Search

PURPOSE Immune checkpoint blockade (ICB) therapy has significantly improved clinical outcomes in bladder cancer. Identification of correlates of benefit is critical to select appropriate therapy for individual patients. METHODS To reveal genetic variables associated with benefit from ICB, we performed whole-exome sequencing on tumor specimens from 88 patients with advanced bladder cancer treated with ICB. RESULTS We identified several genetic factors that correlated with progression-free and overall survival after ICB therapy including ARID1A mutation, tumor mutational burden, intratumoral heterogeneity, the ratio of nonsynonymous to synonymous mutations in the immunopeptidome (immune dN/dS), and tumor cell purity. In addition, we noted that neutrophil-to-lymphocyte ratio and smoking history were negatively associated with overall survival. These genetic characteristics define four molecular subtypes demonstrating differential sensitivity to ICB. We validated the association of these four subtypes with clinical benefit from ICB in an independent cohort (IMvigor210). Finally, we showed that these molecular subtypes also correlate with outcome, although with distinct relationships, among patients not treated with ICB using The Cancer Genome Atlas (TCGA) bladder cancer cohort. Using parallel RNA sequencing data, the subtypes were also shown to correlate with immune infiltration and inflammation, respectively, in the IMvigor210 and TCGA cohorts. CONCLUSION Together, our study defines molecular subgroups of bladder cancer that influence benefit from ICB.

PURPOSE To compare oncologic outcomes and genomic alteration profiles in patients with bladder and urachal adenocarcinoma, urothelial carcinoma (UC) with glandular differentiation, and UC, not otherwise specified (NOS) undergoing surgical resection, with emphasis on response to systemic therapy. METHODS We identified patients with bladder cancer with glandular variants who underwent surgical resection at Memorial Sloan Kettering from 1995 to 2018 (surgical cohort) and/or patients who had tumor sequencing using a targeted next-generation sequencing platform (genomics cohort). Pathologic complete and partial response rates to neoadjuvant chemotherapy (NAC) and recurrence-free and cancer-specific survival were measured. Alteration frequencies between histologic subtypes were compared. RESULTS Thirty-seven patients with bladder adenocarcinoma, 46 with urachal adenocarcinoma, 84 with UC with glandular differentiation, and 1,049 with UC, NOS comprised the surgical cohort. Despite more advanced disease in patients with bladder and urachal adenocarcinoma, no significant differences in recurrence or cancer-specific survival by histology were observed after adjusting for stage. In patients with UC with glandular differentiation, NAC resulted in partial (≤ pT1N0) and complete (pT0N0) responses in 28% and 17%, respectively. Bladder and urachal adenocarcinoma genomic profiles resembled colorectal adenocarcinoma with frequent TP53, KRAS, and PIK3CA alterations while the genomic profile of UC with glandular differentiation more closely resembled UC, NOS. Limitations include retrospective nature of analysis and small numbers of nonurothelial histology specimens. CONCLUSION The genomic profile of bladder adenocarcinomas resembled colorectal adenocarcinomas, whereas UC with glandular differentiation more closely resembled UC, NOS. Differences in outcomes among patients with glandular bladder cancer variants undergoing surgical resection were largely driven by differences in stage. Cisplatin-based NAC demonstrated activity in UC with glandular differentiation, suggesting NAC should be considered for this histologic variant.

659 Background: Combination therapy with I/N can achieve long-term benefit in a sizable proportion of RCC pts, but broad application has been challenged by a high incidence of irAEs. Host features affecting antigen presentation are candidate determinants of irAE pathogenesis, and isolated reports have linked Human Leukocyte Antigen (HLA) variations to organ- specific irAEs. We analyzed data for pts treated on a phase 3 trial to investigate association of class 1 HLA variants with development of irAEs from I/N and develop a tool to predict risk for toxicity. Methods: We analyzed germline HLA + clinical data for 472 pts receiving I/N vs sunitinib (SUT). Based on allelic variants for HLA-A and B pts were categorized into 12 established HLA "Super-Types" (ST), sets of HLA variants with largely overlapping peptide binding specificity. We conducted uni- and multivariate (UV; MV) Cox proportional hazards regression to correlate HLA-ST variables with time to treatment-related AEs (TTrAE, grade 2+) in I/N pts using the Kaplan Meier method. Several MV HLA-ST models were developed on a discovery set (DISC; 2/3 random sample of I/N pts without replacement ) and compared using model concordance (c-index) on a validation set (VAL; remaining 1/3). Model coefficients were used to create a “HLA-ST score” that could be computed for individual pts. Results: 235 pts and 237 pts received I/N vs SUT and had available HLA data, respectively. On UV analysis for I/N, HLA-B07 ST had protective association TTrAE (HR= 0.65, 95% CI: 0.46,0.90; p= 0.010), while B62 associated adversely (HR=1.64, 95% CI: 1.12, 2.40; p=0.014). Relevant to the development of our model we identified interactions between several pairs of HLA ST. The HLA ST model with best performance in DISC (c-index= 0.606) and VAL (c-index=0.595) integrates B07, B62, A01, B08, and two interactions: B07-B08 and B07-A01. The model-generated HLA-ST score was significantly associated with TTrAE after adjusting for race, BMI, region, PDL1 status, and MSKCC risk score (DISC p= vs < median had significantly different TTrAE both in DISC (p=0.004) and VAL (p=0.009); again, no difference was observed in the SUT arm (p=0.5). The weighted HLA-ST score had no association with PFS or OS for I/N pts. Conclusions: In this large sample of I/N-treated patients, class I HLA variants were associated with the risk of developing irAEs. We developed a HLA ST based score that correlated with treatment toxicity independent of relevant clinical and demographic features. No association with TTrAE was seen in SUT-treated pts. These results highlight the potential of characterizing germline features to predict immune related toxicity upfront and deserves further study. Clinical trial information: NCT02231749 .

511 Background: Perioperative platinum-based chemo for UTUC improves pathologic responses and disease-free survival (DFS) while anti-PD1 monotherapy has limited activity to date (PURE-02). Here, we report the results from stage 1 of a phase II neoadjuvant trial of N+I for cisplatin-ineligible patients (pts) with UTUC. Methods: Cisplatin-ineligible pts with histologically confirmed high-grade UTUC and/or radiographically invasive UTUC with positive selective urine cytology were eligible. Pts were treated with I 3mg/kg + N 1mg/kg (weeks [wk] 0, 6), and N 3mg/kg (wk 3) prior to radical nephroureterectomy (NU). The primary endpoint (EP) was pathologic complete response (pCR, ypT0pN0) and secondary EPs included ypT1 or node positive at NU and all 4 pts with TMB>10 were

PURPOSE Recurrent or metastatic adenoid cystic carcinoma (R/M ACC) is a malignant neoplasm of predominantly salivary gland origin for which effective therapies are lacking. We conducted a phase II trial evaluating the multitargeted tyrosine kinase inhibitor lenvatinib in patients with R/M ACC. PATIENTS AND METHODS This study was conducted with a two-stage minimax design. Patients with histologically confirmed R/M ACC of any primary site with radiographic and/or symptomatic progression were eligible. Any prior therapy was allowed except previous lenvatinib. Patients received lenvatinib 24 mg orally per day. The primary end point was overall response rate. Secondary end points were progression-free survival and safety. An exploratory analysis of how MYB expression and genomic alterations relate to outcomes was conducted. RESULTS Thirty-three patients were enrolled; 32 were evaluable for the primary end point. Five patients (15.6%) had a confirmed partial response, 24 patients (75%) had stable disease, two patients (6.3%) discontinued treatment as a result of toxicity before the first scan, and one patient (3.1%) had progression of disease as best response. Median progression-free survival time was 17.5 months (95% CI, 7.2 months to not reached), although only eight progression events were observed. Patients otherwise were removed for toxicity (n = 5), as a result of withdrawal of consent (n = 9), or at the treating physician’s discretion (n = 6). Twenty-three patients required at least one dose modification, and 18 of 32 patients discontinued lenvatinib for drug-related issues. The most common grade 3 or 4 adverse events were hypertension (n = 9; 28.1%) and oral pain (n = 3; 9.4%). Three grade 4 adverse events were observed (myocardial infarction, n = 1; posterior reversible encephalopathy syndrome, n = 1; and intracranial hemorrhage, n = 1). CONCLUSION This trial met the prespecified overall response rate primary end point, demonstrating antitumor activity with lenvatinib in R/M ACC patients. Toxicity was comparable to previous studies, requiring monitoring and management.

4584 Background: We previously reported the clinical outcomes of a positive multi-center phase II trial of neoadjuvant gemcitabine (G) and cisplatin (C) plus atezolizumab (A) in patients with muscle-invasive bladder cancer (Funt, et al. JCO 2022). In this and another trial of neoadjuvant GC with pembrolizumab (Rose et al, JCO 2021), PD-L1 positivity by immunohistochemistry was not predictive of non–muscle-invasive downstaging ( < pT2N0). Therefore, we investigated other pre-treatment tissue-based genomic and gene expression biomarkers of response and resistance. Methods: 36 pts had pre-treatment tissue available for genomic analysis. We performed targeted hybridization capture DNA sequencing using the CLIA-certified MSK-IMPACT platform and whole transcriptome RNA sequencing. We examined genomic and gene expression biomarkers which have been previously investigated in the context of neoadjuvant cisplatin-based chemotherapy or anti-PD-1/L1 immunotherapy for MIBC, including tumor mutation burden (TMB), a DNA damage response (DDR) 9-gene panel (NCT03609216) associated with response to neoadjuvant chemotherapy, and an 8-gene cytotoxic T cell transcriptional signature associated with response to neoadjuvant A (tGE8; Powles et al, Nature Medicine 2019). We also evaluated TGF-β pathway activation, which was associated with resistance to A in pts with metastatic BC (Mariathasan et al, Nature 2018). Putative biomarkers were assessed for correlation with < pT2N0, the trial’s primary endpoint. Results: DNA was available from all 36 pts, and RNA met quality control metrics for 29 pts. TMB was significantly higher in pts with < pT2N0 (median 16 mut/Mb, IQR 12-25) versus ≥ pT2N0 (median 10 mut/ Mb, IQR 8-10; p < 0.01). A single patient had a TMB > 200 Mut/Mb with a POLE hotspot mutation and achieved pT0N0; TMB was still significantly higher in responders after omission of this patient (p < 0.01). Nine of 25 pts (36%) with < pT2N0 had a deleterious DDR mutation versus 1 of 10 pts (10%) with ≥ pT2N0 (p = 0.13). While tGE8 was significantly increased in patients with < pT2N0 compared to those without (p = 0.01), TGF-β pathway activation was not increased in pts with ≥ pT2N0 (p = 0.99). Conclusions: TMB and the tGE8 cytotoxic T cell transcriptional signature were associated with response to combination GC+A in muscle-invasive bladder cancer. More detailed molecular analyses will be reported. Clinical trial information: NCT02989584.

6089 Background: A pilot study showed MEK inhibition could enhance radioiodine (RAI) avidity/efficacy in 5 RAS mutant (MUT), RAI-refractory (RAIR) thyroid cancer (TC) patients (pts). This phase 2 trial with the MEK 1/2 inhibitor trametinib (tram) was conducted to define the efficacy of this “redifferentiation” strategy in RAS MUT RAIR pts and separately in a RAS wild-type (WT) cohort. Methods: Recurrent and/or metastatic, RAIR TC pts w/ RAS MUT (Cohort A) or RAS WT (excluding BRAFV600E) (Cohort B) tumors were treated w/ tram (2 mg orally daily). Progressive disease or new/worsening disease-related symptoms was required for Cohort A pts. 124I PET was performed at baseline and the fourth week of tram. If the second 124I PET showed increased RAI avidity allowing > 2000 cGy to be delivered to a tumor w/ < 300 mCi 131I, pts were treated w/ 131I, guided by whole body and blood dosimetry. Tram was continued through 2 days s/p 131I. Pts who did not qualify for 131I from A/B were taken off study or continued tram alone (Cohort C). For Cohort A (n = 25), the two co-primary endpoints were objective response rate (ORR) and progression-free survival (PFS) 6 months (mos) s/p 131I. Observing either >4 pts w/ confirmed complete or partial response (cCR or cPR) or > 9 progression-free at 6 mos would be considered promising. Secondary endpoints were the proportion of pts w/ increased 124I, safety/tolerability of tram and thyroglobulin changes s/p RAI. The Cohort B primary endpoint was the proportion of pts whose tumors exceeded the lesional dosimetry threshold for 131I w/ tram. An exploratory endpoint for Cohort C was best objective response (BOR) w/ tram. Results: 25 RAS MUT pts enrolled in Cohort A. 23 had at least one (> 1) 124I (-) lesion, 21 had >1 124I (+) lesions and 4 pts had tumors lacking any 124I uptake. After tram treatment, 22/25 had increased 124I uptake; 17/23 had 124I (-) tumors convert positive. Importantly, 15/25 (60%) pts had increased 124I uptake and met lesional dosimetry criteria for 131I on tram. Of 14 pts treated w/ 131I, 8 (57%) achieved cPR, 3 (21%) stable disease (SD) and 3 (21%) progression of disease (PD) 6 mos s/p RAI, translating to 32% ORR and 44% 6-month PFS among all 25 pts. Cohort B had 9 pts (4 Class II BRAF alterations, 4 RET rearrangements, 1 STK11 mutation). 3/4 pts w/ Class II BRAF altered tumors qualified for 131I, leading to 1 cPR, 2 SD 6 mos s/p 131I. 1/4 pts w/ RET rearranged tumors qualified for 131I, producing SD at 6 mos. The STK11 MUT pt did not have increased 124I uptake w/ tram. 7 131I-ineligible pts enrolled to continue tram (Cohort C). Two serious adverse events (grade 3 anemia [Cohort A], grade 3 ejection fraction decrease [Cohort C]) and 3 grade 1 blurred vision/decreased visual acuity AEs were related to tram. Conclusions: Trametinib enhanced RAI uptake/efficacy in a subset of RAS MUT and Class II BRAF altered tumors. Further study to define the efficacy and optimal application of this therapeutic strategy is warranted. Clinical trial information: NCT02152995.

556 Background: Bladder cancer has one of the highest rates of human epidermal growth factor receptor 2 (HER2) alteration. Novel HER2-directed agents are being investigated in metastatic BC. We sought to define the incidence and clinical characteristics of HER2-altered BC across disease states. Methods: We retrospectively analyzed our single-institution, clinically annotated cohort of urothelial BC pts with available genomic profiling data (MSK-IMPACT). We quantified the prevalence of HER2 alterations, defined as driver mutation (based on OncoKB), and/or amplification, across BC disease states. We examined the association between HER2 alteration and disease progression and survival. The Kaplan-Meier method was used for time-to-event analyses. Results: A total of 1073 BC pts underwent MSK IMPACT profiling of tumor tissue derived from the following disease states: 36% (n = 380) non-muscle invasive (NMI)BC, 41% (n = 443) muscle invasive (MI)BC, and 23% (n = 250) (met)BC. At initial diagnosis, the median age was 67 years, 77% (n = 822) were male, 86% (n = 928) were white, and 66% (n = 710) were smokers. Overall, 16% (n = 177) of pts had any oncogenic HER2 alteration (Table), including 11% with a HER2 driver mutation and 7% with HER2 amplification The most frequent mutations were S310F (40%, n = 53) and S310Y (14%, n = 19). The rate of HER2 amplification was different among the three groups (p = 0.002), 9% in MIBC and metBC compared to 3% in NMIBC. Among 514 pts with NMIBC, the median time to progression (TTP) to MIBC or metBC was 111.6 months (95% Cl: 85.7-NR). Among NMIBC pts, TTP was significantly shorter for HER2-amplified (n = 17) vs. non-amplified (n = 497) (HR = 1.99, 95%CI: 1.05-3.76, p = 0.034, median 26 vs. 114 months). Among pts with metBC receiving frontline platinum-based chemotherapy (n = 143), the median overall survival (OS) was 25.3 months (95%CI: 18.5-33.9). OS was numerically higher in pts with any oncogenic HER2 alteration (n = 26) compared to wild-type (n = 117) (HR = 0.59, 95% Cl: 0.33-1.07, p = 0.082), though this difference was not statistically significant. The median OS for platinum-refractory metBC pts receiving 2nd line immunotherapy (n = 63) was 10.3 months (95%CI: 7.2-31.6), and the association between OS and HER2 alteration was not significant (HR = 0.57, 95%CI: 0.24-1.37, p = 0.2). Conclusions: HER2 amplification is more frequent in MIBC and metBC than in NMIBC. In NMIBC, HER2 amplification is associated with shorter TTP to MIBC or metBC. HER2 alteration in metBC is associated with a non-significant trend towards improved OS in frontline platinum-treated pts, though this analysis is limited by small sample size.[Table: see text]

498 Background: Cisplatin-based neoadjuvant chemotherapy followed by radical cystectomy (RC) confers survival benefit but a substantial subset of MIBC pts are cisplatin-ineligible. We conducted a phase II trial of N±I neoadjuvant therapy for cisplatin-ineligible MIBC. Methods: Cisplatin-ineligible pts with MIBC (cT2-T4aN0M0) were enrolled into 2 consecutive cohorts of 15 pts each (C1: N 3 mg/kg q 2 weeks (wk) x 5; C2: I (3mg/kg) + N (1mg/kg) wk 0 and 6, with N (3mg/kg) wk 3 and 9). A 3rd cohort was planned (C3: I (3mg/kg) + N (1mg/kg) q 3 wk x 3). Primary endpoint (EP) was eligibility for planned RC within 60 days without delay from treatment-related adverse events (TRAEs) or progressive disease (PD). Secondary EPs included pathologic downstaging (PaDo

6002 Background: R/M SGCs are a diverse group of malignant neoplasms arising from the major or minor salivary glands and have no standard treatment. The impact of combining PD-1/CTLA-4 checkpoint blockade in R/M SGCs is unknown. Methods: In a Simon's two-stage minimax phase II trial, pts with progressive R/M SGCs (any histology except adenoid cystic carcinoma (ACC)) were enrolled and treated with nivolumab 3 mg/kg every 2 weeks plus ipilimumab 1 mg/kg every 6 weeks (1 cycle = 6 weeks). Imaging, using RECIST v1.1 response assessment, was scheduled to be performed approximately every 12 weeks. The primary endpoint was best overall response (BOR = complete response [CR]+partial response [PR]) per RECIST v1.1. To detect a difference between an unacceptable BOR of 5% and a desirable BOR of 20% (one-sided type I error of 10%, power of 90%), at least 1 in the first 18 pts required an observed response. At least 4 responses of 32 total pts were needed to meet the primary endpoint. Treatment beyond progression of disease (PD) was allowed at the discretion of the investigator. A second cohort of pts with ACC was analyzed and reported separately. Results: From 7/25/2017-7/16/2020, 32 pts were enrolled and evaluable for the primary endpoint. There was 3 confirmed PRs in the first 18 pts, therefore enrollment of the second stage continued. BOR rate was 16% (5/32). Seven pts never reached a first disease assessment and were classified as non-responders: 5 due to clinical PD, 1 due to toxicity, and 1 pt withdrew. Four pts discontinued the trial for toxicities: pancytopenia (1), blurry vision (1), cardiomyopathy/hyperglycemia (1), and neutropenic sepsis (1), and mucositis (1). The 5 confirmed responders had regressions ranging from -66% to -100% in target lesions, with a duration of therapy ranging from 15.7 to 29.5 months (treatment ongoing for one as of 2/6/20). Conclusions: This cohort met its primary endpoint, and the responses observed were dramatic and durable. Paired biopsy and peripheral blood samples will be analyzed to elucidate insights into mechanisms of response and resistance to dual checkpoint blockade. Clinical trial information: NCT03172624.

5050 Background: Only a subset of mUC patients (pts) derives benefit from CPI. We sought to evaluate potential clinical and genomic predictors, with additional focus on extreme outcomes. Methods: CPI-treated mUC with tumor sequencing with MSK-IMPACT (up to 468 genes) were identified. Pre-CPI clinical variables (including hemoglobin [hgb], lymphocyte [lymph], neutrophil [neut] and eosinophil [eos] counts) were extracted. Endpoint was time on treatment (ToT). Poor responders were defined as ToT ≤9 weeks and overall survival ≤3 months; exceptional responders had ToT >20 months. A multivariate Cox model for ToT was performed with clinical variables chosen by stepwise selection minimizing Akaike information criterion. Genomic analyses were adjusted for multiple testing. Results: 166/171 identified mUC were evaluable. Median age was 68 years, 78% were male. 72% had bladder primary, while 17%, 35% and 28% had liver, lung and bone metastases (mets). 67% had prior platinum chemo. 24 pts (14%) were poor and exceptional responders, respectively. In univariate analysis, liver mets were associated with shorter ToT while DNA damage repair (DDR) gene alterations, higher body mass index (BMI), eos, hgb and tumor mutation burden (TMB) were associated with longer ToT. No specific genomic alterations achieved q

6587 Background: Immune checkpoint blockade (ICB) has limited efficacy for radioiodine-refractory thyroid cancer. The high incidence of autoimmune thyroid disease and ICB-induced hypothyroidism suggests that loss of T cell tolerance to thyroid protein epitopes is common and can be activated by ICB to induce immune responses. We hypothesize that RAI can enhance presentation of thyroid protein immunogens and putative neoantigens in thyroid cancers to amplify the effectiveness of ICB. We studied the safety and efficacy of RAI plus the anti-PD-L1 agent durvalumab (durva) in recurrent/metastatic (R/M) patients (pts). Methods: Pts. had at least one RAI-avid tumor on the most recent RAI scan or one tumor on FDG PET with an SUVmax < 10. RECIST measurable disease was required. Any number of prior therapies was allowed. Pts were treated with durva 1500 mg IV every 4 weeks with recombinant human TSH (rhTSH)-stimulated RAI (100 mCi) administered in Cycle 1. Treatment beyond progression was allowed. The primary objective was to assess safety. Durva related dose limiting toxicities (DLTs) were monitored for 6 weeks after the first dose. Since no durva DLTs were observed in the first 6 pts, per protocol rules the trial accrued 11 pts total. Secondary objectives were assessing best overall response (BOR) per RECIST and progression-free survival (PFS). Results: 11 pts (7 female) were enrolled. Eight had prior drug therapy. No DLTs or > Grade 3 durva related adverse events (AEs) were observed. The most common non-laboratory AEs (regardless of attribution) were cough (7), hypertension (7), pain (6), edema (5), and fatigue/nausea/diarrhea/arthralgia/dry skin/dyspnea/edema (4 each). As of 2/6/20, 2 had partial response, 7 stable disease, and 2 progression of disease as BOR. Six pts had tumor regression. Four pts received treatment for > 6 months. Six are still on treatment. Analyses of research biopsies (bxs) (8 had pre-treatment bxs, 6 had an additional on-treatment bx) will be presented. Conclusions: Durva plus RAI is safe and well tolerated. The preliminary efficacy signal in this small cohort is promising. Understanding how RAI plus PD-L1 targeting impacts the tumor immune microenvironment may guide how RAI should be evaluated in future ICB trials. Clinical trial information: NCT03215095 .

742 Background: To investigate the distribution of spinal metastasis in metastatic renal cell carcinoma (mRCC) and to explore relationships between biological and clinical factors and patterns of spinal spread. Methods: An institutional database was queried to identify patients with mRCC and spinal metastatic involvement from June 2005 – November 2018. A blinded radiologist examined all cross-sectional imaging involving the spine and scored each level for absence or presence of disease. Clinical and biologic features including primary tumor size and degree of spinal and non-bony metastatic involvement (including regional lymph node and distant deposits) were collected. Spinal distributions were evaluated by the Kolmogorov Smirnov test and compared across radiographic and clinical parameters. Results: One-hundred patients with 685 spinal levels involved by mRCC were evaluated. A nonuniform spatial distribution was observed across the cohort; a preponderance of thoracolumbar involvement was noted with the mode at L3 (p7cm, patients who had distant spread versus bone-only disease, and patients with increasing number of spine levels involved (1 versus >5 levels) had a significantly different distribution (p

510 Background: FGFR3 alterations are prevalent in non-muscle-invasive bladder cancer (NMIBC). We conducted a marker lesion study of the oral FGFR inhibitor BGJ398 in patients with intermediate-risk NMIBC. Methods: Patients with recurrent clinical high-grade papillary NMIBC following intravesical BCG were eligible. FGFR3 alteration status was determined via testing of pre-treatment or archival tumor tissue. BGJ398 was administered at 125 mg PO daily for a 3 week on, 1 week off schedule (1 cycle). Response was determined after 2 cycles (at 7 weeks). Patients with a complete response (CR) had the option to continue therapy. Results: We enrolled four patients on trial, which we closed secondary to low accrual. Three patients had a CR at the 7-week evaluation. The other patient had a smaller, necrotic lesion after prematurely discontinuing treatment at 4 weeks. Clinically significant toxicities included eye disorders, skin and nail disorders, and elevations in LFTs. Two patients required dose reductions for toxicities. Two patients discontinued treatment before the 7-week evaluation, one for Grade 2 skin toxicity and the other for Grade 3 LFT elevation. All other toxicities were Grades 1 and 2. The other two patients continued treatment beyond a CR at 7 weeks but stopped after 3 and 4 cycles (after 11 and 16 weeks) for vision/skin toxicities and nail infection/mucositis, respectively. Conclusions: The oral FGFR inhibitor BGJ398 showed antitumor activity in a marker lesion study of patients with recurrent high-grade papillary NMIBC. Treatment was discontinued secondary to toxicities in all subjects, after a range of 4 to 16 weeks. Further investigation may be warranted for lower doses of oral BGJ398 or other formulations, such as intravesical delivery. Clinical trial information: NCT02657486. [Table: see text]

538 Background: Nested variant urothelial carcinoma (NVUC) is a historically aggressive variant of bladder cancer with management reliant on radical cystectomy given unclear effectiveness of chemotherapy. Given the aggressive behavior of NVUC, the identification of pathogenic and potentially targetable genomic alterations could suggest novel treatment paradigms for this rare histologic variant. Methods: Query of our institutional database identified 81 cases of clinically localized NVUC. Each case was re-reviewed by GU pathologists to confirm the histology. Targeted capture sequencing was performed to define the spectrum of genomic alterations in known and putative cancer genes. Clinical and sequencing data was then compared between the nested variant cases and a prospective cohort of clinically localized high-grade urothelial carcinomas, not otherwise specified (UC NOS). Results: On central re-review, tumors were classified as either classic NVUC (N=53) or high-grade urothelial carcinoma with nested features (N=28). A cohort of UC NOS patients (N=600) was used as a comparator. Median follow-up of the overall cohort was 4.3 years. Classic NVUC was associated with higher risk of cancer specific mortality compared to UC NOS on multivariate analysis (HR 1.76, p=0.04). Tumor mutational burden was lower in classic NVUC tumors versus UC NOS. Evaluation of 136 cancer-associated genes identified differences in mutational frequencies across the 3 cohorts (Table) with RhoA and FOXA1 alterations enriched in NVUC. Conclusions: Patients with NVUC have significantly higher risk of cancer specific mortality than patients with UC NOS. While there was significant overlap between the landscape of genomic alterations in NVUC and UC NOS tumors, NVUC tumors were enriched for RhoA and FOXA1 alterations among others.[Table: see text]

Purpose Immunotherapy is a standard of care for children with high-risk neuroblastoma, where bone marrow (BM) is the predominant metastatic site. Early response markers of minimal residual disease (MRD) in the BM that are also predictive of survival could help individualize patient therapies. Patients and Methods After achieving first remission (n = 163), primary refractory disease (n = 102), or second remission (n = 95), children with stage 4 neuroblastoma received anti-GD2 3F8 antibody immunotherapy. BM MRD before 3F8 treatment and after cycle 2 (postMRD) was measured using a four-marker panel (B4GALNT1, PHOX2B, CCND1, and ISL1) by quantitative reverse transcription polymerase chain reaction. Progression-free survival (PFS) and overall survival (OS) were estimated using the Kaplan-Meier method. Prognostic variables were tested in both univariable and multivariable analyses, and MRD markers were further assessed individually and in combination as binary composite (postMRD: 0 and 1) and as equal sum (postMRDSum: 0 to 4) using the Cox regression models, and their predictive accuracy was determined by the concordance index. Results When BM was evaluated after cycle 2, individual markers were highly predictive of PFS and OS. The prediction accuracy improved when they were combined in postMRDSum. A multivariable model taking into account all the variables significant in the univariable analyses identified postMRDSum to be independently predictive of PFS and OS. When the model for OS also included missing killer immunoglobulin-like receptor ligand, human antimouse antibody response, and the enrollment disease status, the concordance index was 0.704. Conclusion BM MRD after two cycles of immunotherapy was confirmed as an early response marker and a consistent independent predictor of survival.

e16035 Background: Eosinophils may influence anti-tumor immunity, having been associated with outcomes in CPI-treated melanoma. We sought to examine the association between PEC and outcomes of CPI-treated mUC. Methods: Independent CPI-treated cohorts were identified: discovery (n = 60, Teo et al JCO 2018) and validation (n = 111, non-overlapping CPI-treated mUC patients [pts] from 2014-19). Uni- and multivariate analyses were performed using Cox proportional hazard models to evaluate prognostic value of PEC on (i) overall survival (OS) and (ii) time on treatment (ToT). A platinum-treated metastatic cohort (n = 81, Teo et al, CCR 2017) was used as non-CPI comparator. Results: The cohorts showed comparable OS and pts characteristics, except higher proportion of females and lower neutrophil at baseline (Neut) in the validation cohort. In the discovery cohort, PEC as a continuous variable showed a significant positive association with OS (HR 0.01, 95% CI 0.00-0.29, P = .009). PEC of 100 cells/µL was set as the optimal cut-off point (Eos-L: < 100 cells/µL, n = 9; Eos-H: ≥100 cells/µL, n = 51). Eos-L showed inferior outcomes (OS: HR 3.98, 95%CI 1.85-8.56, P < .001; ToT: HR 2.45, 95%CI 1.17-5.10, P = .017). Similar association was seen in the validation cohort (Eos-L, n = 17; Eos-H, n = 94; OS: HR 2.51, 95%CI 1.31-4.80, P = .006; ToT: HR 2.22, 95%CI 1.29 - 3.80, P = .004). A multivariate model adjusting for other prognostic variables was created (table). In the chemotherapy cohort, continuous Eos or Eos-L was not associated with OS/ToT, with or without adjusting for known prognostic factors (P > 0.5). Conclusions: The association between Eos-L and shorter OS in CPI- but not chemotherapy-treated cohorts might represent a potential negative predictive biomarker for CPI in mUC. [Table: see text]

4559 Background: GC has a high overall response rate (ORR) but a high relapse rate in pts with untreated mUC. Inhibition of programmed death-ligand 1 (PD-L1) with A can lead to long-term survival, but single-agent ORR is modest. We report the outcomes of GC+A in a cohort pts with mUC. Methods: This study was designed to assess the safety of GC + A in 10 pts with untreated mUC prior to testing GC + A in a neoadjuvant study in pts with muscle-invasive disease. The primary endpoint was safety as assessed by a predefined dose limiting toxicity (DLT) rate during the first cycle in the first 6 pts. Total accrual goal was 10 pts to collect preliminary data on ORR and progression-free survival (PFS). RECIST 1.1 assessments were performed every 9 wks. Pts received 6 cycles of GC + A induction and then A maintenance every 3 wks. Results: No DLTs occurred during the first cycle in the first 6 pts. Grades 3-4 neutropenia and anemia occurred in 6/10 and 7/10 pts, respectively. Three pts required gemcitabine dose reductions for hematologic toxicity and 2 pts had febrile neutropenia. One pt discontinued cisplatin after 2 cycles for grade 3 hearing impairment but completed induction with gemcitabine and A. Only 1 pt discontinued study therapy due to treatment-related adverse events (AEs), including A-related grade 4 encephalopathy and grade 3 polyneuropathy. Three of 10 pts had visceral (liver or bone) metastases. Of the 10 pts, 1 pt is completing induction but meets initial criteria for partial response (PR), 8 pts had confirmed PR, and 1 pt had progressive disease (PD). Of 9 pts with confirmatory scans, the median PFS was 10.6 months (95% CI 6.7, N/A). Of 8 pts with confirmed PR, 5 eventually had PD, 1 has just completed induction, 1 remains without PD at 25 months, and 1 had consolidation surgery with a pathologic complete response and remains disease-free at 21 months. Conclusions: This 10 pt study met its primary safety endpoint. The neoadjuvant study is ongoing (NCT02989584). Although there were a substantial number of grade 3-4 toxicities, therapy was discontinued due to treatment-related AEs in only 1 pt. Immune correlative studies are ongoing. Clinical trial information: NCT02989584.

6084 Background: R/M ACC is a malignant neoplasm most commonly of salivary gland origin with no standard treatment. The impact of combined PD-1/CTLA-4 checkpoint blockade in R/M ACC is unknown. Methods: In a two-stage minimax phase II trial, pts with progressive R/M ACC (non-salivary primaries allowed) were enrolled and treated with nivolumab 3 mg/kg every 2 weeks plus ipilimumab 1 mg/kg every 6 weeks (1 cycle = 6 weeks). Imaging, using RECIST v1.1 response assessment, was scheduled to be performed approximately every 12 weeks. The primary endpoint was overall response rate (ORR = complete response [CR]+partial response [PR]) per RECIST v1.1. To detect a difference between an unacceptable ORR of 5% and a desirable ORR of 20% (one-sided type I error of 10%, power of 90%), at least 1 in the first 18 pts required an observed response. At least 4 responses of 32 total pts were needed to meet the primary endpoint. Treatment beyond progression of disease (PD) was allowed at the discretion of the investigator. A second cohort of pts with non-ACC salivary cancer is still accruing for separate analysis. Results: From 6/12/2017-6/20/2018, 32 pts were enrolled and evaluable for the primary endpoint. There was 1 confirmed PR in the first 18 pts, therefore enrollment of the second stage continued. ORR was 6% (2/32). One additional pt had an unconfirmed PR (-31% regression before CNS PD). For best overall response, there were 2 PRs, 15 SD, and 11 PD. Four pts never reached a first disease assessment: 3 due to death from clinical PD and 1 was removed for toxicity. Six pts discontinued the trial for toxicities (Grade 4 (G4) neutropenia/sepsis and G3 adrenal insufficiency (1), G2 hypophysitis (2), G3 arthritis > 7 days (1), G3 colitis (1), and G3 hepatitis/G4 creatinine kinase (CK) elevation (1)). The 2 confirmed PRs consisted of -73.1% and -58.4% regressions, with a duration of therapy of 18.4 and 7.8 months, respectively (treatment ongoing for both). Conclusions: The study did not meet its primary endpoint, though the responses observed were dramatic. Paired biopsy and peripheral blood samples will be analyzed to elucidate insights into mechanisms of response and resistance to dual checkpoint blockade. Clinical trial information: NCT03172624.

Purpose Although gemcitabine and carboplatin (GCa) is a standard option for patients with advanced urothelial cancer (UC) who are ineligible for cisplatin, outcomes remain poor. This trial evaluated the efficacy and safety of bevacizumab with GCa in advanced UC. Patients and Methods Patients with Karnofsky performance status of 60% to 70%, creatinine clearance less than 60 mL/min, visceral metastasis, or solitary kidney were eligible and received a lead-in dose of bevacizumab 10 mg/kg followed 2 weeks later by gemcitabine 1,000 mg/m2 on days 1 and 8 and carboplatin at area under the [concentration-time] curve (AUC) 5.0 or 4.5 and bevacizumab 15 mg/kg on day 1 every 21 days for six cycles. Patients achieving at least stable disease (SD) continued bevacizumab 15 mg/kg every 21 days for 18 additional cycles. The study was powered to detect a 50% improvement in median progression-free survival (PFS) over a historical control. Results Fifty-one patients, median age 67 years (range, 42 to 83 years), were enrolled onto the study and were evaluable for toxicity. Twenty (39%) experienced grade 3 to 4 toxicity, and 10 (20%) had thromboembolic events (deep venous thrombosis or pulmonary embolism). Four received one or fewer cycles leaving 47 evaluable for outcomes. Twenty-three (49%) achieved response (three complete; 20 partial), and 11 had SD. Median PFS was 6.5 months (95% CI, 4.7 to 7.8 months); PFS was greater in the carboplatin AUC 5.0 group (P = .04). Median overall survival (OS) was 13.9 months. Conclusion The 95% one-sided lower confidence bound of 4.77 months for median PFS did not meet the predesignated PFS of more than 4.8 months considered sufficient for further study. Median OS was greater than expected. An ongoing phase III trial in patients who are eligible for therapy with cisplatin will define the role of bevacizumab in UC.

4515Background: Growing evidence highlights the critical role of the TME for RCC biology and response to systemic therapy. We used integrated molecular profiling to characterize immune & stromal TM...

522 Background: PUC is characterized by E-cadherin loss, diffuse growth pattern and aggressive natural history. Management of this entity is controversial. This retrospective study aims to evaluate the impact of PUC histology on clinical outcomes with ctx compared to UC-NOS. Methods: Consecutive cases of nonmetastatic PUC were identified as either (1) localized disease (LD: cT2-3 cN0) or (2) locally-advanced (LA: fixed bladder or radiographic nodal disease). All cases were reviewed by a GU pathologist to confirm PUC histology per WHO classification (2016). A separate cohort of neoadjuvant(NAC)-treated UC-NOS (Tully et al, ASCO GU 2014) served as comparator for clinical outcomes. Kaplan-Meier estimates and logrank test were used for analysis of recurrence free (RFS) and overall survival (OS) defined from date of cystectomy (RC). Results: Between 2000 and 2017, eighty one PUC were identified, with median age of 65 years (range: 22 - 84) and 65% male. Thirty patients (pts) had up-front RC; 51 pts had ctx. In the former group, all pts had LD but upstaging was seen on post-op pathology: 63% pT3, 17% pT4 and 37% pN+. Among ctx pts, 33/51(65%) were LD and had NAC: 70% had ctx doublet and 24% triplet; 82% had 4 cycles and 18% had up to 6 cycles. Response ( < pT2 pN0) was seen in 7/33(22%), four (12%) of which had no residual disease (pCR). Six RC were aborted after NAC for pT4b disease. NAC response was not associated superior RFS (NR vs. 17.3 months; HR 0.52 95% CI 0.18 – 1.47, p = .218) or OS (41.3 vs. 25.5 months; HR 0.65 95% CI 0.23 – 1.87, p = .428). Compared to UC-NOS, PUC responders had significantly inferior outcomes (table). In LA cohort, majority had triplet chemotherapy (72%) and up to 6 cycles (89%). Response was seen in 4/18(22%) pts, with two pCR (11%). Five RC were aborted for pT4b disease. Median RFS and OS were 5.8 and 10.5 months, respectively, with 19% alive at 2-years. Most common sites of recurrence/progression for PUC were peritoneal (42%) and locoregional invasion (26%). Conclusions: Outcomes for PUC are poor with conventional NAC-RC approach. [Table: see text]

469 Background: Genomic characterization of urothelial bladder cancer (UBC) may help to identify alterations associated with tumor stage, novel therapeutic targets and biomarkers to predict outcomes. Methods: 454 UBC patients were sequenced using next generation sequencing assay (MSK−IMPACT). 264 patients (123 NMIBC, 112 MIBC and 29 metastatic) were chemotherapy naïve primary tumors. Results: Patients whose primary tumors harbored TP53, RB1, TP53/MDM2, and cell cycle pathway alterations more frequently presented with advanced disease whereas those with tumors containing FGFR3, and RTK/RAS/RAF pathway alterations generally presented with NMIBC (all FDR < 0.001). TP53, RB1, TP53/MDM2, and cell cycle pathway alterations were associated with worse overall and metastasis free survival whereas FGFR3 alterations were associated with more favorable overall and metastasis free survival (all FDR < 0.05). RTK/RAS/RAF pathway alterations were associated with more favorable metastasis free survival (FDR = 0.01). MIBC with no DDR alterations, DDR alterations of unknown significance and DDR deleterious alterations had pathologic downstaging rates of 37%, 48% and 81% with platinum-based neoadjuvant chemotherapy (p = 0.013). Patients with ERCC2 mutations had significantly higher pathologic downstaging with platinum-based neoadjuvant chemotherapy compared to patients without ERCC2 mutations (p = 0.025). MIBC with DDR alterations had better overall survival with neoadjuvant chemotherapy compared to MIBC without DDR alterations (p = 0.04). Of note, MIBC cases with and without DDR alterations have similar overall survival in the absence of neoadjuvant chemotherapy (p = 0.78). Conclusions: TP53, RB1 and FGFR3 alterations as well as TP53/MDM2, cell cycle and RTK/RAS/RAF pathway alterations showed an association with tumor stage and patient outcomes. MIBC with DDR alterations had higher pathological downstaging as well as better overall survival compared to MIBC without DDR alterations with platinum−based neoadjuvant chemotherapy. In absence of neoadjuvant chemotherapy, MIBC with and without DDR alterations have similar patient outcomes.

4509 Background: Somatic DDR alts are associated with increased mutation load (ML) and improved clinical outcomes for platinum-treated mUC (Teo et al, CCR 2017). We examined the relationship between DDR alts and response to PD1/PDL1 blockade. Methods: mUC pts enrolled to phase 2 trials of atezolizumab or nivolumab who had targeted exon sequencing of 410 genes (MSK-IMPACT) were identified. Pts were dichotomized based on presence of alts in a panel of 34 DDR genes. Analyses were performed based on: (1) any DDR alts and (2) deleterious DDR alts (frameshift, splice site, nonsense or Hotspot point mutations). Study endpoint was overall response (OR) per RECIST. ML was defined as total number of nonsynonymous mutations by MSK-IMPACT. Fisher exact, Wilcoxon rank sum, and stratified logistic regression were used. Results: Fifty two pts were identified (atezo: n=18, nivo: n=34). Median age was 67 years (range: 32 – 84) and majority (44) was male. Median platinum-free interval was 10.2 months (range: 0.3 – 150.4). DDR and deleterious DDR were seen in 25 (48.1%) and 14 (26.9%) pts (including 2 MSI and 1 POLE). OR rate was 46.2%. Responses were associated with DDR alts but not with age, gender, treatment, platinum-free interval or ML (table). In univariate logistic regression model, DDR status was associated with OR (pmedian: p=.027; ≤median: p=.023), indicating that the effect of DDR was independent of ML. Conclusions: DDR alts appeared to be associated with OR to PD1/PDL1 blockade and should be integrated into future validation efforts along with other potential predictors of response. [Table: see text]

302 Background: We examined a cohort of index pre-treatment NMIBC tumors using Next Generation Sequencing to identify genetic alterations with potential clinical implications. Methods: 105 patients on a prospective IRB-approved protocol had their pre-treatment index NMIBC tumor and matched germline DNA sequenced with a 341 cancer-associated gene panel in a CLIA-certified clinical laboratory. A genitourinary pathologist reviewed representative H&E slides to confirm grade, stage, and urothelial histology. Restaging TUR was performed in all HGT1 tumors. Results: To characterize the genomic landscape of NMIBC, we analyzed 105 tumors across the disease spectrum including LGTa (n = 23), HGTis (n = 12), HGTa (n = 32) and HGT1 (n = 38). The most frequently altered genes in NMIBC were the TERT promoter (74%), FGFR3 (50%), KDM6A (47%), ARID1A (28%), PIK3CA (27%), KMT2D (24%), STAG2 (21%), and CDKN2A (17%). 81% of tumors had inactivating alterations in a chromatin-modifying gene. Alterations in the RTK/RAS/PIK3 pathway occurred in 83% of tumors, including 58% of high-grade NMIBC having alterations in either ERBB2 or FGFR3. Of the 105 patients, 62 were treated uniformly with a 6-week induction course of BCG without maintenance. We investigated all genes altered on the 341-gene panel in at least 5 patients for their association with recurrence after BCG therapy in this 62 patient cohort. On cox-regression analysis, only truncating mutations in the chromatin-modifying gene ARID1A were associated with recurrence after BCG (HR = 3.14 [95%CI = 1.51, 6.51] p = 0.002). This remained significant when adjusting for multiple comparisons (p = 0.04) and when including ARID1A missense mutations of unknown significance (p = 0.002). Conclusions: Next Generation Sequencing of index pre-treatment NMIBC tumors identified an association between ARID1A mutations and recurrence after BCG therapy. Further investigation is needed to determine whether ARID1A mutations are a potential predictive/prognostic biomarker or therapeutic target. Moreover, most NMIBC tumors had at least one potentially “actionable” alteration that could serve as a target in rationally designed trials of intravesical or systemic therapy.

356 Background: M-MDSCs promote tumor progression through complex mechanisms, including immunosuppression and the production of mediators of angiogenesis and invasion. M-MDSCs are associated with poor outcomes in a number of malignancies. We conducted an exploratory analysis enumerating M-MDSCs (Lin-CD14+/HLA-DRlow/-) in the blood of pts with mUC and assessed for assay variability and correlation with clinical outcomes. Methods: Whole blood was collected at a single time point for each pt and stabilized in Cyto-Chex tubes. M-MDSC% was calculated by flow cytometric analysis of HLA-DR expression on CD14+ monocytes using an MSKCC developed computational algorithm-based approach (Kitano et al. 2014). Individual samples were run in quadruplicate at MSKCC. The mean MDSC% was used in subsequent analyses and the replicate standard deviation (SD) was considered a reflection of intra-assay variability. Clinical variables were collected and pts followed for OS, which was calculated from time of sample collection. Results: 21 pts with mUC who progressed after prior chemotherapy were included. At the time of collection, pts were on clinical trials with immune checkpoint blockade (n=13) or targeted therapy (n=1), receiving salvage chemotherapy (n=2), or awaiting treatment on clinical trials with immune checkpoint blockade (n=4) or targeted therapy (n=1). The median follow up from the time of collection in surviving patients (n=15) was 11.1 months (range: 10.5-11.5). Median OS was not reached. Mean (SD) M-MDSC% was 29.2 (4.92). The range of replicate SD was 0.13-1.05. M-MDSC% was not higher in pts with visceral mets (p=0.109). Pts with above median M-MDSC% had worse OS (p=0.01; 6 patients died during follow up, all with above median M-MDSC%). Conclusions: In a heterogeneous cohort of pts with mUC, the enumeration of M-MDSCs in stabilized whole blood was feasible and demonstrated marked inter-patient but not intra-assay variability. Pts with higher M-MDSC% values had worse OS. Additional characterization of M-MDSCs in larger cohorts and in pts receiving immunotherapy is ongoing and may have important prognostic and therapeutic implications in mUC.

294 Background: SCCB is a rare, aggressive subtype of bladder cancer. We reviewed our center’s experience in SCCB management and investigated potential markers of clinical outcomes. Methods: SCCB cases were identified and re-reviewed for confirmation by detailed histologic evaluation. A subset of SCCB was sequenced with MSK-IMPACT ( > 279 genes) (Cheng et al J Mol Diagn 2015). Clinical stage, treatment response and proportion of neuroendocrine component (%NE) were correlated with progression-free (PFS) and overall survival (OS), calculated from the time of cystectomy. Distribution of somatic alterations (alts) was compared between responders ( ≤ ypT1pN0) and nonresponders. Results: Between 1985 and 2015, 214 SCCB patients (pts) were identified. Median age was 68 years (range: 38 – 94) and 78% were male. Of 199 pts with sufficient data, 147 (74%) were nonmetastatic (M0) and 52 pts (26%) had metastatic disease (M1). 14 pts did not receive SCCB-directed therapies. Of M0 pts, 108 underwent cystectomy: 71 received neoadjuvant (NAC), 14 received adjuvant, 14 had definitive chemoradiotherapy and 23 did not receive chemotherapy. Of M1 pts, 47 received chemotherapy. Cystectomy only pts were older than NAC pts (76 vs 67 years, p < .01). NAC improved PFS and OS compared to cystectomy only (HR 0.55, p = .04 and HR 0.41, p < .01) although only OS remains significant after adjusting for age (p = .01). For NAC pts, pT0pN0 and ≤ pT1pN0 rates were 38% and 48%, respectively. Responders had superior PFS (HR 0.26 p < .01) and OS (HR 0.32 p = .02). Median %NE was 100 (range: 10 – 100) and did not correlate with NAC response. %NE ≥ 95 is more prevalent among M1 pts (85% vs 68%, p = .04). Genomic data were available for 31 NAC pts (44%). Responders were enriched with alts in ERBB2 (33% vs. 0%, p = .02), PI3K pathway ( PIK3CA, PTEN, AKT1, TSC1, TSC2) (33% vs. 6%, p = .02) and histone acetyltransferases (60% vs. 25%, p = .07). Of 22/31 tumors sequenced with a larger gene set (n = 341), ERCC2 and RBM10 alts were enriched in responders (54% vs. 11%, p = .07) and nonresponders (44% vs. 0%, p = .02), respectively. Conclusions: In SCCB, NAC improves survival and response might be characterized by specific somatic alts. M1 pts had higher %NE but it is not predictive of outcome.

TO THE EDITOR: We suggest that the far-reaching conclusion based on a landmark phase III trial concerning high-risk neuroblastoma (HR-NB) should be reconsidered, given the major statistical error in the recent update. 1 The initial report in 1999 prompted the worldwide adoption of myeloablative therapy with autologous stemcell transplantation (ASCT) as the standard of care for patients with HR-NB in first remission. Now, 15 years later, we learn that this intervention made no statistically significant difference in overall survival (OS) when tested in a randomized fashion (P .39 by log rank; P.08 at 5 years). 1 OS stands out as the gold standard for evaluation of treatment efficacy, the acid test for US Food and Drug Administration drug approval, and the driving force for advances in cancer therapeutics. 2 OS is 100% accurate for event and time; it takes into account both safety(toxiccomplications)andefficacy.Event-freesurvival(EFS)isa surrogateendpointthatallowsanearlierassessmentofresults,butits validity requires confirmation, either through correlation with OS or by meta-analysis. 2-4 Unless improvement in quality of life can be documented, delaying asymptomatic events without prolonging survival is not clinically meaningful. 3 In short, a surrogate end point should be used with caution as the basis to establish standard of care. In this landmark study, 1 why did OS fail to show significance, whileEFSdid?Thisdiscrepancypointstothewelcomepossibilitythat

10567Background: Gangliosides GD2 and GD3 are the tumor-associated glycolipid antigens found in a broad spectrum of human cancers, including pediatric embryonal tumors (neuroblastoma, retinoblastom...

5011Background: Individual DDR genetic alterations have been reported to be predictive for neo chemotherapy response in UC. We sought to identify whether alterations in a DDR gene set correlate wit...

436 Background: Cisplatin-based chemotherapy before radical cystectomy (RC) improves survival in pts with MIBC. DD-GC therapy is active in the metastatic setting [6 cycles (cy), 18 months median survival; Bamias et al, 2012)] and as neo therapy (3 cy, 44% ≤ pT1 rate; Plimack et al, 2014), but the optimal dose and number of cy of neo therapy has not been defined. We prospectively evaluated the activity and safety of 6 cy of neo DD-GC over 12 weeks in MIBC. Methods: Pts with T2-4aN0 disease received six 14-day cy of DD-GC as follows: G 2500 mg/m2 day 1, C 35 mg/m2 days 1 and 2, pegfilgrastim day 3. RC with bilateral pelvic lymph node dissection was planned within 8 weeks of DD-GC completion, regardless of clinical response. The primary endpoint was pathologic response ( ≤ pT1) rate ≥ 55% (exact Binomial one-sided test). Pts not undergoing RC were deemed non-responders regardless of clinical stage after DD-GC. Pts receiving < 3 cy were inevaluable and replaced. All pts were evaluable for toxicity. Results: 49 pts (40 male) were enrolled. Median age was 64 (range: 37-78). Clinical stage was T2N0 (32 pts), T3N0 (12 pts), and T4aN0 (5 pts). Toxicities resulting in cy delay and/or dose modifications included thrombocytopenia (9 pts), renal insufficiency (5 pts), vascular access complication (2 pts), ototoxicity (1 pt), significant urinary symptoms (1 pt), and transient ischemic attack (1 pt). Three pts are inevaluable for the primary endpoint ( < 3 cy). As of 9/7/15, 2 pts are pending RC. Of the 44 pts evaluable for response to date, 31 completed 6 cy of DD-GC, 6 pts completed 5, 3 pts completed 4, and 4 pts completed 3 (median: 6 cy). The median time to RC was 46 days. Four of 44 pts did not undergo RC (consent withdrawal, pt refusal, disease progression prior to RC, death from other causes). Trial accrual has closed with completion of clinical and pathologic data expected by 11/1/15. Of 40 pts with RC pathology available to date, 24 (60%) were ≤ pT1 and 7 (18%) were pT0. Conclusions: Six cy of DD-GC is an active well-tolerated neo chemotherapy regimen in pts with MIBC. The pathologic response rate is encouraging. Thrombocytopenia was the most common toxicity resulting in cy delays/dose modifications. Clinical trial information: NCT01589094.

473 Background: Approximately 30% of patients with advanced germ cell tumor (aGCT) will progress after first-line chemotherapy. Nearly half of these patients will die of progressive GCT. We describe potentially actionable mutations in a cohort of patients with platinum-resistant aGCT through targeted sequencing. Methods: 76 patients with cisplatin-resistant (CR) disease were sequenced using the Memorial Sloan Kettering Integrated Mutation Profiling of Actionable Cancer Targets (MSK-IMPACT) assay that examines 341 cancer-related genes. Patients were categorized as CR if they met any of the following criteria: 1) incomplete response to first-line cisplatin-based chemotherapy; 2) nonteratomatous tumor progression after standard chemotherapy; 3) nonteratomatous GCT identified at postchemo surgery. We grouped all somatic mutations into core signal transduction pathways or canonical cell functions to identify potential precise targets for therapy. Results: The majority of patients had nonseminoma histology (n = 64, 84%). International Germ Cell Cancer Collaborative Group risk was good, intermediate, and poor in 34%, 13%, and 53% of patients. 17 patients died of disease. In total, 51 potentially actionable alterations were identified in 36/76 (47%) patients. In the TP53 pathway, 7 MDM2 amplifications and 4 MYCN amplifications that may sensitize to nutlin-3 inhibitors were identified. Within the receptor tyrosine kinase pathway, 3 KIT mutations, 1 KDR amplification, and 1 MET amplification were seen that may sensitize to tyrosine kinase inhibitors. Eleven KRAS mutations, 3 NRAS mutations, 3 BRAF mutations, and 2 RAC1 mutations were see among the RAS pathway with preclinical data suggesting efficacy towards respective inhibitors. Actionable targets were also among the PI3-K, WNT, and cell cycle pathways. Potential targets with chromatin modifying or tumor suppressor functions were also seen. Conclusions: We describe actionable alterations that may guide treatment selection in a significant proportion of patients with CR aGCT. Targeted sequencing of these patients may allow us to enrich future clinical trials with patients whose tumors harbor alterations in the drug target of interest.

605 Background: Clear cell renal cell carcinoma (ccRCC) is among the most highly infiltrated tumors despite a relatively low mutation burden. Despite success in treating subsets of these patients with immunotherapy, the high immune-infiltration of ccRCC remains a conundrum that has not been sufficiently addressed. Methods: We utilized an original RNAseq-based aggregate immune score (Senbabaoglu Y, bioRxiv, 2015) to compare immune infiltration levels across 20 tumor types profiled in The Cancer Genome Atlas (TCGA) and validated our results with an independent cohort of ccRCC patients (Sato Y, Nat Genet, 2013). We also evaluated the immunogenic effect of intratumoral heterogeneity (ITH) through an analysis of clonal architecture in ccRCC tumors from both of the aforementioned data sets using the SciClone (Miller C, PLoS Comput Biol, 2014) algorithm. Results: We identified ccRCC as an immunogenic outlier that is unique from other highly infiltrated tumors in its exceptional overexpression of antigen presentation machinery compared to normal tissue. In a focused decomposition of immune cell levels within the 415 ccRCC tumors from TCGA, three unique clusters of tumors emerged primarily separated by varying degrees of T cell infiltration; termed (1) T-cell-enriched, (2) heterogeneous, and (3) non-infiltrated groups. These clusters were validated in an independent ccRCC cohort of 101 patients. A comparison of the degree of clonality with immune infiltration levels revealed that tumors with less ITH (i.e. fewer subclones) had significantly higher immune infiltration in both the TCGA and the validation cohort . Conclusions: Our findings provide novel insight into the immune landscape of ccRCC tumors and explore a potential mechanism for the immunogenicity of these tumors. Such information can potentially be used to aid in predicting therapeutic response to immunomodulatory agents.

4510 Background: 20% of patients (pts) with aGCT progress after initial chemotherapy (chemo), displaying at least partial cisR. We sought to characterize the genetic alterations in aGCT and identif...

4532 Background: Platinum-based chemotherapy (PBC) is standard of care in the neoadjuvant and metastatic UC settings. Recent work identified mutations in ERCC2, RB1, FANCC and ATM as potential pred...

322 Background: Gemcitabine and split-dose cisplatin (GC) is standard first-line therapy for patients with advanced urothelial cancer (UC). Everolimus is an inhibitor of the PI3K/AKT/mTOR pathway, dysregulated in ~35% of UCs. The drug has been shown to have single-agent activity in phase II trials of advanced UC. The safety of combining everolimus with GC in the first-line setting for UC is unknown. Methods: Previously untreated cisplatin-eligible patients with advanced UC were enrolled. Patients received gemcitabine 800 mg/m2 and cisplatin 35mg/m2 on days 1 and 8 with everolimus at dose levels (DL1: 5 mg QOD, DL2: 5 mg daily or DL3: 10 mg daily) for six 21-day cycles. Patients with at least SD continued maintenance everolimus until progression. Restaging evaluations were performed every 2 cycles. The primary objective was to establish the dose-limiting toxicity (DLT) and maximum tolerated dose (MTD) of the three-drug combination. Results: 12 patients were enrolled, 7 with bladder and 5 with upper tract primaries. 4/12 patients completed at least 6 cycles. 0/3 patients at DL1 had DLTs. 3/3 patients at DL2 had DLTs (pancytopenia, hypersensitivity reaction to everolimus and anemia). Following de-escalation to DL1, 2/6 patients in a new DL1 cohort had DLTs (neutropenia and diarrhea, 1/6 patients inevaluable). Grade 3/4 events included anemia (67%), neutropenia (42%), lymphopenia (42%), thrombocytopenia (25%), urinary tract infection (25%), hypomagnesemia (17%), and hypophosphatemia (17%). Tumor responses included: 4 PR (33%), 5 SD (42%), 1 PD (8%), and 2 inevaluable (17%). 2 patients with PR underwent consolidative surgery but eventually progressed. Responses by MSKCC Risk Scores of 0, 1, and 2 were seen in 3/5 (60%), 1/6 (17%), and 0/1 patients, respectively. Median PFS was 3.9 months (95% CI, 3.7-no reached) and median OS was 10.8 months (95% CI, 6.9-not reached). Conclusions: MTD was reached at DL1 for combination gemcitabine, split-dose cisplatin and everolimus, with 2/8 evaluable patients exhibiting DLTs at this dose level. Clinical trial information: NCT01182168.

373 Background: Cisplatin-based chemotherapy is the standard of care first-line therapy for locally advanced or metastatic urothelial cancer (mUC) with reported response rates (ORR) of 50%, median progression free survival (mPFS) 7 months and median overall survival (mOS) 14 months in phase III trials using cisplatin 70 mg/m2 every 3 weeks (q3wk). Many patients cannot receive cisplatin due to impaired performance status, renal dysfunction, or other co-morbidities. Carboplatin is an alternative but is associated with lower response rates and shorter survival. Administering cisplatin 35 mg/m2 day 1+8 (i.e., split dose) every 3 weeks may be associated with lower toxicity and allow more patients receive the drug. In this retrospective analysis we examined the safety and efficacy of cisplatin/gemcitabine using split dose cisplatin. Methods: Patients who received split dose cisplatin/gemcitabine between 2007-2013 were identified. We recorded baseline demographics, clinical characteristics, and number of cycles administered. Responses were determined retrospectively using RECIST v1.1; OS and PFS were assessed using the Kaplan-Meier method. Results: 50 patients were identified and baseline characteristics are outlined in Table 1. The median number of cycles administered was 4 (range 1-7). 25 patients (50%) discontinued therapy due to completion of planned course, 14 (28%) due to progressive disease and 11 (22%) due to toxicity. ORR was 54%, mPFS was 5.4 months (95% CI: 4.2 – 8.8) and mOS was 13.3 months (95% CI: 9.6 – 21.9). Conclusions: In this analysis, split-dose cisplatin was administered safely in patients who were deemed ineligible for the q3wk regimen and efficacy did not appear to be compromised. The split-dose regimen may allow more patients tolerate cisplatin, the most active FDA-approved drug in this disease. [Table: see text]

4510 Background: Cisplatin-based combination chemotherapy is the standard of care for patients with muscle invasive urothelial carcinoma. Pathologic downstaging to pT0/pTis after neoadjuvant cispla...

355 Background: NAC and RC-PLND improves survival in MIBC and GC is a standard NAC option. However, little is known about GC efficacy endpoints and the individual contribution of NAC and surgery. Methods: Pts with clinical T2-T4aN0M0 MIBC treated from 1/2000 to 10/2012 with a planned 4 cycles of GC plus RC-PLND within 90 days (D) of NAC were evaluated retrospectively for the number (#) of cycles, dose delivered, D from end of NAC to RC-PLND, margin status, LN status and # of LN identified. Post-NAC pathologic endpoints included complete response (pT0), residual Non-MIBC disease (pTa/Tis/T1;N0) and ≥MIBC disease (≥pT2N0). Associations with overall survival (OS) and disease-free survival were analyzed using Cox regression; non-linear associations with # of resected LN used linear and quadratic terms. Results: 154 pts met inclusion criteria. 5-year (yr) OS was 61% (95% CI 53-71%). Post-NAC pT0 was achieved in 21% (32/154) and Non-MIBC in 25% (39/154 - pTa (2), pTis (25), pT1 (12)). Post-NAC pT0 and Non-MIBC had similar 5-yr OS (85% and 89%, respectively) and combined (

342 Background: Cisplatin-based NAC in UC confers a survival benefit, but prediction of which patients (pts) will benefit is not possible. We attempted to validate pharmacogenomic (PGx) markers of cisplatin sensitivity derived from a cell-based model and previously associated with response in a population of pts receiving NAC. Methods: Three germline single nucleotide polymorphisms (SNPs) were previously associated with pathologic response at cystectomy in a single-institution discovery set of 59 cT2 UC pts. These SNPs were tested for association with NAC response in a prospectively-identified, multi-institution, independent cohort. The primary analysis tested for association between rs244898 and rs7937567 and pathologic complete response (pT0). A replication set of 134 pts would provide 80% power to detect effects of both SNPs at p=0.05 independently using a multivariate logistic model. Results: N=146 pts with ≥cT2 N0 disease were identified from three institutions. All pts received ≥3 cycles of cisplatin-based NAC and underwent definitive surgery. Rates of pT0 and 5 odds ratio of effect on pT0 in the discovery set, neither was associated with achievement of pT0 in the replication set. The third SNP (rs10964552) which was associated with pathologic downstaging in the discovery set, also failed to replicate. All three SNPs were in Hardy-Weinberg equilibrium, and minor allele frequencies were similar between discovery and validation. Conclusions: Three germline SNPs previously associated with platinum sensitivity in pre-clinical and clinical models were not associated with NAC response in a large replication cohort of UC pts. Reasons for failure may include unmeasured clinical differences between the pt cohorts, a higher proportion of MVAC use in the replication set compared to gemcitabine/cisplatin in discovery pts, or simply spurious association in the discovery cohort. These results emphasize the need for replication when evaluating PGx markers, and demonstrate that multi-institutional efforts are feasible and will likely be necessary to achieve advances in UC PGx.

395 Background: The exonic single nucleotide variant rs11762213 located in the METoncogene has recently been identified as a prognostic marker in clear cell renal cell carcinoma (ccRCC) (Schutz et al, Lancet Oncol 2013). We sought to validate this finding and explore the biologic implications using The Cancer Genome Atlas (TCGA) cohort. Methods: The variant call file (VCF) and expression data was available for 272 patients. We then extracted the data for rs11762213 as follows for the "normal" sample: allelic fraction for variant allele > 80% is homozygous variant while allelic fraction for variant allele 30% to 70% is heterozygous. Paired tumor-normal materials, genomic data (whole exome, RNAseq, and DNA methylation) and clinical information were acquired from publically available TCGA datasets. Kaplan-Meier method was used to estimate the survival probabilities. Cancer specific survival (CSS) was analyzed using the competing risk method and Cox proportional hazard regression was used for analysis of time to recurrence. Multivariate competing risk models were also fitted in the TCGA cohort in order to adjust for the Mayo Clinic SSIGN score. Results: Overall, the variant allele of rs11762213 was detected in 10.3% of the cohort and was associated with higher nuclear grade (p=0.03) and trended toward higher clinical stage (p=0.07). After adjusting for the prognostic SSIGN score, the risk allele remained a significant predictor for adverse cancer specific survival (CSS; p

322 Background: No standard therapy exists for platinum refractory mUC. Single agent pemetrexed (Pmt) had objective response rates (ORR) of 8% and 28% in two phase II studies (enrolling 13 and 47 patients) and is commonly used in this setting. To address the difference in reported ORRs, we performed a retrospective analysis of Pmt use at MSKCC to evaluate objective response in a larger cohort of patients. A secondary aim was to explore whether neutrophil-lymphocyte ratio (NLR) added prognostic value beyond known factors (time from prior chemotherapy (TFPC), ECOG performance status (PS), liver metastases, and hemoglobin) in the UC salvage setting. Methods: Patients who received Pmt for platinum refractory mUC between 2008 and 2013 were identified. Baseline demographics, clinical characteristics, prior therapies, Pmt dose, and number of cycles were recorded. ORR was determined according to RECIST 1.1. Kaplan-Meier method and Cox regression were used to analyze associations with overall survival (OS). Results: 135 patients were identified with median age 66 (range 45-88), male 76%, ECOG 0 in 14% / 1 in 54% / ≥2 in 32%. Pmt was administered as 2nd-line chemotherapy in 56% / 3rd line in 30% / ≥4th line in 14%. ORR was 7% with median duration of response 10.2 months. There was no significant difference in ORR by line of therapy or PS. Median progression free survival was 2.4 months and median OS was 6.6 months. In this dataset, TFPC was not prognostic, while liver metastases, PS, and hemoglobin were prognostic. Higher NLR was significantly associated with worse OS independent of other known factors. Conclusions: In the largest reported series to date, Pmt had an ORR of 7% in metastatic UC regardless of line of therapy or ECOG performance status. This limited activity highlights the urgent need to develop novel therapeutic strategies. NLR was identified as an independent prognostic factor in this setting. [Table: see text]

343 Background: Germline single-nucleotide variations (SNVs) in DNA repair genes may have a role in response to platinum-based chemotherapy in bladder cancer (BC). As a pilot project we created a discovery cohort using an extreme phenotype model of BC response to neoadjuvant chemotherapy; we identified candidate SNVs in DNA repair genes via whole exome sequencing, and sought to replicate these findings in a validation cohort. Methods: We sequenced exomes of a discovery cohort with ≥ cT2 BC and neoadjuvant chemotherapy [6 complete responders (CR, pT0N0 and 1 year disease free interval) vs. 9 non-responders]. We identified SNVs within 377 DNA repair pathway genes. We filtered for SNVs with high impact effects, and moderate impact missense variants. The discovery phase resulted in 16 candidate SNVs which we genotyped (matrix-assisted laser desorption/ionization-time of flight) within a validation cohort of 303 BC patients treated with platinum-based neoadjuvant or first-line chemotherapy. Primary outcome was pathologic (neoadjuvant) or radiographic (first-line) CR, with secondary outcome CR+partial response (PR; RECIST criteria or pathologic downstaging). We used additive genetic model and logistic regression to analyze association between individual SNVs and both CR and CR+PR. Multivariate models included MSK risk score for CR or chemotherapy (cis- vs carboplatin) for CR+PR. Results: We included 303 patients (186 neoadjuvant, 117 first-line; 272 cisplatin, 31 carboplatin), median age 64, including 67 (22%) CR and 153 (50%) CR+PR. Genotyping was adequate for analysis in 14 of 16 SNVs—none were associated with CR. SNVs associated with CR+PR were rs2072454 (EGFR, p=0.031 unadjusted, p=0.035 adjusted for chemotherapy) and rs1805321 (PMS2, p=0.014 unadjusted, p=0.019 adjusted for chemotherapy). Findings were not significant after adjusting for multiple comparisons. Conclusions: Using an extreme phenotype model we identified SNVs in genes of the DNA repair pathway that may have an association with platinum-based chemotherapy response in BC. A larger validation study is required to increase power and reduce the impact of multiple testing prior to confirming significance.

316 Background: VTE occur in ~13% of pts receiving Cis-based therapy for UC. Cb-based therapy is considered less thrombogenic but no definitive data exist to support this conclusion. While Bev added to chemotherapy (Ch) increases VTE in other tumors, VTE impact when added to UC Ch is unknown. This study evaluated the incidence of and etiologic factors for VTE in UC pts treated with GCb, GCbBev and GCis. Methods: UC pts treated from 6/2006 to 6/2010 on a GCbBev protocol were analyzed prospectively. Planned therapy was 6 cycles of GCbBev then Bev alone; pts who had ≥ 3 cycles were included. Similar UC pts treated with GCb or GCis during the same time were retrospectively evaluated. Type of VTE and potential contributing clinical factors were collected. VTE was defined as pulmonary embolism, vascular thrombosis, myocardial infarction or cerebral vascular accident. Associations with Ch regimen were tested using Fisher’s exact test or linear regression. Factors associated with VTE were analyzed using conditional logistic regression stratified by Ch regimen. Results: 198 pts were analyzed. VTE occurred in 13/51 (26%) GCbBev pts, 22/92 (24%) GCb pts, and 8/55 (15%) GCis pts. Age (≤65 vs >65; p

300 Background: Little is known about germline variants predictive of response to chemotherapy in bladder cancer (BC). We investigated germline variants in patients (pts) treated with neoadjuvant gemcitabine/cisplatin (GC) under the hypothesis that responders and non-responders may differ in rare alleles predictive of treatment outcome. Methods: The Hiseq platform (≥ 20x coverage) was used to sequence the exomes of muscle-invasive (≥cT2) BC pts representing “extreme phenotypes” defined by response to neoadjuvant GC (complete responders vs. non responders). Responders were pT0N0 at cystectomy and remained disease free >1 year. Non-responders had residual disease in the bladder at cystectomy and metastatic disease either at cystectomy or within 1 year. Rare, potentially deleterious single nucleotide variants (SNVs) unique to these phenotypes were confirmed with Sequenom and validated in an independent group of BC pts treated with GC in the neoadjuvant or first line setting. One-sided Fisher’s exact test and multivariate logistic regression were used to estimate association between SNVs and response. Results: The discovery cohort had 6 responders (4 male) with a mean age of 62 and 9 non-responders (6 male) with a mean age of 63. Filtered exome sequences identified 93 rare/potentially deleterious SNVs unique to non-responders and 45 unique to responders of which 84 and 35 were confirmed by Sequenom, respectively. The independent validation cohort consisted of 173 pts, mean age 60, 72% male, and 53% responders (complete or partial). Among 119 SNVs genotyped in this cohort, only 3 responder and 9 non-responder SNVs were identified in ≥ 3 pts. None of the responder SNVs were associated with response. One non-responder SNV (rs150721903) was significantly associated with resistance to GC (OR: 9.7, 90% CI: 1.7-56.8, p=0.011), and remained significant after adjusting for risk score (p= 0.017). rs150721903 is a missense variant in an uncharacterized protein on chromosome 8. Conclusions: Exome sequencing of extreme phenotypes identified a variant that predicts resistance to GC, demonstrating the feasibility of this pharmacogenomics approach. Further evaluation of this SNV is warranted.

279 Background: PPARy is a steroid hormone receptor encoded by the PPARG gene (Chr 3p25). Data are conflicting regarding its role in UC. Some studies suggest activation has antiproliferative effects, and preclinical data indicate a potential role for PPARy agonists in cancer treatment. However, recent findings show that use of the PPARy agonist thiazolidinones (TZD) in type 2 diabetes is associated with increased risk of developing UC. To define the spectrum of PPARy aberrations in UC, DNA was evaluated to detect copy number gains/losses (CNG/CNL) of PPARG in three clinically annotated UC cohorts. Methods: Cohort A contained 97 patients with high-grade UC. Cohort B contained 94 patients with primary bladder tumors who developed metastatic UC. Cohort C had 33 patients with metastatic UC, including both primary and paired metastatic tissue for 14 patients. DNA from all tumors was subjected to copy number analysis (array comparative genomic hybridization for cohorts A and B and molecular inversion probe array for cohort C) to identify focal areas of CNG/CNL. Statistical analysis using RAE (cohort A) or GISTIC 2.0 (with CNG defined as log2 copy number ratio >0.8) in cohorts B/C was used to identify significant alterations. Results: The region containing PPARG showed CNG in 7%, 10%, and 15% of cohorts A, B and C, respectively. In cohort A, RAE p value was 5.36E-05 at the amplicon containing PPARG. In cohorts B and C GISTIC 2.0 q values were 9.75E-12 and 6.04E-10. PPARG CNG had no impact on survival in the cohorts with available survival data (A and B). Paired biopsies from primary and metastatic sites showed discordance in PPARG CNG in 2 cases. In 1 case CNG was present in the metastasis but not in the primary and in 1 case the opposite occurred. Conclusions: PPARG shows significant CNGs in 7-15% of UC. Notably the RAF1 locus (a known oncogene) lies close to PPARG and could also represent the driver alteration within this amplicon. Investigation of the functional relevance of these CNGs is clearly warranted as it may be a driver of bladder cancer growth and progression. Determination of PPARG CNG status in UCs in a TZD treated population may provide mechanistic insights into the increased incidence of UC in patients treated with these agents.

290 Background: Despite level 1 evidence demonstrating survival benefit for cisplatin-based neoadjuvant chemotherapy in urothelial cancer, its utilization remains low, likely due to the modest benefit and potentially significant toxicities. To improve selection of patients (pts) most likely to benefit, we evaluated the association of pharmacogenomic markers of cisplatin sensitivity with neoadjuvant therapy outcome. Methods: Pts receiving standard neoadjuvant cisplatin-based chemotherapy for muscle-invasive disease were recruited to provide a germline DNA sample. Nine single nucleotide polymorphisms (SNPs) selected for their associations with platinum sensitivity in various clinical and pre-clinical studies were tested for association with complete pathologic response (pT0) rate at cystectomy. Results: Sixty-three Caucasian pts were analyzed (median age=64; 67% male; all bladder primary and cN0). The overall pT0 rate was 27%; 52% were 58% of cases. Demonstrating the apparent independent nature of the two SNPs, pts carrying either favorable genotype had pT0 OR=8.5 (95% CI 2.5-31.8, P=0.0008). The combined effect of detecting both SNPs was most informative, as 2/3 pts with both favorable genotypes achieved pT0 (67%), compared to 9/15 pts with one favorable genotype (60%), and only 6/43 lacking both favorable genotypes (14%). The negative predictive value considering both SNPs was 86%. Conclusions: Two germline SNPs having prior evidence of governing platinum sensitivity in pre-clinical and clinical models were demonstrated to confer strong associations with complete response to cisplatin-based neoadjuvant chemotherapy. Validation testing in an independent, multi-institutional cohort of urothelial cancer pts is underway. If reproducible, these SNPs deserve consideration as predictive clinical biomarkers to inform the use of neoadjuvant chemotherapy in this disease.

360 Background: We have previously shown that mutations in the epigenetic modifiers PBRM1, BAP1, SETD2 and KDM5C are associated with adverse tumor characteristics and, in some cases, worse cancer specific survival in clear cell renal cell carcinoma (ccRCC). We analyzed publically available data from the Cancer Genome Atlas Project (TCGA), to assess the impact of mutations in these genes on cancer-specific survival. Methods: We analayzed the genomic and clinical data from the TCGA cohort of 424 patients with primary ccRCC. The Kaplan-Meier method was used to estimate the survival probabilities, and log-rank test was used to test the univariate association between mutation status and overall survival. Cancer specific survival (CSS) was analyzed using the competing risk method. Multivariate Cox proportional hazard regression and competing risk models were also fitted to adjust for the validated Mayo Clinic SSIGN prognostic score. Results: Mutations in these epigenetic modifiers are frequent (PBRM1, 33.7%; SETD2, 11.6%; BAP1, 9.7%, KDM5C, 5.7%). BAP1 (p=0.002, HR 2.21 [1.34-3.62]), SETD2 (p=0.036, HR 1.68 [1.03-2.72]) and KDM5C (p=0.016, HR 2.18 [1.16-4.11]) are associated with worse CSS by competing risk. When adjusting for the prognostic SSIGN score, only mutations in KDM5C remain significant (p

e15022 Background: A marker of response to Bacillus Calmette-Guérin (BCG) intravesical therapy in high-risk non-muscle invasive bladder cancer (NMIBC) could help avoid unsuccessful BCG treatment and potentially enhance care. This study examines germline single nucleotide polymorphisms (SNPs) as markers of BCG-refractory NMIBC. Methods: Saliva or blood was prospectively collected from patients treated with BCG for NMIBC between 1991 and 2010, treated either at MSKCC or referring centers prior to MSKCC registration. BCG-refractory disease was defined as the presence of pathologically documented tumor in the bladder six months after BCG administration. SNPs were selected based on either reported associations with BCG response or putative mechanisms of BCG activity. Eighty candidate SNPs were genotyped (using the Sequenom Mass ARRAY iPLEX platform) in a discovery set (n=158) and those significantly associated were analyzed in a validation set (n=168). Univariate logistic regression was used to test the association between refractory disease, selected clinical variables, and SNPs. Because center of treatment was found to be a confounding variable (p=0.02), primary analysis was limited to cases treated at a single center (MSKCC), consisting of 276 patients (108 from discovery set and 168 from validation set). Results: Median age of diagnosis of the 276 cases was 65.7 years, 94% diagnosed with high-grade superficial bladder cancer, 35% stage Ta, 30% stage T1, and 35% stage Tis. At six month follow-up, 37 (13.4%) had BCG–refractory disease. Nine SNPs showed significant associations with outcome in the discovery cohort (two SNPs failed genotyping). In primary analysis, One SNP, rs11615, an intronic variant within the ERCC1 gene, was associated with BCG-refractory disease. Patients with CC genotype were 3.6 times more likely to be refractory to BCG compared to TT genotype (OR for each C allele was 1.8: 95% CI 1.1-3; p=0.01). No other factors were found to be predictive. Conclusions: A single nucleotide polymorphism of ERCC1 may be associated with BCG-refractory disease. Validation in independent datasets is needed.