Association between sites of metastases (mets) and outcomes with immune checkpoint inhibitor (ICI) therapy for advanced urothelial carcinoma (aUC).

445 Background: Different metastatic sites have variable prognostic implications in aUC. However, details on response and outcomes with ICI for particular mets is still unknown. We hypothesized that bone and liver mets would have poor response and outcomes with ICIs. Methods: We performed a retrospective cohort study in patients (pts) with aUC who received ICI. We compared overall response rate (ORR) and overall survival (OS) between pts with different mets at ICI initiation. We developed 4 different models: 1) lymph node (LN) only vs other; 2) visceral mets (bone, lung, liver) vs other; 3) bone + liver mets vs bone without liver vs liver without bone vs neither and 4) 6 factor model: a. LN +/- soft tissue/locoregional recurrence b. lung +/- (a) c. bone +/- (b) d. liver +/- (c) e. central nervous system (CNS) +/- (d) and f. other. ORR and OS were compared among groups using multivariable (adjusting for ECOG PS and hemoglobin<10g/dl) logistic regression and cox regression, respectively. Results: We identified 984 pts (24 institutions); 703 and 696 were included in OS and ORR analyses, respectively. Median age at ICI start was 71 (range 32-93), 77% white race, 74% men, 67% ever smokers, 72% pure UC, 18% upper tract UC, 55% extirpative surgery. Prevalence of LN, lung, bone and liver mets at ICI start was 74%, 32%, 27% and 21%, respectively. LN-only mets had significantly higher ORR (44% vs 22%, OR 2.6, p<0.05) and longer mOS (22 vs 8 months, HR 0.5, p<0.05) vs other mets. Visceral mets had significantly lower ORR (21% vs 35%, OR 0.5, p<0.05) and shorter mOS (7 vs 17 months, HR 1.8, p<0.05) vs non-visceral mets. Pts with bone and liver mets had significantly lower ORR and shorter OS vs those with bone or liver mets, which both had significantly lower ORR and shorter OS vs those with neither and with LN +/- local recurrence (Table). Conclusions: In the context of ICI treatment, bone, liver, lung or CNS mets were associated with lower ORR and/or shorter OS, and bone and liver mets were particularly associated with low ORR and short OS. LN-only mets were associated with higher ORR and longer OS. Further work is needed to interrogate site-specific tumor-host immune interactions and identify biomarkers. Research Sponsor: None[Table: see text]

Impact of subsequent therapy on survival in KEYNOTE-361: Pembrolizumab (pembro) plus chemotherapy (chemo) or pembro alone versus chemo as first-line therapy for advanced urothelial carcinoma (UC).

439 Background: The phase III KEYNOTE-361 study examined the efficacy and safety of 1L pembro + chemo or pembro alone vs chemo for pts with advanced UC. The PFS and OS benefit of pembro + chemo vs chemo did not reach statistical significance; no further formal tesing was done. We present an exploratory analysis of OS by subsequent therapy in KEYNOTE-361 (NCT02853305) to assess how 1L and 2L therapy selection affected survival outcomes; no formal comparisons were conducted. Methods: OS was estimated for pts by whether they received subsequent therapy, and by whether subsequent therapy included an anti–PD-(L)1 agent. Results: 351 pts were randomized to pembro + chemo, 307 pts to pembro, and 352 pts to chemo. As of Apr 29, 2020, the median (range) time from randomization to data cutoff was 31.7 (22.0-42.3) mo. 124/351 pts (35%) in the pembro + chemo arm, 126/307 pts (41%) in the pembro arm, and 215/352 pts (61%) in the chemo arm received any subsequent therapy. Similar rates of subsequent therapy (pembro + chemo: 32%; pembro: 43%; chemo: 59%) were observed for pts who experienced progressive disease (PD) by blinded independent central review (BICR). A higher rate of pts (169/352 [48%]) in the chemo arm received subsequent anti–PD-(L)1 therapy than in either the pembro + chemo arm (23/351 [7%]) or pembro arm (14/307 [5%]). Due to the small pt numbers, pts in the pembro + chemo or pembro arms who received subsequent anti−PD-(L)1 were not considered further. This analysis included all pts who received 2L therapy (465/1010 pts [46%]); the rate of 2L therapy was similar in pts with PD by BICR (274/615 [45%]). Chemo agents alone or in combination, specifically carboplatin, cisplatin, docetaxel, doxorubicin, gemcitabine, and paclitaxel, were the most commonly received subsequent therapies for pts who did not receive anti–PD-(L)1 in 2L. Pts who received 1L chemo followed by subsequent anti–PD-(L)1 had longer mOS (19.1 mo [95% CI 16.2-22.2]) than pts with 1L pembro followed by 2L therapy not including an anti−PD-(L)1 agent (16.0 mo [95% CI 11.8-19.2]) (Table). Conclusions: In this exploratory analysis, favorable survival outcomes were observed for pts who received 1L chemo followed by anti–PD-(L)1 compared with pts who received 1L pembro followed by 2L therapy not including an anti–PD-(L)1 agent. These data underline the continued importance of immunotherapy as 2L therapy for advanced UC. Clinical trial information: NCT02853305 . Research Sponsor: Merck & Co., Inc[Table: see text]

A Review of Probabilistic Opinion Pooling Algorithms with Application to Insider Threat Detection

We retrospectively explore the effectiveness of various probabilistic opinion pools against a set of insider threat detection modeling data from a recently completed, multiyear, sponsored research effort. We explored four opinion pools: the linear opinion pool (likely the most popular), the beta-transformed linear opinion pool, the geometric opinion pool, and a multiplicative method based on odds called Bordley’s formula. The data for our study came from our recent work in the inference of insider threats for our research sponsor. In this work, we created a multimodeling inference enterprise modeling (MIEM) process to either predict threats within a population or, given the threats, predict how well the enterprise system can detect those threats. As part of larger research challenges designed by the research sponsor, we applied the MIEM process quarterly to respond to a sequence of varying challenge problems (CPs). Via MIEM, we developed multiple, independent computation forecast models. These models generated certainty intervals to answer CP questions. These intervals were fused into a single interval for each question via an expert panel prior to submission. The sponsors scored the responses against ground truth. In this paper, we (a) ask which pooling functions work best on these data and consider why, and (b) compare this performance to the actual submissions to determine if one of the pooling functions performed better than our judgment-based fusion.

More than a Box to Check: Research Sponsor and Clinical Investigator Perspectives on Making GCP Training Relevant

Background: Good clinical practice (GCP) training is the industry standard for ensuring the quality conduct of registrational clinical trials. However, concerns have been raised about whether the current structure and delivery of GCP training sufficiently prepares clinical investigators and their delegates to conduct clinical trials. Methods: We conducted qualitative semi-structured interviews with 13 clinical investigators and 10 research sponsors to 1) examine characteristics of the quality conduct of sponsored clinical trials, including critical tasks and concerns perceived as essential for trial quality, 2) identify key knowledge and skills required to perform critical tasks, and 3) identify gaps and redundancies in GCP training and areas of improvement to ensure the quality conduct of clinical trials. We used applied thematic analysis to analyze the data. Results: The top three tasks identified as critical for the quality conduct of clinical trials were obtaining informed consent, ensuring protocol compliance, and protecting participants&rsquo; health and safety. Respondents acknowledged that GCP principles address each of these critical tasks; however, they described many challenges and burdens of GCP training, including high training frequency and repetitive content. Respondents suggested moving beyond GCP training as a mere check-box activity by making it more effective, engaging, and interactive. They also emphasized that applying GCP principles in a real-world, skills-based environment would increase the relevance of GCP training to investigators and their delegates. Conclusion: Our findings indicate that although investigators and sponsors recognize that GCP training addresses critical tasks necessary to the quality conduct of clinical trials, they articulated the need for significant improvement in the design, content, and presentation of GCP training.