Purpose: DNA damage repair (DDR) defects are common across cancer types and can indicate therapeutic vulnerability. Optimal exploitation of DDR defects in prostate cancer requires new diagnostic strategies and a better understanding of associated clinical genomic features. Experimental Design: We performed targeted sequencing of 1,615 plasma cell-free DNA samples from 879 patients with metastatic prostate cancer. Depth-based copy-number calls and heterozygous SNP imbalance were leveraged to expose DDR-mutant allelic configuration and categorize mechanisms of biallelic loss. We used split-read structural variation analysis to characterize tumor suppressor rearrangements. Patient-matched archival primary tissue was analyzed identically. Results: BRCA2, ATM, and CDK12 were the most frequently disrupted DDR genes in circulating tumor DNA (ctDNA), collectively mutated in 15% of evaluable cases. Biallelic gene disruption via second somatic alteration or mutant allele–specific imbalance was identified in 79% of patients. A further 2% exhibited homozygous BRCA2 deletions. Tumor suppressors TP53, RB1, and PTEN were controlled via disruptive chromosomal rearrangements in BRCA2-defective samples, but via oncogene amplification in context of CDK12 defects. TP53 mutations were rare in cases with ATM defects. DDR mutations were re-detected across 94% of serial ctDNA samples and in all available archival primary tissues, indicating they arose prior to metastatic progression. Loss of BRCA2 and CDK12, but not ATM, was associated with poor clinical outcomes. Conclusions: BRCA2, ATM, and CDK12 defects are each linked to distinct prostate cancer driver genomics and aggression. The consistency of DDR status in longitudinal samples and resolution of allelic status underscores the potential for ctDNA as a diagnostic tool.

Investigation of histopathology slides by pathologists is an indispensable component of the routine diagnosis of cancer. Artificial intelligence (AI) has the potential to enhance diagnostic accuracy, improve efficiency, and patient outcomes in clinical pathology. However, variations in tissue preparation, staining protocols, and histopathology slide digitization could result in over-fitting of deep learning models when trained on the data from only one center, thereby underscoring the necessity to generalize deep learning networks for multi-center use. Several techniques, including the use of grayscale images, color normalization techniques, and Adversarial Domain Adaptation (ADA) have been suggested to generalize deep learning algorithms, but there are limitations to their effectiveness and discriminability. Convolutional Neural Networks (CNNs) exhibit higher sensitivity to variations in the amplitude spectrum, whereas humans predominantly rely on phase-related components for object recognition. As such, we propose Adversarial fourIer-based Domain Adaptation (AIDA) which applies the advantages of a Fourier transform in adversarial domain adaptation. We conducted a comprehensive examination of subtype classification tasks in four cancers, incorporating cases from multiple medical centers. Specifically, the datasets included multi-center data for 1113 ovarian cancer cases, 247 pleural cancer cases, 422 bladder cancer cases, and 482 breast cancer cases. Our proposed approach significantly improved performance, achieving superior classification results in the target domain, surpassing the baseline, color augmentation and normalization techniques, and ADA. Furthermore, extensive pathologist reviews suggested that our proposed approach, AIDA, successfully identifies known histotype-specific features. This superior performance highlights AIDA’s potential in addressing generalization challenges in deep learning models for multi-center histopathology datasets.

Purpose: DNA mismatch repair defects (MMRd) and tumor hypermutation are rare and under-characterized in metastatic prostate cancer. Furthermore, since hypermutated MMRd prostate cancers can respond to immune checkpoint inhibitors, there is an urgent need for practical detection tools. Experimental design: We analyzed plasma cell-free DNA targeted sequencing data from 433 metastatic prostate cancer patients with circulating tumor DNA (ctDNA) purity ≥2%. Samples with somatic hypermutation were subjected to 185× whole exome sequencing and capture of mismatch repair gene introns. Archival tissue was analyzed with targeted sequencing and immunohistochemistry. Results: 16 patients (3.7%) had somatic hypermutation with MMRd etiology, evidenced by deleterious alterations in MSH2, MSH6, or MLH1, microsatellite instability, and characteristic trinucleotide signatures. CtDNA was concordant with mismatch repair protein immunohistochemistry and DNA sequencing of tumor tissue. Tumor suppressors such as PTEN, RB1, and TP53 were inactivated by mutation rather than copy number loss. Hotspot mutations in oncogenes such as AKT1, PIK3CA and CTNNB1 were common, and the AR ligand binding domain was mutated in 9/16 patients. We observed high intra-patient clonal diversity, evidenced by subclonal driver mutations and shifts in mutation allele frequency over time. Patients with hypermutation and MMRd etiology in ctDNA had a poor response to Androgen Receptor inhibition and inferior survival compared to a control cohort. Conclusions: Hypermutated MMRd metastatic prostate cancer is associated with oncogene activation and subclonal diversity, which may contribute to a clinically aggressive disposition in selected patients. In patients with detectable ctDNA, cell-free DNA sequencing is a practical tool to prioritize this subtype for immunotherapy.

Large cell neuroendocrine carcinoma (LCNEC) of the urinary tract is a rare disease. We present a relatively large retrospective cohort of urinary LCNEC, 20 from the urinary bladder, and 2 from the ureter, from a single institution. The patients included 16 men and 6 women with a median age of 74.5 years. Most LCNEC presented at an advanced stage with tumors invading the muscularis propria and beyond (21/22). Eight cases were pure LCNEC, while 14 cases were mixed with other histologic types, including conventional urothelial carcinoma (n=9), carcinoma in situ (n=7), small cell carcinoma (n=6), and urothelial carcinoma with glandular (n=3) features. Most LCNEC expressed neuroendocrine markers synaptophysin (22/22), chromogranin (13/16), CD56 (7/7), TTF1 (8/8), and INSM1 (2/3). They were negative for common urothelial markers including HMWCK (0/3), p40/p63 (0/6), CK20 (0/10), and had variable GATA3 staining (4/8). Ki-67 stained 25% to nearly 100% tumor cell nuclei. Patient survival was associated with cancer stage, and pure LCNEC showed worse survival than mixed LCNEC. Compared with small cell carcinoma at similar stages from a prior study, LCNEC had a worse prognosis only when patients developed metastatic disease. For organ-confined LCNEC, neoadjuvant chemotherapy followed by radical resection is the treatment option to achieve long-term survival.