Citations

TPS5107 Background: Despite a growing number of treatment options for first line mCRPC, approximately 40% of patients (pts) have radiographic progression within the first year. Androgen receptor (AR) signaling inhibition increases genomic instability with double-strand DNA breaks & co-inhibition of AR & PARP induces synthetic lethality in multiple preclinical models. Homologous recombination repair (HRR) gene aberrations do not appear to be necessary for this synergy as demonstrated in a ph 3 clinical trial of abiraterone & olaparib where this combination improved radiographic progression-free survival (rPFS) in HRR-wild-type pts compared with abiraterone alone. A ph 1b trial has since shown that enzalutamide plus rucaparib has acceptable safety profile & no significant drug-drug interactions (S-DDI). Methods: CASPAR/A031902 (NCT04455750) is a ph 3 study in which 984 pts will be randomized 1:1 to enzalutamide plus rucaparib or placebo. HRR gene aberration is not required for enrollment. All pts will undergo next-generation targeted-exome sequencing from archival tumor tissue (new biopsy only required if no archival tissue available). Treatment will be continued until disease progression & crossover is not allowed. Key eligibility criteria are age ≥ 18 years, ECOG PS 0-2, biopsy-proven prostate adenocarcinoma, progressive (PSA or radiographic) disease per Prostate Cancer Working Group 3 guidelines, measurable or nonmeasurable disease per RECIST 1.1, no prior treatment for mCRPC (prior docetaxel, abiraterone, darolutamide, or apalutamide in non-mCRPC setting is allowed), no significant uncontrolled comorbidity, & no medications with S-DDI with enzalutamide/rucaparib. Hierarchical co-primary endpoints are rPFS & overall survival (OS). The OS analysis will be undertaken as a primary endpoint if the rPFS endpoint is met. For a one-sided logrank test with a type 1 error rate equal to 0.025, the study has 90% power to detect a hazard ratio (HR) of 0.71 in rPFS (median rPFS of 15 & 21 months in control & combination arms, respectively) & 80% power to detect an HR of 0.80 in OS (median OS of 32 & 40 months, respectively). Key secondary endpoints are rPFS & OS in pts with vs without pathogenic BRCA1, BRCA2, or PALB2 alterations; & differences in adverse events & quality of life (QOL) outcomes between the treatment arms. QOL assessments include Functional Assessment of Cancer Therapy–Prostate (FACT-P), Brief Pain Inventory Short Form (BPI-SF) & EQ-5D-5L. A key correlative endpoint is the sensitivity of ctDNA-based testing for alterations in HRR genes. Enrollment to CASPAR began in July 2021 & the study is available for participation to all US-NCTN sites with a projected enrollment of 3 years. Support: U10CA180821, U10CA180882, U24CA196171; U10CA180888. Clinical trial information: NCT04455750.

5083 Background: ARASENS and PEACE-1 trials suggests treatment intensification with novel hormonal therapy (NHT) in addition to androgen deprivation therapy (ADT) and docetaxel (DOC) provides survival benefit as compared to DOC+ADT in patients with metastatic castration sensitive prostate cancer (mCSPC). However, the performance of triplet therapy as compared to NHT+ADT remains unexplored. Methods: MEDLINE, EMBASE and recent conference proceedings were searched to include phase III trials evaluating triplet therapy in patients with mCSPC and reporting treatment effects in subgroup of patients with and without docetaxel. Outcomes included overall survival (OS) and radiographic progression free survival (rPFS). Precomputed hazard ratios (HRs) and confidence intervals (CIs) from non-randomized subgroup comparisons were pooled after logarithmic transformation using inverse-variance weighting approach. A DerSimonian-Laird random-effects meta-analysis was then performed to assess subgroup differences. Interaction between subgroups was assessed using P-value of heterogeneity. Mixed treatment comparisons were computed using a fixed-effect model within the frequentist framework using subgroup effect estimates from eligible trials. Results: This meta-analysis included five clinical trials (ARASENS, PEACE-1, ENZAMET, ARCHES and TITAN) with a total of 5804 patients (docetaxel: 2836; no docetaxel: 2836). Subgroup analysis showed statistically significant difference between treatment effects in patients who received docetaxel (NHT + DOC + ADT; HR: 0.74; 95% CI: 0.66-0.84; I2: 0%) and those who did not (NHT + ADT; HR: 0.61; 95% CI: 0.53-0.70; I2: 0%) for OS (p-value of interaction: 0.04). However, no statistically significant interaction was observed in terms of rPFS (p-value: 0.46). Mixed treatment comparisons showed improved OS with NHT + DOC + ADT (HR: 0.74; 95% CI: 0.66-0.84) as compared to DOC + ADT, but not when compared to NHT + ADT (HR: 0.97; 95% CI: 0.78-1.20). NHT + ADT significantly improved OS compared to DOC + ADT (HR: 0.77; 95% CI: 0.64-0.92). Consistently, significant rPFS improvement was observed with NHT + DOC + ADT when compared DOC + ADT (HR: 0.49; 95% CI: 0.42-0.57) but not when compared to NHT + ADT (HR: 0.82; 0.65-1.04). NHT + ADT was observed to improve rPFS compared to DOC + ADT (HR: 0.60; 95% CI: 0.50-0.72). Conclusions: This exploratory and hypothesis generating analysis suggests that addition of docetaxel (triplet therapy) may not delay progression or prolong survival compared to NHT-based doublets. These findings provide direction for future clinical trials in this space and suggest an equipoise to the question of how triplet regimens compare with NHT-doublets. The results should be interpreted with caution as this analysis does not account for potential confounding relationships such as volume of disease.

e16550 Background: Improved computational power and modern algorithms have generated significant interest in radiomics for cancer diagnosis and staging. Here we assess the performance of deep learning (DL) models as a means for feature extraction in combination with supervised machine learning (ML) algorithms for accurate staging and chemotherapy response assessment of bladder cancer. Methods: Deidentified grayscale CT images from bladder cancer patients scheduled to undergo radical cystectomy were included in this retrospective study. These images were manually annotated with two regional masks (normal region and cancer region). Five DL models- namely, AlexNet, GoogleNet, InceptionV3, ResNet-50, and XceptionNet pre-trained on the ImageNet dataset, a public dataset, were then fine-tuned on our bladder CT scan data to extract features. Through feature selection process, the subset of the features was used to build ML classifiers for classification. The classification was performed using five different ML classifiers, namely k-Nearest Neighbor (KNN), Naïve-Bayes (NB), Support Vector Machine (SVM), Linear Discriminant Analysis (LDA), and Decision Tree (DT). The classification task was performed with 10-fold cross-validation, and each of the experiments contained a different but not mutually exclusive subset of samples. The evaluation metrics include accuracy, sensitivity, specificity, precision, and F1-score. Results: A total of 200 deidentified grayscale CT images of 100 patients with histologically proven bladder cancer, were included in this study. For experiment (1) normal vs. cancer, the LDA classifier on XceptionNet based features provides the best performance with an accuracy of 86.07%, sensitivity of 96.75%, specificity of 69.65%, precision of 83.07%, and F1-score of 89.39%. For experiment (2) non-muscle invasive Cancer (NMIBC) vs. muscle invasive bladder cancer (MIBC), the LDA classifier on XceptionNet based features provided the best performance with an accuracy of 79.72%, sensitivity of 66.62%, specificity of 87.39%, precision of 75.58%, and F1-score of 70.81%. For experiment (3) T0 lesion vs. MIBC, the LDA classifier on XceptionNet based features provides the best performance with an accuracy of 74.96%, sensitivity of 80.51%, specificity of 70.22%, precision of 69.78%, and F1-score of 74.73%. Conclusions: Our proposed model has shown good results in differentiating normal vs cancer and promising performance in differentiating T0 vs MIBC after chemotherapy treatment. We are expanding our dataset to further improve the performance in differentiating T0 vs MIBC. In addition, we will investigate the applicability of GAN for data augmentation to address data limit. We believe the hybrid DL and ML framework may facilitates radiologists' decisions and clinical decision-making in patients with bladder cancer.

4594 Background: Cisplatin based neoadjuvant chemotherapy (NAC) followed by definitive therapy improves survival in patients with muscle invasive bladder cancer (MIBC). However, the clinical benefit of NAC might vary with the choice of definitive therapy. Therefore, we assessed the absolute benefit of NAC followed by radical cystectomy or radical radiotherapy separately using the totality of evidence. Methods: MEDLINE and EMBASE were systematically searched to identify randomized trials assessing cisplatin based neoadjuvant chemotherapy followed by either radical cystectomy or definitive radiotherapy in patients with MIBC. Outcomes of interest included overall survival (OS) and disease-free survival (DFS). Treatment effects were expressed as hazard ratios (HR) with 95% confidence interval (CI). Incidence rate ratios were calculated to estimate time to event outcomes for trials not reporting HR. A random-effects DerSimonian-Laird meta-analysis was conducted. Absolute effects were then obtained using baseline risks from the control arm of RCTs. Results: Of 4887 studies identified, 13 trials with 2529 patients were included in this meta-analysis. Most trials included patients with T2-4 and N0 patients and only 3 trials included patients with node positive disease. Total of 180 (47%) DFS events were observed with NAC+RC compared to 213 (56%) events in RC alone (HR: 0.72; 95% CI: 0.59-0.89) and 346 (58%) OS events were observed with NAC+ RC compared to 385 (52%) events in RC alone (HR: 0.80; 95% CI: 0.69-0.92). Total of 186 (70%) DFS events were observed with NAC + radiotherapy compared to 205 (71%) events in radiotherapy alone (HR: 0.91; 95% CI: 0.74-1.12) and 263 (58%) OS events were observed with NAC+ radiotherapy compared to 294 (61%) events in radiotherapy alone (HR:0.93; 95% CI: 0.79-1.08). Conclusions: The choice of definitive therapy after cisplatin-based NAC impacts survival in patients with MIBC. RC after NAC improved DFS (114 fewer events per 1000 events) and OS (76 fewer per 1000 events) whereas radiotherapy after NAC showed no survival benefit. [Table: see text]

3586 Background: HSP90 mediated chaperoning is a well-conserved biological mechanism for stabilization and activation of kinases. More than 60% of human kinases including VEGFR, CRAF, CSF1R, and FGFR are target of HSP90 (client kinases), whereas EGFR is non-client. CDC37 is a specific co-chaperone determining selectivity of client kinases recognized by HSP90. We hypothesized that gene expression levels of CDC37 have predictive values for anti-angiogenic therapies in mCRC. Methods: The subjects of this study were mCRC patients treated with regorafenib (REGO, Japanese retrospective cohort) and those treated with bevacizumab (BEV) or cetuximab (CET) in combination with first-line chemotherapy (CALGB/SWOG 80405 trial cohort). CDC37 expression levels were measured using RNA isolated from FFPE samples by nCounter gene expression profiling (Nanostring) and HiSeq 2500 (Illumina) in the Japanese and CALGB/SWOG 80405 cohorts, respectively. Overall survival (OS) and progression-free survival (PFS) were compared between patients with high CDC37 expression ( CDC37-H) and those with low expression ( CDC37-L), grouped by median cutoff value in each cohort. Results: In total, 484 patients were included (50 treated with REGO, 227 treated with BEV, and 207 treated with CET). In REGO-treated patients, CDC37-H showed significantly better OS (median 11.3 vs 6.0 months, adjusted hazard ratio [HR] 0.24, 95% confidence interval [CI] 0.11-0.54, p <0.01) and PFS (median 3.5 vs 1.9 months, adjusted HR 0.51, 95% CI 0.28-0.94, p = 0.03) compared to CDC37-L. Similarly, in BEV-treated patients, CDC37-H showed significantly better PFS (median 13.5 vs 9.6 months, adjusted HR 0.59, 95% CI 0.43-0.79, p <0.01) and numerically better OS (median 34.1 vs 29.4 months, adjusted HR 0.81, 95% CI 0.60-1.11, p = 0.20) compared to CDC37-L. However, in CET-treated patients, CDC37-H and CDC37-L patients showed similar OS (median 33.7 vs 26.1 months, adjusted HR 1.00, 95% CI 0.73-1.38, p = 0.98) and PFS (median 11.3 vs 11.0 months, adjusted HR 1.08, 95% CI 0.81-1.45, p = 0.60). Significant interaction was observed between CDC37 expression and treatment in terms of PFS in the CALGB/SWOG 80405 cohort ( p = 0.01). Conclusions: Our results suggest patients with CDC37-dependent ( CDC37-H) tumors may derive more benefit from REGO and BEV both of which target HSP90 client kinases or signaling pathways, but not from CET which target HSP90 non-client kinase. Further validation studies are warranted to develop a novel personalized approach for targeted therapies based on CDC37 expression in mCRC patients. Support: U10CA180821, U10CA180888; Pfizer, Genentech; https://acknowledgments.alliancefound.org. ClinicalTrials.gov Identifier: NCT00265850.

3580 Background: Monoamine oxidases (MAOs), including MAOA and MAOB, are mitochondrial enzymes responsible for catalyzing monoamine oxidation. Increased expression of MAOs were found in several cancer types and high MAOB was associated with worse disease stage and poorer survival in CRC. Positive and negative correlations of MAOB expression with mesenchymal type and epithelial type gene expressions, respectively, have been reported. Hence, we investigated whether MAOB expression is predictive for targeted therapies in mCRC. Methods: 430 mCRC pts treated with either bevacizumab (BEV, n = 224) or cetuximab (CET, n = 206) in combination with first-line chemotherapy within the CALGB/SWOG 80405 trial were included in the analysis. MAOB RNA was isolated from FFPE tumor samples and sequenced on the HiSeq 2500 (Illumina). Overall survival (OS) and progression-free survival (PFS) were compared between groups of pts categorized by tertiles of MAOB expression into high (H), medium (M) and low (L). Hazard ratios (HR) and 95% confidence intervals (CI) were computed from multivariable Cox proportional hazards model, adjusting for age, sex, location, number of metastases, KRAS, MSI status, and treatment with FOLFOX or FOLFIRI. Sensitivity analyses were conducted after stratifying by sex. Logrank P-values describe differences without adjustment for patient characteristics. Results: In CET-treated pts, MAOB-L showed significantly longer OS (median 39.2 vs 30.9 vs 15.9 months, logrank P = 4.7E-05, L vs H (as reference) adjusted HR 0.42, 95% CI [0.27, 0.65]) and PFS (median 13.2 vs 11.8 vs 7.6 months, logrank P = 0.006, L vs H adjusted HR 0.59 [0.40, 0.88]) compared to MAOB-M and MAOB-H, respectively. Similar results were observed when evaluating MAOB expression as a continuous variable. In BEV-treated pts, no significant differences were observed when comparing MAOB expression tertiles; however, pts with lower MAOB expression had significantly better OS, but not PFS, when evaluating MAOB as a continuous variable (Cox LRT P = 0.015, covariate adjusted). In CET-treated pts, the effect of MAOB expression was observed in male but not female pts (OS: median 40.3 vs 30.9 vs 16.1 months by MAOB-L, M, H, respectively, logrank P = 6.8E-05, L vs H adjusted HR 0.33 [0.19, 0.59]; PFS: median 13.8 vs 12.6 vs 7.9 months, logrank P = 0.001, L vs H adjusted HR 0.46 [0.28, 0.79]). A significant interaction was observed between MAOB expression and treatment for OS ( P = 0.010) in males and females combined, but only in males ( P = 0.018) when stratified by sex. Conclusions: Our results suggest that pts with MAOB-L tumors may have greater benefit from CET-based treatment and that targeting MAOB may be a promising strategy to improve patient outcomes. Further validation studies are warranted to develop a novel personalized approach based on MAOB expression in mCRC pts. Clinical trial information: NCT00265850.

3572 Background: Early onset colorectal cancer (eoCRC: disease diagnosed < 50) has been increasing over the past 2 decades. Currently, standard treatment recommendations for eoCRC patients (pts) with metastatic disease does not differ from late-onset CRC (loCRC) pts although outcomes data in eoCRC pts is limited. Methods: Individual patient data on 5,761 treatment-naive metastatic eoCRC pts was pooled from 8 phase II and III randomized EGFRi studies (2000 - 2012) from the ARCAD mCRC database. The distribution of demographics, clinicopathological features, and biomarkers were summarized by age groups. Progression-free survival (PFS) was compared between age groups by stratified Cox models, adjusting for potential confounders. Predictive value of age group was evaluated by testing interaction effect between treatment and age variables based on a subset of trials with concurrent randomizations between regimens with and without EGFRi Results: eoCRC (n=756) were more evenly distributed between gender, had improved performance status (PS), increased likelihood of metastatic resection, and distant lymph node metastasis, but were less likely to have lung metastasis or KRAS mutation compared to loCRC (n=5,005, table 1). eoCRC and loCRC patients had similar distribution of primary tumor sidedness, primary resection, liver and/or peritoneal involvement, number of metastatic sites involved, and BRAF mutations (MT). No difference in PFS for eoCRC versus loCRC pts was noted (7.8 vs. 7.9 months [M], adjusted hazard ratio [HRadj], 1.02, 95% confidence interval [CI], 0.93-1.11). Among pts with KRAS wild type (WT) and left sided primary tumors, univariable analysis of EGFRi demonstrated improved mPFS in loCRC (9.9 vs 8.5M, HR = 0.74, p<0.001), but this benefit was not seen in eoCRC (8.3 vs 8.9 months, HR 1.20, p=0.36). The same pattern was observed upon multivariable analysis (Table). Conclusions: In our pooled analysis, EGFRi + chemotherapy significantly improved PFS in treatment-naïve loCRC patients but not in left sided, KRAS WT, eoCRC patients. Further validation in an independent cohort is warranted. [Table: see text]

3553 Background: Higher levels of plasma 25-hydroxyvitamin D [25(OH)D)] are associated with better outcomes in mCRC, but underlying biologic mechanisms are unknown. Key components of the VitD metabolic pathway include CYP27B1 (encodes 1-α-hydroxylase, converts 25(OH)D to active calcitriol), VitD receptor (VDR), and CYP24A1 (encodes 24-hydroxylase, degrades calcitriol and 25(OH)D into excreted metabolites). Since these factors may affect 25(OH)D levels and potentially mediate VitD activity in mCRC, we examined the relationship between tumoral gene expression (GEx) of CYP27B1, VDR, and CYP24A and pt outcome in a study nested in a randomized phase III trial of first-line chemotherapy plus biologics in mCRC pts, CALGB/SWOG 80405. Methods: We determined GEx of CYP27B1, VDR, and CYP24A1 by RNA sequencing (RNA-Seq) of archival tumor samples using the Illumina TruSeq platform. Primary endpoints were overall (OS) and progression-free survival (PFS). Cox proportional hazard models were used to calculate hazard ratios (HRs) and 95% confidence intervals (CIs), adjusting for prognostic and molecular characteristics. Results: 562 pts with RNA-Seq data were included. Pts with higher CYP27B1 expression (>median) were less likely to have BRAF wild type (WT) (79% vs 90%) compared to pts with lower expression (p=0.0007). Pts with higher VDR expression (>quartile 1 [Q1]) were younger (median age 59 vs 62 years; p=0.03), more likely to have left-sided (63% vs 46%; p=0.0005) and BRAF WT tumors (89% vs 70%; p<0.0001), and less likely to have RAS WT tumors (70% vs 80%; p=0.02) compared to pts with lower VDR. Pts with higher CYP24A1 expression (>median) were more likely to have left-sided tumors compared to pts with lower expression (63% vs 54%; p=0.03). On multivariable analysis, pts with higher CYP27B1 expression had significantly improved OS (HR 0.84; 95% CI, 0.75-0.93; p=0.002) and PFS (HR 0.89; 95% CI, 0.80-0.99; p=0.04). Higher VDR expression (up to Q1) was associated with significantly improved PFS (HR 0.69; 95% CI, 0.53-0.91; p=0.007) but not OS (HR 0.85; 95% CI, 0.66-1.09; p=0.20). Above Q1, this improvement attenuated. Higher CYP24A1 GEx was not associated with improved OS (HR 0.98; 95% CI, 0.88-1.08; p=0.66) or PFS (HR 0.98; 95% CI, 0.89-1.08; p=0.68). We found no significant interactions between GEx of CYP27B1, VDR, or CYP24A with baseline plasma 25(OH)D levels (p for interaction ≥0.10 for all). Conclusions: Our findings suggest an association between GEx of VitD pathway markers, particularly CYP27B1 and VDR, and survival in pts with mCRC, lending biologic plausibility to a role of VitD in CRC pathogenesis. Future studies are needed to confirm these findings and elucidate underlying mechanisms of action. Clinical trial information: NCT00265850 .

AbstractBackgroundUropathogen resistance, fluoroquinolone‐resistance (FQR), and extended spectrum beta‐lactamase (ESBL), has been observed to be emerging worldwide with prevalences above recommended thresholds for routine empirical treatment. The primary aim of our study was to determine the prevalence of FQR from a geographically diverse sample of United States emergency departments (EDs).MethodsWe conducted a multi‐center, observational cohort study using a network of 15 geographically diverse US EDs. All patients ≥18 years of age with the primary or secondary diagnosis of urinary tract infection (UTI) in the ED identified using International Classification of Diseases (ICD‐10) diagnosis code of cystitis, pyelonephritis, or UTI from 2018 to 2020 were included. We calculated descriptive statistics for uropathogens and susceptibilities. Logistic regression analysis was used to identify antimicrobial resistance risk factors associated with FQR Escherichia coli.ResultsAmong 3779 patients who met inclusion criteria, median age was 62.9 years (interquartile range [IQR]: 41–77.6) and 76.3% were female. The most common diagnoses were complicated (41.2%) and uncomplicated cystitis (40.3%). E. coli was the most common pathogen (63.2%), followed by Klebsiella pneumoniae (13.2%) and Enterococcus species (5.8%). Across all sites, overall E. coli FQ‐resistance prevalence was 22.1%, ranging from 10.5 to 29.7% by site. The prevalence of ESBL‐producing uropathogen was 7.4%, ranging from 3.6% to 11.6% by site. Previous IV or oral antimicrobial use in the past 90‐days and history of a multi‐drug resistant pathogen were associated with FQ‐resistant E. coli (odds ratio [OR] 2.68, 95% confidence interval [CI]: 2.04–3.51, and OR 6.93, 95% CI: 4.95–9.70, respectively). Of the patients who had FQ‐resistant E. coli or an ESBL‐producing uropathogen isolated, 116 (37.1%) and 61 (36.7%) did not have any documented risk factors for resistance.ConclusionFQ‐resistant E. coli is widely prevalent across US sites highlighting the need for ongoing monitoring of antimicrobial resistance and, at some locations, modification of empirical treatments.