Erdafitinib versus pembrolizumab in pretreated patients with advanced or metastatic urothelial cancer with select FGFR alterations: cohort 2 of the randomized phase III THOR trial

BACKGROUND

Erdafitinib is an oral pan-fibroblast growth factor receptor (FGFR) tyrosine kinase inhibitor approved to treat locally advanced/metastatic urothelial carcinoma (mUC) in patients with susceptible FGFR3/2 alterations (FGFRalt) who progressed after platinum-containing chemotherapy. FGFR-altered tumours are enriched in luminal 1 subtype and may have limited clinical benefit from anti-programmed death-(ligand) 1 [PD-(L)1] treatment. This cohort in the randomized, open-label phase III THOR study assessed erdafitinib versus pembrolizumab in anti-PD-(L)1-naive patients with mUC.

PATIENTS AND METHODS

Patients ≥18 years with unresectable advanced/mUC, with select FGFRalt, disease progression on one prior treatment, and who were anti-PD-(L)1-naive were randomized 1 : 1 to receive erdafitinib 8 mg once daily with pharmacodynamically guided uptitration to 9 mg or pembrolizumab 200 mg every 3 weeks. The primary endpoint was overall survival (OS). Secondary endpoints included progression-free survival (PFS), objective response rate (ORR), and safety.

RESULTS

The intent-to-treat population (median follow-up 33 months) comprised 175 and 176 patients in the erdafitinib and pembrolizumab arms, respectively. There was no statistically significant difference in OS between erdafitinib and pembrolizumab [median 10.9 versus 11.1 months, respectively; hazard ratio (HR) 1.18; 95% confidence interval (CI) 0.92-1.51; P = 0.18]. Median PFS for erdafitinib and pembrolizumab was 4.4 and 2.7 months, respectively (HR 0.88; 95% CI 0.70-1.10). ORR was 40.0% and 21.6% (relative risk 1.85; 95% CI 1.32-2.59) and median duration of response was 4.3 and 14.4 months for erdafitinib and pembrolizumab, respectively. 64.7% and 50.9% of patients in the erdafitinib and pembrolizumab arms had ≥1 grade 3-4 adverse events (AEs); 5 (2.9%) and 12 (6.9%) patients, respectively, had AEs that led to death.

CONCLUSIONS

Erdafitinib and pembrolizumab had similar median OS in this anti-PD-(L)1-naive, FGFR-altered mUC population. Outcomes with pembrolizumab were better than assumed and aligned with previous reports in non- FGFR-altered populations. Safety results were consistent with the known profiles for erdafitinib and pembrolizumab in this patient population.

Olaparib (O) in patients (pts) with prostate cancer with BRCA1/2 inactivating mutations: Results from the Targeted Agent and Profiling Utilization Registry (TAPUR) study

5567

Background: TAPUR is a phase II basket study evaluating anti-tumor activity of commercially available targeted agents in pts with advanced cancers with genomic alterations. Advanced prostate cancer (PC) pts with germline or somatic BRCA1/2 inactivating mutations treated with O are reported. Methods: Eligible pts had advanced PC, no remaining standard treatment (tx) options, measurable disease, ECOG Performance Status (PS) 0-2, and adequate organ function. Tumor genomic testing was performed in CLIA-certified, CAP-accredited site selected labs. Pts received O tablets or capsules dosed at 300 mg (n=24) or 400 mg (n=5), respectively, orally twice daily until disease progression. Simon 2-stage design tested the null disease control (DC) (objective response (OR) or stable disease at 16+ weeks (wks) (SD16+) according to RECIST) rate of 15% vs. 35% (power = 0.85; α = 0.10). If ≥2 of 10 pts in stage 1 have DC, 18 more pts are enrolled. If ≥7 of 28 pts have DC, the tx is worthy of further study. Pts had radiographic evaluations at 8 and 16 wks and then every 12 wks. Secondary endpoints are progression-free survival (PFS), overall survival (OS) and safety. Results: 29 pts with BRCA1/2 inactivating mutations were enrolled from Aug 2016 to Jul 2019; 4 were identified as ineligible after enrollment due to bone only disease and removed from analyses. Demographics and investigator-reported outcomes are summarized in the Table. Nine pts with OR and 8 with SD16+ were observed for DC and OR rates of 68% (90% CI: 53% - 77%) and 36% (95% CI: 18% - 57%), respectively. Six pts had at least one grade 3 AE or SAE at least possibly related to O including anemia, aspiration, dehydration, diabetic ketoacidosis, fatigue, and neutropenia. Conclusions: Monotherapy with O showed anti-tumor activity in heavily pre-treated PC pts with germline (1/2 pts with OR or SD16+) or somatic (16/23 pts with OR or SD16+) BRCA1/2 inactivating mutations. These findings extend results from recent trials of O in advanced prostate cancer pts with germline only BRCA1/2 mutations. Clinical trial information: NCT02693535 . [Table: see text]

ERDAFITINIB in locally advanced or metastatic urothelial carcinoma (mUC): Long-term outcomes in BLC2001

5015

Background: Erdafitinib (JNJ-42756493; ERDA) is the only pan-FGFR kinase inhibitor with US FDA approval for treatment of adults with mUC with susceptible FGFR3/2 alterations (alt) and who progressed on ≥ 1 line of prior platinum-based chemotherapy (chemo). Approval was based on data from the primary analysis of the pivotal BLC2001 trial1. Here we report long-term efficacy and safety data from the 8 mg/d continuous dose regimen in BLC2001. Methods: BLC2001 (NCT02365597) is a global, open-label, phase 2 trial of pts with measurable mUC with prespecified FGFR alt, ECOG 0-2, and progression during/following ≥ 1 line of prior chemo or ≤ 12 mos of (neo)adjuvant chemo, or were cisplatin ineligible, chemo naïve. The optimal schedule of ERDA determined in the initial part of the study was 8 mg/d continuous ERDA in 28-d cycles with uptitration to 9 mg/d (ERD 8 mg UpT) if a protocol-defined target serum phosphate level was not reached and if no significant treatment-related adverse events (TRAEs) occurred. Primary end point was the confirmed objective response rate (ORR=% complete response + % partial response). Key secondary end points were progression-free survival (PFS), duration of response (DOR) and overall survival (OS). Results: Median follow-up for 101 patients treated with ERDA 8 mg UpT was ~24 months. Confirmed ORR was 40%. Median DOR was 5.98 mos; 31% of responders had DOR ≥ 1 yr. Median PFS was 5.52 mos, median OS was 11.3 mos. 12-mos and 24-mos survival rates were 49% and 31%, respectively. Median treatment duration was 5.4 mos. The ERDA safety profile was consistent with the primary analysis. No new TRAEs were seen with longer follow-up. Central serous retinopathy (CSR) events occurred in 27% (27/101) of patients; 85% (23/27) were Grade 1 or 2; dosage was reduced in 13 pts, interrupted for 8, and discontinued for 3. On the data cut-off date, 63% (17/27) had resolved; 60% (6/10) of ongoing CSR events were Grade 1. There were no treatment-related deaths. Conclusions: With a median follow-up of 2 yrs, ERDA in mUC + FGFR alt showed a manageable safety profile and consistent efficacy, with median OS of 11.3 mos. 31% had a DOR ≥12 mos and 31% were alive at 24 mos. ERDA monotherapy vs. immune checkpoint inhibitor (PD-1) or chemo is being further analyzed in a randomized control study (THOR; NCT03390504).Reference: Loriot Y, et al. N Engl J Med. 2019;381:338-48. Clinical trial information: NCT02365597 .

Comparison of germline mutations in African American and Caucasian men with metastatic prostate cancer

5568

Background: The relevance of germline mutations in metastatic prostate cancer is well established; however, comparison of germline genetics in African American (AA) versus Caucasian (CA) men with metastatic prostate cancer (PCa) is limited. Methods: Germline data from self-identified AA and CA metastatic PCa patients (pts) were collected from 5 academic cancer centers. Various commercial cancer-specific germline testing panels were used to evaluate 12-86 genes. Pathogenic (P) or likely pathogenic (LP) mutations, and variants of unknown significance (VUS), were reported according to ACMG guidelines. Self-reported family history (FH) was annotated for 99% of pts. Statistical analyses included Chi-squared and Fischer’s exact tests. Results: A total of 821 metastatic PCa pts were assessed: 152 AAs and 669 CAs. For P/LP alterations, AAs had a frequency of 11.2% (17/152) as compared to a frequency of 14.6% (98/669) in CAs (p = 0.302). AA pts were more likely to have a VUS than CA pts, 61% vs 43% respectively (OR = 2.09, 95%CI [1.45, 2.99], p < 0.001). BRCA mutations were similar between races, but AA were more likely to have a BRCA1 P/LP alteration (OR = 6.00, 95% CI [1.33, 27.09], p = 0.025). AA pts were less likely to have a P/LP alteration in a non-BRCA gene (OR = 0.34, 95% CI [0.15, 0.80], p = 0.013). Among DNA repair genes, there were no significant difference between AA and CA pts (p = 0.574); however, there was a trend toward AA pts having fewer P/LP alteration in a non-BRCA DNA repair genes (OR = 0.26, 95% CI [0.06, 1.08], p = 0.071). In pts with >1 first degree relative (FDR) with ovarian cancer, P/LP germline alterations were more likely in CAs (OR = 2.33, 95% CI [1.05, 5.17], p = 0.043); but there were no significant differences in AAs (p = 0.098). Those with >2 FDRs with PCa were more likely to have a P/LP change in CAs (OR = 2.32, 95% CI [1.04, 5.15], p = 0.043), but there were no difference in AAs (p = 0.700). In pts with ≥2 FDRs with breast cancer, P/LP germline alterations were more likely in both AAs (OR = 9.36, 95% CI [1.72, 50.84], p = 0.019) and CAs (OR = 3.92, 95% CI [1.79, 8.59], p = 0.001). Conclusions: We did not observe a difference in the overall frequency of germline P/LP alterations between AA and CA men with metastatic PCa but VUSs were more common in AA men. These AA men have an overall frequency of BRCA mutations similar to CA men; however, BRCA1 mutations were more prevalent in these AAs. Non-BRCA P/LP mutations are significantly less frequent in AA pts. A positive family history of >2 FDRs with breast cancer was associated with P/LP alterations in both AA and CA pts.

Study EV-103: New randomized cohort testing enfortumab vedotin as monotherapy or in combination with pembrolizumab in locally advanced or metastatic urothelial cancer

TPS5092

Background: Cisplatin-based chemotherapy is the standard for first-line (1L) patients (pts) with locally advanced/metastatic urothelial cancer (LA/mUC). PD-1/PD-L1 inhibitors have promising durability of responses but 1L use is restricted to pts ineligible for cisplatin-containing therapy and whose tumors express PD-L1 (CPS ≥10) or pts ineligible for platinum-containing chemotherapy regardless of PD-L1 status. Enfortumab vedotin (EV), an antibody-drug conjugate, delivers the microtubule-disrupting agent monomethyl auristatin E to cells expressing Nectin-4, which is highly expressed in UC. EV recently received FDA accelerated approval based on tumor response rates for adults with LA/mUC who have previously received a PD-1/PD-L1 inhibitor and a platinum-containing chemotherapy. In the ongoing phase 1b/2 study EV-103/KEYNOTE-869 (NCT03288545), the safety and antitumor activity of EV are investigated as monotherapy (mono) (for the first time in the 1L setting) and in combination with PD-1 inhibitor pembrolizumab (P) +/- chemotherapy in UC. An initial analysis of EV (1.25 mg/kg) + P (200 mg) (both drugs in investigational use here) in this study showed a 73.3% confirmed ORR in 45 1L cisplatin-ineligible LA/mUC pts (dose-escalation + expansion Cohort A) (Rosenberg ASCO 2020). Methods: A new Cohort K randomized 1:1 to 1.25 mg/kg EV mono or 1.25 mg/kg EV + 200 mg P provides additional information on EV + P and the contribution of activity from EV in cisplatin-ineligible pts with LA/mUC in the 1L setting. This cohort will enroll 150 adults (≥18 years) with LA/mUC and measurable disease per RECIST v1.1, and exclude pts with prior systemic treatment for LA/mUC, active CNS metastases, ongoing sensory or motor neuropathy (Grade ≥2), or uncontrolled diabetes. Cisplatin-ineligibility in this study is based on ≥1 of the following: ECOG of 2, creatinine clearance of ≥30 and < 60 mL/min, or hearing loss/dysfunction. In each 3-week cycle of this study, EV is administered on days 1 and 8, and P on day 1. The primary endpoint is ORR per RECIST v1.1 by BICR. Secondary endpoints include ORR per RECIST v1.1 by investigator assessment, DOR, DCR, PFS per RECIST v1.1 by BICR and investigator assessment, OS, safety, and tolerability. Sample size is not based on power calculation for formal hypothesis testing but is selected based on ORR estimate precision based on 95% CIs. Efficacy is summarized by treatment arm with no formal statistical comparisons between arms. The study opened in Oct 2017. Cohort K opened in Jan 2020. Clinical trial information: NCT03288545 .

Pathogenic variants in PTEN to predict for increased risk of relapse and death in patients with limited stage small cell bladder cancer

526

Background: Small cell bladder cancer (SCBC) is a rare histologic subtype with insufficient genomic characterization. Patients with limited stage (LS) SCBC have a poor prognosis, and no biomarker exists to optimize treatment selection. We sought to identify genomic aberrations in patients with LS-SCBC using a comprehensive next generation sequencing (NGS) platform. Mutations in the PTEN/AKT pathway are important in urothelial tumor biology but have an undefined role in SCBC. Methods: 23 LS-SCBC cases were identified. NGS was performed on diagnostic transurethral bladder tumor resection or cystectomy specimens containing SCBC. Detected variants were filtered by in silico algorithms predicting for a deleterious impact on protein function. Variant allele frequencies (VAF) greater than 2% were permitted in this analysis. Variants in the PTEN gene were assessed for association with relapse-free survival (RFS) and overall survival (OS) using Kaplan-Meier techniques and Cox proportional hazards models. Results: Median follow up for the cohort was 4.02 years. 14/23 (60.9%) patients have died. Six unique deleterious PTEN mutations were observed in 9/23 (39.1%) patients. p.W274C was the most common PTEN variant and was detected in 5 (21.7%) patients. Three variants were detected at > 10% VAF. All 9 patients with a deleterious PTEN variant died. The presence of deleterious PTEN variants [HR = 4.68 [(1.54, 14.27), p = 0.003]] predicted for inferior OS. In the 19 patients with known relapse history, 6/7 (85.7%) with and 3/12 (25%) without any deleterious PTEN mutation relapsed. The presence of deleterious PTEN variants [HR = 9.41 [(2.32, 38.23), p < 0.001]] also predicted for inferior RFS. Conclusions: Pathogenic variants in tumor suppressor PTEN were associated with inferior RFS and OS in this pilot cohort of patients with LS-SCBC, suggesting that disruption of PTEN function may be a critical genomic event underlying the progression of small cell bladder cancer. Our findings also support prior reports that pathogenic gene variants detected at low allele frequencies may be clinically important.

Contemporary racial disparities in PSA screening in a large, integrated health care system

308

Background: The USPSTF prostate cancer screening guidelines have changed significantly in the past decade, from a recommendation of do not screen in 2012 to a 2018 recommendation that focuses on shared decision making. Additionally, most guidelines further acknowledge that African American men should be screened more intensively than Caucasian men due to increased incidence of prostate cancer and increased prostate cancer mortality. Our objective was to characterize racial disparities in PSA screening in a large healthcare system with a diverse patient population to understand contemporary trends. Methods: This retrospective cohort study used data from the Atrium Health Enterprise Data Warehouse, which includes clinical records from over 900 care locations across North Carolina, South Carolina, and Georgia. Participants included all men ≥ 40 years seen in the ambulatory or outpatient setting during 2014-2018. PSA testing was determined through laboratory data. Clinical and demographic data were collected. Between-group comparisons were conducted using generalized estimating equations models to account for within-subject correlation. Statistical significance was defined as p < 0.05. Results: There were 582,846 individual men seen from 2014-2018, including 416,843 Caucasians (71.5%) and 85,773 African Americans (14.7%). Screening rates declined among all groups from 2014-2018 (see table). African American men were screened at a similar or lower rate than Caucasian men in each year (from 18.6% vs 19.0% in 2014 to 11.9% vs 12.2% in 2018, respectively). Conclusions: PSA screening declined significantly between 2014 and 2018. African American men screened at a similar or lower rate than Caucasian men each year. Given the consensus that African American men should be more intensively screened for prostate cancer, significant racial disparities remain in prostate cancer screening. Further study is warranted to understand patient, provider, and system factors that contribute to disparities in prostate cancer care and outcomes.[Table: see text]

Discordance of HER2 expression in distant metastatic tumors (MTs) versus primary tumors (PTs) in urothelial carcinoma (UC)

550

Background: Available HER2 targeted agents have limited benefit for UC, but exceptional tumor responses are occasionally seen. HER2 expression has been tested as a biomarker of response to these drugs. Trial eligibility is often based on HER2 expression in PTs. Previous studies found moderate to high concordance of HER2 expression in PTs and synchronous regional nodes at cystectomy, but the concordance rate with distant MTs is poorly defined and further complicated by shifting definitions of HER2 positivity. Methods: Immunohistochemical (IHC) staining for HER2 (Ventana Pathway clone 4B5) was performed on PTs and matched MTs in 79 patients with UC. IHC staining was scored (0-3+) using intensity and % positive tumor cells using 2018 ASCO/CAP guidelines for breast cancer. Fisher’s Exact Test assessed association of PT and MT HER2 scores, and Cohen’s kappa statistic (κ) assessed agreement in HER2 3+ between groups. MTs were defined as synchronous if biopsied <60 days after the PT biopsy. Results: There were 67 metachronous and 12 synchronous paired biopsies; median time between metachronous biopsies was 418 days. MT biopsy sites included lymph node (30%), bone (22%), lung (18%), pelvic/peritoneal soft tissue (11%), liver (8%), brain (4%), other (6%). HER2 2+ and 3+ expression was seen in 30.4% and 12.7% of PTs and 12.7% and 7.6% of MTs, respectively. Of HER2 3+ PTs (n=10), HER2 expression in paired MTs was 0 (n=3), 1+ (n = 3), 2+ (n = 2), and 3+ (n = 2). Of HER2 3+ MTs (n=6), HER2 expression in paired PTs was 0 (n=1), 2+ (n=3), and 3+ (n=2). An association was observed between PT and MT HER2 scores (p=0.004), however there was low agreement of HER2 3+ rates between PT and MT lesions (κ = 0.17, p=0.113). Discordance was noted in both synchronous and metachronous biopsies. Conclusions: Based on the current HER2 3+ definition, HER2 overexpression is spatially and temporally discordant in UC between PTs and distant MTs. These findings suggest that both PTs and MTs should be tested for HER2 expression in future clinical trials assessing the efficacy of HER2 targeted agents in UC. The high discordance may explain previously negative therapeutic studies.

Comparison of Caucasian and African-American DNA repair alterations in men with metastatic prostate cancer

199

Background: Data on germline DNA repair defects and VUS rates are sparse in African American (AA) men with metastatic prostate cancer (PCa). Methods: Germline testing data from two centers with a significant percentage of metastatic AA PCa patients were combined and compared to Caucasian American (CA) with metastatic PCa. Fourteen canonical DNA repair genes (ATM, BARD1, BRCA1, BRCA2, BRIP1, CHEK2, MLH1, MSH2, MSH6, NBN, PALB2, PMS2, RAD51C, RAD51D) were assessed in all tested patients (pts) using a pathogenic/likely pathogenic (P/LP) classification. Variants of unknown significance (VUS) were assessed in an Invitae-derived dataset with consistent VUS reporting. Results: A total of 105 AA men with metastatic disease were evaluated and 7/105 of these men (6.67%) had P/LP alteration. Among the AA pt alterations, there were 4 pts with BRCA2, 2 pts with BRCA1, and 1 pt with PALB2. A total of 39/417 (9.3%) of CA metastatic patients had P/LP alterations in the canonical 14 genes. No differences were detected in the AA vs CA metastatic comparison (p=0.39). A total of 1/105 (0.95%) AA pts and 23/418 (5.5%) CA had non-BRCA P/LP mutations. The number of non-BRCA P/LP mutations were lower in the AA as compared to the CA men (p=0.045). When evaluating VUS calls in the metastatic AAs using Invitae multi-gene panels, 28/92 (30.43%) pts had a VUS in the canonical 14 genes as compared to 67/366 (18.31%) of the CA men. AAs were more likely than CA to have a VUS (p=0.010). These data indicate that metastatic AA pts and CA are not significantly distinct in the P/LP alterations in 14 canonical DNA repair genes but that there were lower percentages of P/LP in the AA non-BRCA gene subset. Further, when assessing these genes, it is clear that a VUS is more likely to be called in the AA men. Conclusions: Among men with metastatic PCa, AAs have similar rates of inherited P/LP alterations in 14 well accepted DNA repair genes as compared to CA men, however the non-BRCA gene P/LP alterations were less frequent among the AAs. Variants classified as a VUS were clearly higher in these AA pts as compared to the CA pts.

Molecular profiling of aggressive variant urothelial carcinoma

378

Background: The WHO recognizes multiple variant histologies of urothelial carcinoma (vUC), many of which have been associated with poor outcomes compared with urothelial carcinoma (UC). We aimed to explore molecular differences between aggressive vUC and UC. Methods: 23 micropapillary (MP), 16 plasmacytoid (P), 23 sarcomatoid (S), 7 nested (N), 6 clear cell (CC), and 2 giant cell (GC) vUC specimens were tested between 2012 to 2018 via a multiplatform profiling service (Caris Life Sciences, Phoenix, AZ) consisting of gene sequencing (next generation sequencing [NGS]), gene amplification (CISH or NGS), and protein expression (immunohistochemistry [IHC]). Findings were compared to 435 control UC specimens using the Chi-square test. Results: 84% of samples were from primary tumor. Alterations identified are summarized in Table 1, and are notable for high rates of TP53 mutations across histologic subtypes, varying rates of RB1, ERBB2 and FGFR mutations, and overall low rates of DNA damage repair (DDR) mutations (29 genes reported) except in S. There were more ARID1A mutations detected in MP than UC (100% [3 specimens] v. 41.3%, p=0.044), and more CDH1 mutations in P than UC (50% [4 specimens] v. 2%, p<0.001). CISH ERBB2 (HER2) amplification was seen in 27.3% MP compared with only 10.4% in UC (p=0.005). Compared to UC, PD-L1 IHC (SP142 assay) was positive (>5%) in a high proportion of S (55.6%, p=0.002) but in a lower proportion of other vUC (e.g. absent in P). Tumor mutational burden (TMB) was high in a lower proportion of vUC: 18.4% UC vs. 14.3% MP, 0% P, 16.7% S. Conclusions: Aggressive variant histology UCs have a differential profile of molecular aberrations compared to UC, with notable differences in potential targets such as HER2 and DDR genes as well as immunotherapy biomarkers. Further studies are needed to confirm these findings, and may support therapy development for these rare, aggressive UC subtypes. Aberrations (%) in Variant Histology UC. [Table: see text]

Impact of Aurora kinase A and B expression on response to neoadjuvant chemotherapy and patient outcome in muscle-invasive bladder cancer (MIBC)

393

Background: Identification of robust biomarkers that predict likelihood of benefit from neoadjuvant chemotherapy (NAC) in patients with MIBC remains an unmet need. Previous studies have implicated tumor overexpression of aurora kinases A (AURKA) and B (AURKB) as a mechanism of chemo-resistance. Because overexpression of AURKA has also emerged as a potential biomarker for detection of high-risk urothelial carcinoma, we sought to characterize the expression of AURKA, AURKB with clinical outcomes in MIBC patients who received NAC and to test the hypothesis that tumor overexpression of AURKA and AURKB would predict for residual MIBC. Methods: 47 patients with MIBC who received NAC prior to cystectomy were retrospectively identified. Immunohistochemistry for AURKA and AURKB was performed on pre-treatment diagnostic transurethral resection of bladder tumors and matched cystectomy specimens. Logistic regression models were estimated to determine the impact of pre-NAC expression on pathologic staging at cystectomy. Receiver Operator Characteristic curves (ROC) were calculated to assess diagnostic predictive ability of AURKA and AURKB. AURKA and AURKB were assessed for association with relapse-free survival (RFS) and overall survival (OS) using Kaplan-Meier techniques and Cox proportional hazards models. Results: 22 of 47 [46.8%] patients had residual muscle-invasive (ypT2-4) urothelial carcinoma after NAC. Neither baseline tumor expression of AURKA (ROC = 0.54, p = 0.71) nor AURKB (ROC = 0.46, p = 0.98) predicted for ypT2-4 status. However, baseline expression of AURKA above the 75th percentile for this cohort, determined by the percentage of tumor cells positive, was associated with an inferior RFS, [HR = 3.79, [1.40, 10.26] p = 0.005] and OS, [HR = 5.84, [2.14, 15.98], p < 0.001]. Similar trends for worse survival outcomes were also observed for high AURKB levels. Conclusions: Although baseline tumor AURKA and AURKB expression did not predict for pathologic residual MIBC after NAC, high expression of AURKA and AURKB predicted for inferior RFS and OS. Further evaluation of AURKA and AURKB as potential biomarkers and therapeutic targets in MIBC is warranted.

Non-urothelial bladder cancer: Genomic alterations and patient outcomes

399

Background: Adenocarcinoma (ADA) and squamous cell carcinoma (SCC) are rare, aggressive subtypes of bladder cancer, for which no clear standard of care exists. We report on survival of pts with ADA and SCC and identify potential therapeutic targets using molecular profiling via next generation sequencing (NGS). Methods: Survival trends, demographics, pt characteristics were obtained from the Surveillance, Epidemiology, and End Results (SEER) Database. In a separate cohort, NGS results from 72 specimens (50% metastatic) were also analyzed, using either a hotspot 47 gene panel or a 592 gene assay (Caris Life Sciences, Phoenix, AZ). Results: In SEER, 235,537 cases of bladder cancer were extracted from 1988-2008, of which 3096 were SCC and 671 were ADA. 90% of pts were white, although more African-American patients (15%) were seen in those with ADA. Among all stages, median overall survival (mOS) and 5-yr survival rates were 17.9 mos and 58% for ADA and 15 mos and 37% for SCC. Via NGS testing, 43 patients (28 ADA, 15 SCC) were tested with a 47 gene panel and 29 (21 ADA, 8 SCC) with a 592 gene panel. In the 47 gene panel, among ADA pts, the highest mutation rates were TP53 (57.1%), KRAS (21.4%), SMAD4 (14.8%), PIK3CA (10.7%) and BRCA2 (7.7%). Among SCC pts, the highest mutation rates were TP53 (66.7%), PIK3CA (33.3%), HRAS (14.3%), FBXW7 (6.7%) and AKT1 (6.7%). In the 592 gene assay, the genes with the highest mutation rates in pts with ADA were TP53 (81%), SMAD4 (33.3%), KRAS (23.8%), KMT2C (11.8%), ARID1A (11.1%), BRAF (9.5%), CTNNB1 (9.5%), KMT2D (9.5%), TSC1 (9.5%), KDM6A (5.9%), CDKN2A (5%). Among SCC pts, the highest mutation rates were TP53 (75%), CDKN2A (42.9%), FGFR3 (25%), PIK3CA (25%), CIC (14.3%), KDM6A (14.3%), BRAF (12.5%), BRCA1 (12.5%), FH (12.5%), HRAS (12.5%) and KMT2D (12.5%). Only 1 pt had high TMB. Conclusions: Genomic profiling identifies differences in underlying tumor biology of bladder ADA and SCC, which on a population level are rare with poor survival. Overall, the alterations in the PIK3CA/ AKT/ mTOR and TP53 pathways are similar to what has been reported in UC. Future analyses of these malignancies should investigate the emerging actionable targets, such as TSC1, FGFR3, BRCA1/2 and BRAF.

Discordance rate of PD-L1 expression between primary and metastatic lesions in urothelial carcinoma (UC)

493

Background: Immune checkpoint inhibitors (ICI) targeting PD-1 or PD-L1 are effective in select patients with advanced UC. High PD-L1 expression enriches for response to ICIs; however, the predictive value of PD-L1 expression is limited, which may be due in part to dynamic expression of PD-L1 in the tumor environment. We sought to characterize PD-L1 expression in primary UC and paired metastatic lesions to gain insight into the potential temporal discordance of tumor PD-L1 expression during the metastatic process. Methods: Immunohistochemical (IHC) staining for PD-L1 using the SP-142 antibody was performed on primary tumors and matched metastatic specimens in 83 patients with advanced UC. IHC staining was scored for the percentage of cells positive ( < 5%, ≥5%) in tumor cell (TC) and immune cell (IC) compartments. Correlation of PD-L1 expression in TCs and ICs was estimated using Spearman’s correlation coefficients (ρ). Cohen’s kappa statistics (κ) were utilized to assess the agreement in PD-L1 expression between groups. Results: High (≥5%) PD-L1 expression in primary and metastatic biopsies, respectively, was observed in 6.1% and 14.6% of TCs and in 7.8% and 11.7% of ICs. High co-expression of PD-L1 in both TC and IC compartments was infrequent in primary and metastatic lesions (3.6% and 2.6%, respectively). PD-L1 expression in TCs was positively correlated with PD-L1 expression in ICs in primary tumors (ρ = 0.47) and in metastatic lesions (ρ = 0.27). TC PD-L1 expression in primary tumors was correlated with TC PD-L1 expression in paired metastatic lesions (ρ = 0.44) but there was minimal agreement in high expression rates between primary and metastatic lesions in the TC compartment (κ = 0.147). IC PD-L1 expression in primary tumors was not correlated with IC PD-L1 expression in paired metastatic lesions (ρ = 0.05) and there was poor agreement in high expression rates between primary and metastatic lesions in the IC compartment (κ = 0.086). Conclusions: High PD-L1 IC expression is temporally discordant between primary and metastatic UC lesions. Future studies of PD-1/PD-L1 targeted therapies in patients with metastatic UC should utilize recent biopsies of metastatic lesions to define PD-L1 expression when feasible.

Erdafitinib (ERDA; JNJ-42756493), a pan-fibroblast growth factor receptor (FGFR) inhibitor, in patients (pts) with metastatic or unresectable urothelial carcinoma (mUC) and FGFR alterations (FGFRa): Phase 2 continuous versus intermittent dosing

411

Background: Although immune checkpoint inhibitors (ICI) have improved outcomes in some pts with platinum-resistant mUC, many pts (eg, pts with TCGA luminal 1 tumors, many of whom are FGFRa) may not benefit. ERDA, a pan-FGFR (1-4) inhibitor, demonstrated promising phase 1 activity: 11 partial responses among 24 FGFRa mUC pts. We report efficacy and safety of ERDA in the ongoing global open-label phase 2 study BLC2001 (NCT02365597). Methods: Pts had measurable mUC with specific FGFR2/ FGFR3 mutations or translocations per central lab Janssen assay, ECOG 0-2, and were chemorefractory (progressed during/following ≥ 1 line of prior systemic chemo or ≤ 12 mos of [neo]adjuvant chemo). Cisplatin-ineligible, chemo-naïve pts, and prior ICI treatment were allowed. Pts were randomized 1:1 to 28-d cycles of oral 6 mg/d continuous dosing (6 C) or 10 mg/d intermittent 7 d on/7 d off dosing (10 I) ERDA; the dose was further uptitrated if no significant treatment-related adverse events (TRAEs) were observed. The primary end point was ORR. Results: 78 pts received 6 C and 33 pts received 10 I (10 I cohort stopped early) ERDA. 31 pts in 6 C arm were further uptitrated. Across arms, 50% had ≥ 2 prior lines of therapy; 93% were chemorefractory. Confirmed ORRs (RECIST 1.1) were 35% and 24%, and disease control rates (CR+PR+SD) were 74% and 73% in the 6 C and 10 I arms, respectively. Adverse events (AEs) were manageable, and there were no treatment-related deaths (Table). Treatment is ongoing in 10 pts. Conclusions: ERDA (6 C or 10 I) has promising efficacy and tolerability in pts with FGFRa mUC. Based on these results and ERDA pharmacometric modeling, dosing was optimized at 8 mg/d (continuous), and this cohort is ongoing. Phase 3 study is planned. Clinical trial information: NCT02365597. [Table: see text]

Molecular profiles of small cell bladder and prostate cancer and comparisons with small cell lung cancer

264

Background: Small cell bladder cancer (SCBC) and small cell prostate cancer (SCPC) are rare and aggressive cancers. Standard therapy remains a platinum agent combined with etoposide, with few options after recurrence. Advances in molecular genomics and drug development have altered our approach to cancer. These same novel approaches may alter how we approach SCBC and SCPC. The purpose of this study is to identify potential targets and compare molecular profiles of SCBC and SCPC to SCLC. Methods: In total, 21 SCBC and 19 SCPC were identified from a de-identified database (Caris Life Sciences). Specimens were evaluated for genetic aberrations (Sanger or next generation sequencing [NGS], ISH) and/or protein expression (immunohistochemistry [IHC]). Comparisons were made against a de-identified cohort of SCLC (n = 428). Results: Pathogenic/presumed pathogenic mutations in SCBC were found in TP53 (91.7%, 11/12), RB1 (18.2%, 2/11), PTEN (8.3%, 1/12), EGFR (7.7%, 1/13), and PIK3CA (7.1%, 1/14). SCPC genetic aberrations were detected in TP53 (72.7%, 8/11) and RB1 (25.0%, 2/8). No carcinomas in this cohort had a high mutational burden or MSI-high status (0%, 0/7). Amplified genes found in SCBC included DDR2 (50%, 1/2) and EGFR (25.0%, 1/4). In SCPC, gene amplification was found in AKT2 (20%, 1/5), CCNE1 (20%, 1/5), FGFR1 (20%, 1/5), and MYC (20%, 1/5). The highest protein expression rates in SCBC involved MRP1 (100%, 5/5), TOP2A (94.1%, 16/17), and RRM1 (81.3%, 13/16). The highest protein expression rates in SCPC were MRP1 (100%, 6/6), TUBB3 (100%, 9/9), and TOP2A (94.4%, 17/18). Comparisons between SCBC and SCPC with SCLC revealed more similarities than differences. Significant differences were found in RRM1 by IHC between SCBC and SCLC. Also, significant differences were found between SCPC and SCLC in AR and PTEN by IHC. Conclusions: Comparisons of GU small cell carcinomas reveal similarities to SCLC. Both TP53 and RB1 mutations were found in both SCBC and SCPC. Amplification in genes CCNE1 and FGFR1, frequently identified in SCLC, were also found in SCPC. The high protein expression in biomarkers like MRP1 may explain the poor response to cytotoxic chemotherapy. Prospective studies are urgently needed.

Efficacy of programmed death 1 (PD-1) and programmed death 1 ligand (PD-L1) inhibitors in patients with FGFR mutations and gene fusions: Results from a data analysis of an ongoing phase 2 study of erdafitinib (JNJ-42756493) in patients (pts) with advanced urothelial cancer (UC)

450

Background: FGFR3 mutations appear to be enriched in luminal 1 UC, which have low expression of markers associated with an immune response, including CD8-T-effector gene expression levels. Objective response rate (ORR) to PD-1 and PD-L1 inhibitors in luminal 1 tumors appear lower (10-19%) than in infiltrated luminal 2 and basal 3 tumors (ORR 31-34%). This suggests FGFR3 mutations occur within a group of tumors less likely to benefit from immune checkpoint inhibition (ICI). Here we describe outcomes of a subgroup of pts with FGFR2/3 mutations and gene fusions, previously treated with PD-1/PD-L1 inhibitors and included in an ongoing erdafitinib phase 2 study. Methods: Patients who had received prior immunotherapy for advanced UC were selected from those enrolled on BLC2001, an ongoing phase 2, open-label study of the pan-FGFR inhibitor erdafitinib in subjects with advanced UC with specific FGFR2/3 gene alterations (NCT02365597). We explored investigator-reported clinical outcomes including ORR and median time to progression (TTP) of pts treated with ICI preceding erdafitinib therapy. Results: 28/203 pts had previously received anti PD-1 or anti PD-L1 treatment. In this subgroup, the median age was 67.0 y; 82% were male, 79% had an ECOG score ≤1, 36% had liver metastases, and 93% had ≥2 prior lines of systemic therapy. The response rate to anti PD-1/PD-L1 agents was 3.6% (95% CI, 0.1%-18.3%) with one partial response. Median TTP was 3.4 mo (range 2-15 mo; 95% CI, 2.3-4.9). Conclusions: In this post-hoc analysis, FGFR mutations and gene fusions seem to select for a group of pts less likely to respond to immune checkpoint inhibition. Although caution is needed due to the retrospective nature of the data, the presence of these alterations may reflect an unmet need with currently approved agents. This finding highlights the need for FGFR alteration testing and an effective FGFR-targeted therapy. Clinical trial information: NCT02365597.