BACKGROUND AND OBJECTIVE:Detecting localized prostate cancer (PC) with metastatic potential-defined as unfavorable-histology PC (uhPC), comprising Grade Group (GG) ≥3 disease or GG 2 disease with cribriform/intraductal features-is critical for guiding appropriate intervention. We evaluated the utility of the Prostate Imaging Reporting and Data System (PI-RADS) and the automated Restriction Spectrum Imaging restriction score (RSIrs; a biophysics-based quantitative MRI biomarker) for uhPC detection and localization. METHODS:We evaluated patient-level detection of uhPC in a multicenter cohort with biopsy as the reference standard and lesion-level localization in a separate cohort with whole-mount histopathology (WMHP) from radical prostatectomy. The area under the receiver operating characteristic curve (AUC) was calculated to compare patient-level detection of uhPC using PI-RADS and RSIrs. PI-RADS and RSIrs were used to evaluate sensitivity for the most aggressive tumor within the prostate (index tumor) and for all uhPC tumors on WMHP. KEY FINDINGS AND LIMITATIONS:The AUC for patient-level detection of uhPC did not differ significantly between PI-RADS and RSIrs in 1022 patients from five centers (p = 0.13). At the lesion level (n = 103 patients), sensitivity for the index tumor was 87% (95% confidence intervals [CI], 79-94) for PI-RADS, 85% (95% CI, 78-93) for RSIrs, and 93% (95% CI, 86-98) for the two combined. For all uhPC tumors, sensitivity was 81% (95% CI, 73-90) for PI-RADS, 86% (95% CI, 78-93) for RSIrs, and 90% (95% CI, 82-97) for the two combined. A limitation of the lesion-level analyses was that only patients who opted for surgery could be included. CONCLUSIONS AND CLINICAL IMPLICATIONS:MRI showed high sensitivity for detecting uhPC, reinforcing its value for identifying biologically aggressive disease. Both PI-RADS and automated RSIrs may be useful for targeted biopsy and tumor-focused treatment, such as focal radiation dose escalation to aggressive intraprostatic lesions.

BACKGROUND:We aim to evaluate whether increased lymph node yield at prostatectomy (RP) is associated with improved outcomes in NRG/RTOG 9601, a randomized clinical trial of men who underwent either radiation (RT) alone or RT + bicalutamide for PSA elevation following RP for pT2/T3 prostate cancer. METHODS:We reviewed available pathology reports for patients in NRG/RTOG 9601 to determine the nodal count at RP. Cox proportional hazards models were used to assess effect of lymph nodes yield, arm (RT alone or RT + bicalutamide), Gleason score, positive margins, and seminal vesicle invasion on the following endpoints: times to local and distant failure and overall and disease-specific survival. RESULTS:Of 760 patients, 552 (73%, 276 in each arm) had complete data available. Median node count in the entire cohort was 6 (range: 0-33, IQR: 3-9). There were no significant differences between arms in terms of patient demographic or clinical characteristics, including total lymph nodes removed in either arm. There was no significant association between total lymph nodes and overall or disease-specific survival with both arms combined and when adjusting for arm. Notably, interaction analysis revealed that in seminal vesicle invasion, there was a significant association between lymph node yield and OS and DSS (HR = 0.91, 95% CI: 0.83-0.99, p = 0.034; HR = 0.87, 95% CI: 0.77-0.99, p = 0.029, respectively). CONCLUSIONS:Although lymph node yield in NRG/RTOG 9601 did not show association with adverse outcomes in the entire cohort or either arm alone, there was significant association between lymph node yield and adverse outcomes when seminal vesicle invasion was present. The therapeutic benefit of extensive lymph node dissection remains uncertain but could be more relevant in higher risk patients.

Histopathological features and genomic biomarkers are integral to the management of patients with high-grade urothelial carcinoma (HGUC). Morphological features are often reflective of underlying genomic alterations. In this study, we investigate the correlation between tumor mutation burden (TMB) and specific histopathological features in HGUC to identify surrogate markers for TMB when molecular testing may not be available. We hypothesized that increased nuclear pleomorphism and intense lymphocytic infiltration might correlate with higher TMB, aiding treatment decisions. In this retrospective study, 119 cases with TMB analysis by Next Generation Sequencing (NGS) were evaluated for cytological and nuclear pleomorphism, lymphocytic infiltration, and histological subtypes. Pleomorphism and lymphocytic infiltrate were classified into binary categories (mild versus marked) for improved reproducibility. TMB > 10 mutations/Mb was defined as high TMB. Both marked pleomorphism and marked lymphocytic infiltration were associated with higher TMB (10.3 vs 7.2 muts/Mb, p=0.027 and 12.4 vs. 8.9 muts/Mb, p=0.005 respectively). Micropapillary, squamous, and sarcomatoid subtypes demonstrated higher median TMB, although the differences did not reach statistical significance. In summary, our findings suggest that marked nuclear pleomorphism and lymphocytic infiltrate are associated with higher TMB, indicating their potential as surrogate markers for immunotherapy response.

INTRODUCTION:A recent study proposed a system of unfavourable and favourable histology categories to risk-stratify prostate cancer patients. We hypothesized that unfavourable histology would be associated with higher Decipher radical prostatectomy (RP) scores and would improve biochemical recurrence (BCR) prediction. MATERIALS AND METHODS:The RP slide with Decipher testing was classified as having either unfavourable or favourable histology, with unfavourable histology defined as the presence of Gleason pattern 5, large cribriform (diameter >0.25 mm), intraductal carcinoma, complex intraluminal papillary architecture, anastomosing cords of epithelium with or without cribriform spaces and/or grade 3 stromogenic carcinoma. Favourable histology was defined as the absence of unfavourable histology patterns. The association between favourable/unfavourable histology and Decipher RP score was evaluated using the Mann-Whitney test. Univariable and multivariable analyses to test for a statistically significant association between the predictors favourable/unfavourable histology, age, PSA at diagnosis, race, RP grade group, pT and pN stage and margin status and the risk of BCR were performed. RESULTS:Four hundred and eleven RP cases from 2014 to 2020 were classified as 306 (74.5%) unfavourable and 105 (25.5%) favourable. Median Decipher scores were 0.7 for unfavourable and 0.52 for favourable (P < 0.001), corresponding to Decipher assay high and intermediate risk categories, respectively. On univariate analysis, unfavourable histology compared to favourable histology was significantly associated with decreased time to BCR, which remained significant in a multivariable model (HR 1.65; 95% CI 1.04-2.64, P = 0.035). CONCLUSIONS:Unfavourable histology was associated with higher Decipher scores and increased BCR risk. The addition of the histology category improved BCR prediction in a multivariable model.

413 Background: Effective prostate cancer (PCa) risk stratification is paramount to guide treatment considerations across all stages of PCa. Serum PSA, Gleason score (GS), and bulk-profiling tests (e.g. Decipher) form the cornerstone of risk stratification in clinical practice. New techniques such as Imaging Mass Cytometry (IMC) now enable tumor profiling with unprecedented resolution. We developed a novel IMC assay with the goal of identifying single cell subtypes prognostic of clinical outcomes in localized PCa. Methods: Spatial single-cell expression profiling with IMC was performed on primary PCa tumor biopsies and paired benign prostate samples obtained from patients with localized PCa who subsequently underwent radical prostatectomy. Cell subtypes were defined using the Phenograph cell-clustering approach. Per-sample “cell fraction” (percent abundance) of each cell subtype was defined as cell count of that subtype divided by total number of cells in the sample. Association between cell fraction of each cell subtype and both GS and Canary risk score was assessed using one-way analysis of variance (ANOVA) with Tukey's multiple comparison test. Biochemical progression-free survival (bPFS) and cancer-specific survival (CSS) were prespecified clinical endpoints. Analyses stratified by cell fraction tertiles were performed using the Kaplan-Meier method with Cox proportional hazards testing for significance. All hypothesis tests were performed using a two-tailed significance level of 0.05. Results: Co-expression patterns of 40 selected proteins in 3,429,844 cells comprising 604 biopsy samples obtained from 393 patients were measured. 28 of 393 patients (7%) were assigned a GS of 4+3 or higher. Single-cell clustering analysis revealed 16 distinct prostate, stromal, and immune cell subtypes including androgen-driven (AR+ PSMA+ KLK2+ CD46+) luminal prostate cancer cells (Subtypes 1, 8, 10), basal epithelial cells (Subtype 9), CD8+ T-cells (Subtype 14), CD4+ T-cells (Subtype 15), and antigen-presenting cells (Subtype 12). Subtypes 2, 7, and 9 were enriched in benign and low-risk samples; Subtypes 1, 6 and 12 were enriched in samples with GS of 4+3 or higher (P<0.001) and Canary high-risk ( P <0.01) samples. Patients with tumors enriched for Subtypes 6 and 12 demonstrated shorter bPFS ( P =0.008 and P =0.001 respectively) and shorter CSS ( P =0.02 and P =0.002 respectively) than those with tumors not enriched for these cell subtypes. Multivariable analysis including GS and Canary risk score revealed that Subtype 6 and Subtype 12 cell fraction were both independently prognostic of bPFS and CSS ( P <0.05). Conclusions: In our cohort of men with localized PCa, we identified single-cell features associated with biochemical recurrence and CSS that complement existing risk stratification approaches. Future work includes validation in independent cohorts and further investigation of spatial co-localization patterns between these prognostic cell types and neighboring cells of the tumor microenvironment.

BACKGROUND AND OBJECTIVE:Biochemical recurrence (BCR) after radical prostatectomy (RP) is a heterogeneous disease state in prostate cancer with multiple treatment options. Improved risk stratification could enable more personalized decision-making. We developed and validated a digital pathology-based multimodal artificial intelligence (MMAI) model to predict outcomes in post-RP BCR patients undergoing salvage therapy. METHODS:An MMAI model was trained to predict distant metastasis (DM) using prostate histopathology image features and clinical variables (pathologic grade group, pathologic T stage, prostate-specific antigen level before salvage radiotherapy [SRT], age, and surgical margin). The locked model was validated in 533 patients from NRG/RTOG 9601 and 0534 treated with SRT ± hormone therapy (HT), using Cox regression and time-dependent area under the receiver operating characteristic curve. KEY FINDINGS AND LIMITATIONS:With a median follow-up of 9.3 yrs, MMAI score was significantly associated with DM (subdistribution hazard ratio = 2.17 per standard deviation [95% confidence interval 1.65-2.85]; p < 0.001) and remained independently prognostic after adjusting for clinical variables and treatment. The 10-yr time-dependent area under the receiver operating characteristic curve for MMAI was 0.74 compared with 0.68 for a clinical nomogram. Binary risk categorization demonstrated higher 10-yr DM incidence in the MMAI high-risk (25%) than in the low-risk (8.8%) group. The absolute reduction in 10-yr DM incidence with HT plus SRT versus SRT alone was 21% in the high-risk group versus 2.5% in the low-risk group. Limitations include the use of archived trial cohorts. CONCLUSIONS AND CLINICAL IMPLICATIONS:The post-RP MMAI model provides individualized risk estimates after SRT ± HT and may support shared decision-making about salvage treatment. External and prospective validation are ongoing.

Background The SAKK 09/10 trial randomized biochemically recurrent prostate cancer patients to salvage radiation 64 Gy versus 70 Gy, and the NRG/RTOG 0126 randomized intermediate-risk prostate cancer patients to definitive radiation 70.2 Gy versus 79.2 Gy. We investigated a previously developed Post-Operative Radiation Therapy Outcomes Score (PORTOS) to identify preferential benefit from radiation dose escalation (DE). Materials and methods PORTOS was evaluated in patients enrolled in SAKK 09/10 and NRG/RTOG 0126 with available tissue that passed quality control (n = 226, 215). PORTOS was evaluated in the published post-operative groups in SAKK 09/10 and in tertiles in NRG/RTOG 0126 as cut-offs had not been established for biopsy samples and definitive radiation patients. Clinical and molecular correlates in a real-world dataset of 42 407 prostatectomy and 31 107 biopsy samples were also analyzed. Results In SAKK 09/10, the biomarker-treatment interaction was statistically significant between PORTOS (lower versus higher) and treatment arm for clinical progression-free survival. Only patients in the higher PORTOS group benefited from DE. In NRG/RTOG 0126, in patients with a lower tertile PORTOS, there was no difference in Phoenix biochemical failure (BF). However, for patients in the average and higher tertile PORTOS range, there was a significant benefit for DE for Phoenix BF. An interaction test indicated a significant difference in benefit for DE between higher and lower PORTOS groups. PORTOS was not strongly associated with clinicopathological variables in either trial or the large real-world dataset. In the latter, PORTOS was modestly associated with hypoxia signatures and strongly associated with immune signatures and subtypes. Conclusion In the SAKK 09/10 and RTOG 0126 randomized controlled trials, we demonstrated that PORTOS can potentially identify a subset of patients who benefit from DE, a subgroup that cannot be identified using clinicopathological or prognostic variables. These results suggest that PORTOS could be used clinically as a predictor of radiation response.

AIMS:A dichotomous classifier for prostatic adenocarcinoma, favourable and unfavourable histology, was recently proposed to optimize metastasis prediction. We evaluate interobserver agreement for its classification. METHODS AND RESULTS:Fifty biopsies with prostatic adenocarcinoma were selected. One slide per biopsy was scanned using the Aperio AT2 scanner (20x) and viewed in Leica Web Viewer. Ten pathologists reviewed each slide after completing a training module. Biopsies were scored for Grade Group (1-5) and presence of unfavourable histology. Forty-one of 50 biopsies (82%) had unanimous consensus [13 (32%) unfavourable and 28 (68%) favourable], 47 (94%) for at least 9 of 10 reviewers, and 48 (96%) for at least 8 of 10 reviewers. Percent agreement for favourable/unfavourable classification was 94%, kappa = 0.88 (95% CI: 0.8, 0.96), and Gwet's AC1 coefficient = 0.90 (95% CI: 0.82, 0.97) [both P-value < 0.001]. Inter-rater agreement for Grade Group designation was evaluated in biopsies with 100% consensus. In the favourable consensus group (n = 28), kappa for Grade Group was 0.80 (95% CI: 0.62, 0.98) and Kendall's W was 0.83 (95% CI: 0.81, 0.89), with both P-value < 0.001. For the unfavourable consensus group (n = 13), kappa for Grade Group was 0.56 (95% CI: 0.36, 0.75) and Kendall's W was 0.78 (95% CI: 0.76, 0.85), with both P-value < 0.001. CONCLUSIONS:The classification of prostatic adenocarcinoma as favourable/unfavourable histology has high interobserver agreement, even when learnt through a simple tutorial. GG designation was more variable. This provides evidence that the favourable/unfavourable classification could be utilized in routine practice given its ease of use.

BACKGROUND AND OBJECTIVE:A biopsy diagnosis of Gleason grade group (GG) 3 prostate cancer (PC) automatically classifies patients as having at least unfavorable intermediate-risk disease warranting definitive treatment. We hypothesized that GG3 PCs are not equally unfavorable. METHODS:The Urologic Outcomes Database at University of California-San Francisco was queried for men with localized, nonmetastatic PC diagnosed after 2000 who underwent radical prostatectomy (RP). The primary outcome was recurrence, defined as either biochemical failure (two prostate-specific antigen results ≥0.2 ng/ml) or salvage treatment. Multivariable Cox proportional-hazards regression models were used to calculate associations with the risk of recurrence, adjusted for clinicodemographic and postoperative factors. KEY FINDINGS AND LIMITATIONS:We included 4934 men who underwent RP in the analysis, of whom 862 (17%) were diagnosed with GG3 PC on biopsy. Cancer of the Prostate Risk Assessment postsurgery (CAPRA-S) scores overall increased over time, but remained broadly distributed. Multivariable analysis controlled for postoperative factors with CAPRA-S revealed that favorable biopsy Gleason histology (not expansile cribriform or intraductal carcinoma) was the strongest factor associated with lower risk of recurrence after RP (hazard ratio [HR] 0.61, 95% confidence interval [CI] 0.41-0.91), independent of the percentage of pattern 4. A higher percentage of positive cores (PPC) was also significantly associated with the risk of recurrence (HR per 10% increment: 1.06, 95% CI 1.01-1.11). Limitations include the retrospective nature of the single-institution study and the homogeneous study population. CONCLUSIONS AND CLINICAL IMPLICATIONS:Patients with GG3 PC on diagnostic biopsy have heterogeneous risk. Unfavorable biopsy histology and higher PPC were significantly associated with the risk of recurrence after RP after controlling for CAPRA-S scores. Not all GG3 cancers are equally unfavorable, and differential management may be warranted.

PURPOSE:Long-term androgen deprivation therapy (ADT) improves survival in men with high-risk localized prostate cancer (PCa) receiving radiotherapy (RT). Predictive biomarkers are needed to guide ADT duration. METHODS:A multimodal artificial intelligence (MMAI)-derived predictive biomarker was trained for long-term (LT) versus short-term (ST) ADT using pretreatment digital prostate biopsy images and clinical data (age, prostate-specific antigen, Gleason, and T stage) from six NRG Oncology phase III randomized radiotherapy trials. The novel MMAI-derived biomarker was developed to predict the differential benefit of LT-ADT on the primary end point, distant metastasis (DM). MMAI predictive utility was validated on a seventh randomized trial, RTOG 9202 (N = 1,192), which randomly assigned men to RT + ST-ADT (4 months) versus RT + LT-ADT (28 months). Fine-Gray and cumulative incidence analyses for DM, and secondarily, death with DM, were performed. Deaths without DM were treated as competing risks. RESULTS:In the validation cohort (median follow-up, 17.2 years), LT-ADT significantly improved DM from 26% to 17% (subdistribution hazard ratio [sHR], 0.64 [95% CI, 0.50 to 0.82], P < .001). A significant biomarker-treatment predictive interaction was observed (P = .04) for DM, whereby MMAI biomarker-positive men (n = 785, 66%) had reduced DM with LT-ADT versus ST-ADT (sHR, 0.55 [95% CI, 0.41 to 0.73], P < .001), whereas no treatment benefit was observed for MMAI biomarker-negative men (n = 407; sHR, 1.06 [95% CI, 0.61 to 1.84], P = .84). The estimated 15-year DM risk difference between RT + LT-ADT and RT + ST-ADT was 14% in MMAI biomarker-positive men and 0% in MMAI biomarker-negative men. The MMAI biomarker was also prognostic for DM, irrespective of treatment (sHR, 2.35 [95% CI, 1.72 to 3.19], P < .001). CONCLUSION:To our knowledge, the MMAI model is the first validated predictive biomarker to guide ADT duration with RT in localized/locally advanced PCa. Approximately one third of men with high-risk PCa could safely be spared the additional 24 months of ADT and the associated morbidity.

PURPOSE Artificial intelligence (AI) tools could improve clinical decision making or exacerbate inequities because of bias. African American (AA) men reportedly have a worse prognosis for prostate cancer (PCa) and are underrepresented in the development genomic biomarkers. We assess the generalizability of tools developed using a multimodal AI (MMAI) deep learning system using digital histopathology and clinical data from NRG/Radiation Therapy Oncology Group PCa trials across racial subgroups. METHODS In total, 5,708 patients from five randomized phase III trials were included. Two MMAI algorithms were evaluated: (1) the distant metastasis (DM) MMAI model optimized to predict risk of DM, and (2) the PCa-specific mortality (PCSM) MMAI model optimized to focus on prediction death in the presence of DM (DDM). The prognostic performance of the MMAI algorithms was evaluated in AA and non-AA subgroups using time to DM (primary end point) and time to DDM (secondary end point). Exploratory end points included time to biochemical failure and overall survival with Fine-Gray or Cox proportional hazards models. Cumulative incidence estimates were computed for time-to-event end points and compared using Gray's test. RESULTS There were 948 (16.6%) AA patients, 4,731 non-AA patients (82.9%), and 29 (0.5%) patients with unknown or missing race status. The DM-MMAI algorithm showed a strong prognostic signal for DM in the AA (subdistribution hazard ratio [sHR], 1.2 [95% CI, 1.0 to 1.3]; P = .007) and non-AA subgroups (sHR, 1.4 [95% CI, 1.3 to 1.5]; P < .001). Similarly, the PCSM-MMAI score showed a strong prognostic signal for DDM in both AA (sHR, 1.3 [95% CI, 1.1 to 1.5]; P = .001) and non-AA subgroups (sHR, 1.5 [95% CI, 1.4 to 1.6]; P < .001), with similar distributions of risk. CONCLUSION Using cooperative group data sets with a racially diverse population, the MMAI algorithm performed well across racial subgroups without evidence of algorithmic bias.