating the impact of perioperative antibiotic prophylaxis on the microbiome in patients with cutaneous malignancy

TPS9602

Background: Preoperative antibiotic prophylaxis is commonly used to reduce surgical site infections (SSIs). However, the rate of SSIs following surgical procedures classified as clean is only 2-3%. Overuse of antibiotics is associated with several potential adverse effects, including dysregulation of the gut microbiome. Disruption of the composition and function of the native gut microbiota, referred to as dysbiosis, has been implicated in a number of inflammatory and autoimmune disorders, as well as gastrointestinal (GI) and non-GI cancers. Recent studies have demonstrated that antibiotics have a profound and persistent effect on the gut microbiota, as evidenced by diminished overall abundance and diversity, as well as alteration of community composition that includes a decreased relative abundance of bacteria in the Ruminococcaceae family. In melanoma, diversity of gut microbiota and relative abundance of Ruminococaceae have been linked to improved survival and enhanced response following immune checkpoint blockade. In this study, we seek to determine the impact of preoperative prophylactic antibiotic use on the gut microbiome in patients following surgery for stage I or II melanoma. Methods: In this non-comparative randomized pilot trial, the impact of prophylactic antibiotic use at the time of surgical intervention on gut microbiome diversity and composition will be studied. Patients diagnosed with clinical stage I or II melanoma undergoing wide excision with or without lymphatic mapping and sentinel lymph node biopsy are randomized 1:1 to either receive preoperative cefazolin or no preoperative antibiotics. Stool samples and peripheral blood are collected before surgery, the day of surgery (optional), on post-operative day 3 (optional), and 2 weeks and 3 months following surgery. The primary endpoint for the study is change in microbiome alpha diversity at 2 weeks following surgery. Secondary endpoints are change in relative abundance of microbes at 2 weeks and 3 months after surgery and SSI rates according to whether or not prophylactic antibiotics were administered at time of surgery. Exclusion criteria include recent antibiotic use (within 3 months), allergy to beta-lactam or cephalosporin antibiotics, increased risk of infection due to medical comorbidity or use of immunosuppressive medication. Enrollment began in October 2021. As of January 2022, 22 of 30 patients have been accrued to ensure complete sample collection for 20 patients. Study findings may inform a larger trial evaluating interventions to mitigate antibiotic impact. Clinical trial information: NCT04875728.

Neoadjuvant and adjuvant nivolumab (nivo) with anti-LAG3 antibody relatlimab (rela) for patients (pts) with resectable clinical stage III melanoma

9502

Background: Neoadjuvant therapy (NT) for pts with clinical stage III melanoma remains an active area of research interest. Recent NT trial data demonstrates that achieving a pathologic complete response (pCR) correlates with improved relapse-free (RFS) and overall survival (OS). Checkpoint inhibitor (CPI) NT with either high or low dose ipilimumab and nivolumab regimens produces a high pCR rate of 30-45% but with grade 3-4 toxicity rate of 20-90%. In metastatic melanoma (MM), the combination of nivo with rela (anti Lymphocyte Activation Gene-3 antibody) has demonstrated a favorable toxicity profile and responses in both CPI-naïve and refractory MM. We hypothesized that NT with nivo + rela will safely achieve high pCR rates and provide insights into mechanisms of response and resistance to this regimen. Methods: We conducted a multi-institutional, investigator-initiated single arm study (NCT02519322) enrolling pts with clinical stage III or oligometastatic stage IV melanoma with RECIST 1.1 measurable, surgically-resectable disease. Pts were enrolled at 2 sites and received nivo 480mg IV with rela 160mg IV on wks 1 and 5. Radiographic response (RECIST 1.1) was assessed after completion of NT; surgery was conducted at wk 9 and specimens were assessed for pathologic response per established criteria. Pts received up to 10 additional doses of nivo and rela after surgery, with scans every 3 mo to assess for recurrence. The primary study objective was determination of pCR rate. Secondary objectives included safety, radiographic response by RECIST 1.1, event-free survival (EFS), RFS, and OS analyses. Blood and tissue were collected at baseline, at day 15, day 28, and at surgery for correlative analyses. Results: A total of 30 pts (19 males, median age 60) were enrolled with clinical stage IIIB/IIIC/IIID/IV (M1a) in 18/8/2/2 pts, respectively. 29 pts underwent surgery; 1 pt developed distant metastatic disease while on NT. pCR rate was 59% and near pCR ( < 10% viable tumor) was 7% for a major pathologic response (MPR, pCR + near pCR) of 66%. 7% of pts achieved a pPR (10-50% viable tumor) and 27% pNR (≥50% viable tumor). RECIST ORR was 57%. With a median follow up of 16.2 mos, the 1 -year EFS was 90%, RFS was 93%, and OS was 95%. 1-year RFS for MPR was 100% compared to 80% for non-MPR pts (p = 0.016). There were no treatment related gr 3/4 AEs that arose during NT; 26% of pts had a gr 3/4 AE that began during adjuvant treatment. Conclusions: Neoadjuvant and adjuvant treatment with nivo and rela achieved high pCR and MPR rates with a favorable toxicity profile in the neoadjuvant and adjuvant settings. Pts with MPR had improved outcomes compared to non-MPR pts. Translational studies to discern mechanisms of response and resistance to this combination are underway. Clinical trial information: NCT02519322.

Preliminary results of a phase II study of neoadjuvant checkpoint blockade for surgically resectable undifferentiated pleomorphic sarcoma (UPS) and dedifferentiated liposarcoma (DDLPS)

11505

Background: is a randomized, phase II non-comparative trial evaluating the efficacy of neoadjuvant checkpoint blockade [nivolumab (N) or ipilimumab/ nivolumab (I/N)] in patients (pts) with surgically resectable retroperitoneal DDLPS or extremity/truncal UPS treated with concurrent neoadjuvant radiation therapy (RT, UPS only). Methods: Primary endpoint was pathologic (path) response. Secondary endpoints were safety, RECIST response, recurrence-free survival, overall survival and patient-reported outcomes. Biospecimens (tumor, blood, fecal microbiome) at baseline, on therapy, and at time of surgery were collected and will be assessed for immune-based prognostic biomarkers. We assessed correlation between radiographic and pathologic response by linear regression. Correlative analyses includes assessment of tumor PD-L1 expression, characterization of tumor immune infiltrates by multiplex immunohistochemistry, and transcriptomic and genomic analyses. Results: Of the 25 pts enrolled; 24 are evaluable for response (14 DDLPS, 9 UPS). Clinical activity was variable by histologic subtype and treatment with RT. Median path response in the UPS cohort was 95% [95% CI 85–99] and was similar between the N/RT and I/N/RT groups (Table). Median path response in the DDLPS cohort was 22.5% [95% CI 85–99; Table]. Median change in tumor size (radiographic response) was -4% and +9% in the UPS and DDLPS cohorts, respectively. There was no correlation between path response and radiographic response (R2 0.0309; p = 0.43). Of 8 pts with path response ≥ 85%, there was 1 partial response, 5 stable disease and 2 progressive disease by RECIST criteria. There was 1 delay to surgery due to grade 3 hyperbilirubinemia (Arm B). There was no difference in toxicity between N/RT and I/N/RT. Conclusions: N/RT and I/N/RT have significant clinical activity in UPS; more than expected compared to historic controls. Toxicity profiles were as expected and the majority of patients underwent resection without delay. Larger studies evaluating N/RT in UPS are warranted given the significant path response in this cohort. RECIST was not associated with path response and better markers of on-treatment clinical activity are needed. Correlative analyses that may guide combination strategies are ongoing and will be presented at the meeting. Clinical trial information: NCT03307616 .

Efficacy of immune checkpoint inhibitors for the treatment of metastatic melanoma (MM) in patients with concurrent chronic lymphocytic leukemia (CLL)

e22044

Background: Patients with CLL have immune impairment with abnormalities in T-cell subset composition and immune synapse formation. The impact of these defects on response to immune checkpoint inhibitors (CPI) is not known. Given the high incidence of melanoma in patients with CLL we sought to evaluate the response to CPI in patients with concomitant MM and CLL. Methods: Retrospective analysis of 24 patients (pts) with concurrent CLL and MM who received a total of 38 CPI therapies between July 1997 and July 2019. Primary objective was to determine objective response rate (ORR), defined as complete response (CR) or partial response (PR) by RECIST1.1. Secondary outcomes included event-free survival; overall survival (OS), and duration of response (DOR). Results: The median age at CLL and melanoma diagnosis was 62 and 63 years, respectively. 71% of patients were male. Most presented with early stage CLL at diagnosis (67%), 60% had mutated IGVH, and 47% had deletion of 13q by FISH. 71% remained on observation for their CLL. Median time from melanoma diagnosis to CPI initiation was 13.5 months. 83% had stage IV MM and 17% stage III MM at the time of therapy. 17% had increased LDH. The most common melanoma mutations were BRAF(35%), BRAFV600 (26%), TP53 (30%) and NRAS (26%). Median follow up was 37 months and the ORR was 24% (Table). Median DOR was 41 months and median OS is 26.4 months. Immune-mediated adverse events occurred in 42%, including 13% fever, 11% thrombotic events, 8% endocrine dysfunction. 13 pts are alive and 11 pts died (8 pts due to MM progression). There were no significant changes in absolute lymphocyte counts during CPI therapy. 2 pts received CPI while on ibrutinib or ibrutinib+venetoclax therapy with ongoing CLL responses. Conclusions: Our experience indicates that CPIs can be effective for the treatment of MM in patients with concurrent CLL, achieving durable responses. Immuno-mediated toxicities were frequently observed. A lower ORR was observed in first-line CPI in MM, however the numbers of pts are small. Further studies are needed to determine if initial or concurrent treatment for CLL could improve CPI outcomes and survival. Additional studies evaluating T cells function and tissue infiltration in these patients are ongoing. [Table: see text]

Clinical activity of pembrolizumab (P) in undifferentiated pleomorphic sarcoma (UPS) and dedifferentiated/pleomorphic liposarcoma (LPS): Final results of SARC028 expansion cohorts

11015

Background: Immune checkpoint inhibitors have demonstrated activity in multiple tumor types but their activity in soft tissue sarcomas remains limited. In the multicenter phase II study, SARC028, the anti-PD-1 antibody, P demonstrated objective responses that were largely restricted to UPS and LPS subtypes. We now report outcomes from 2 expansion cohorts of SARC 028 in advanced UPS and LPS. Methods: To further confirm the clinical activity of P in UPS and LPS, we enrolled an additional 30 pts in each of 2 expansion cohorts for a total of 40 UPS and 40 LPS pts. Primary endpoint was investigator-assessed response by RECIST v1.1. Secondary endpoints were safety, progression-free survival (PFS), 12-week PFS rate, and overall survival (OS). An ORR of 25% was considered clinically meaningful and < 10% was considered to show lack of efficacy. P was to be considered a success if 8 or more of 40 enrolled patients had a PR or better (1-sided α = 0.042, 82% power). Pts age ≥18 with advanced, refractory UPS or LPS received 200 mg of P IV every 3 weeks until progression or unacceptable toxicity. Results: Preliminary results from the first 10 pts in each of the UPS and LPS cohorts have been reported. We now present summary data after enrolling an additional 30 pts in each cohort. The ORR in the UPS cohort was 23% (9/40), with an additional 5/30 PRs observed in the expansion cohort (total 2 CRs, 7 PRs). In the LPS cohort, the ORR was 10% (4/39 evaluable pts), with an additional 2/30 PRs observed (total 4 PRs). Median PFS for the UPS group was 3 months [95% CI: 2, 5] and 2 months [95% CI: 2, 4] for the LPS group. 12-week PFS rate was 50% in UPS [95% CI: 35, 65] and 44% in LPS [95% CI: 28, 60]. The UPS group had a median OS of 12 months [95% CI: 7, 34] and 13 months [95% CI: 8, NR] for the LPS group. P was well tolerated with no unexpected toxicities. Conclusions: The UPS cohort achieved its primary endpoint, however the activity of P in UPS deserves further evaluation in a randomized study. The activity of P was not confirmed in the LPS cohort. Ongoing biomarker analyses may direct better patient selection and guide future combination strategies. Clinical trial information: NCT02301039.

A phase II study of oral azacitidine (CC-486) in combination with pembrolizumab (PEMBRO) in patients (pts) with metastatic melanoma (MM)

9560

Background: Immune checkpoint blockade (ICB) have improved survival for many pts with MM, offering durable responses in up to 35% of pts, but many have short-lived or no response ( > 40%). A potential immune escape mechanism is the subversion of cellular epigenetic machinery to impact multiple aspects of the immune response such as suppression of Cancer Testis Antigen (CTA), which can be reversed in preclinical models by DNA hypomethylating agents (HMA), thereby increasing cancer cell immunogenicity. HMAs can also increase T cell infiltration and T cell-mediated tumor killing, and they achieve synergy with CBI in preclinical models. This suggests that epigenetic therapy with CBI is a rational combination to target MM. We hypothesize that CC-486 (an oral HMA) + PEMBRO will be tolerated at biologically relevant doses and enhance response to PEMBRO in pts with MM who are PD-1 naïve and reverse resistance to (ICB) in pts refractory/resistant to PD-1. Methods: This study (NCT02816021) evaluated the safety and efficacy of CC-486 (300 mg PO QD on days 1-14/21 day cycle) + PEMBRO (200mg IV Q 21 days) defined by Objective Response Rate (ORR) by RECIST 1.1 in pts with MM. PD-1 naïve pts were assigned to Arm A and pts with progression on prior PD-1 therapy to Arm B. Unlimited prior systemic therapies were allowed on Arm B. Continuous monitoring for toxicity and futility was performed and assumes an ORR of > 35% (Arm A) and > 15% (Arm B) at 95% power. Tumor biopsies at baseline and post treatment were mandated. Results: 22pts, 11 in each arm, have been treated. The most common AEs were nausea, vomiting, diarrhea, fatigue, and anemia. The most common gr 3/4 toxicities were neutropenia (3), diarrhea (2), dehydration (2), and rectal hemorrhage (1). 5 of 9 evaluable pts in Arm A achieved a PR (55% ORR); 0 of 9 evaluable pts in Arm B pts have responded. Conclusions: Although this regimen was tolerated in both arms, Arm B met futility stopping rules and was closed. The initial response rate in Aim A (55%) is promising, and accrual to this Arm continues. Analyses of longitudinally collected tumor biopsies are underway to interrogate the effects of HMA on the immune response to both arms. Clinical trial information: NCT02816021.

In-hospital and long-term outcomes in patients with malignancy undergoing percutaneous endoscopic gastrostomy (PEG)

9086

Background: PEG is widely performed in cancer patients as a means of providing nutrition or palliation. Although considered safe, PEG-associated outcomes in these patients remain poorly described. We examined the safety and benefits of PEG placement in this patient population. Methods: A five year retrospective review of all patients with malignancy (excluding head/neck, thoracic) who underwent attempted PEG at our institution was performed. Results: PEG was placed in 187 of 189 patients; 64 with hematologic malignancy (H-M), 125 non-hematologic malignancy (NH-M). Median age at time of PEG was 60.8 years. Indication was nutritional support (100%) in H-M, enteral access (59.2%) and management of obstructive symptoms (38.4%) in NH-M. A minority were able to return home (27.5%), discontinue parenteral nutrition (22%), advance diet for nutrition (24.6%) or comfort (17.1%) with the rest remaining NPO. Overall rates of PEG-related major (aspiration, tube dislodgement/leakage, bleeding, visceral injury, respiratory failure after procedure, cardiac arrest) and minor (superficial infection, ileus) complications were 21.4% and 11.3%, respectively, with higher rates in H-M (34.4% and 20.3% vs 14.4% and 6.4%). All cause in-hospital mortality was high: 31.3% H-M, 13.6% NH-M. Median time from PEG placement to death was 54 days. Leading cause of death differed by malignancy: respiratory failure and sepsis in H-M (31% and 21%), primary malignancy in NH-M (75%). Overall one year mortality was 56%. Code status was changed in 21% of patients after PEG (“Full Code” to “Do Not Resuscitate/Do No Intubate” or “Comfort Measures Only”). Multivariate/subgroup analyses will be presented at time of meeting. Conclusions: PEG placement in this study population was associated with significant procedure-related complications and failed to achieve TPN independence or advancement of diet. Nearly 25% of patients declined aggressive resuscitation strategies after undergoing surgery for PEG. Thus, higher burden of counseling is needed to carefully weigh the risk and benefit of PEG placement in these patients. Further studies are needed to elucidate the factors affecting the decision process and patient selection.