Biology subjects

Costello, J. C.

Publications and source records attributed to Costello, J. C..

17 recordsLinked to original sources

bioRxiv2026

Stromal innervation is prognostic in non-muscle-invasive bladder cancer and predicted by a distinct microenvironmental transcriptomic signature

Peripheral nerves are increasingly recognized as active components of the tumor microenvironment (TME). Yet, their role in non-muscle-invasive bladder cancer (NMIBC) and response to Bacillus Calmette-Guerin (BCG) remains poorly defined. Using multispectral imaging, we characterized autonomic and sensory nerve populations within the tumor-adjacent stroma of 241 treatment-naive high-risk NMIBC tumor regions from 141 patients. Stromal tumor-associated nerve (STaN) abundance varied substantially across patients and was associated with aggressive clinicopathological features, BCG failure, and disease progression. Unsupervised clustering resolved distinct STaN phenotypic groups with differential clinical and spatial associations. Noradrenergic and a neurochemically heterogeneous STaN populations showed the strongest associations with adverse outcomes. In contrast, a mixed population of nerves with high expression of the synaptic vesicle protein synaptophysin exhibited prominent spatial remodeling associated with BCG failure, despite limited prognostic value by abundance alone. Integration with an independent single-cell atlas nominated cancer-associated fibroblasts and mast cells as candidate cellular intermediaries of neural communication within the TME. Matched transcriptomic profiling revealed coordinated neuronal, extracellular matrix, lipid metabolism, and immune programs in STaN-high tumors, whereas STaN-low tumors preferentially enriched proliferative and hypoxic stress-response pathways. Leveraging these transcriptional features, we developed a transcriptomic classifier that distinguished STaN-high from STaN-low tumors in an independent test cohort (AUC = 0.87) and stratified clinical outcomes in additional patient cohorts. Together, these findings establish tumor innervation as a heterogeneous and clinically relevant feature of the NMIBC microenvironment, with neural abundance, phenotype, and spatial organization capturing distinct aspects of tumor behavior and clinical outcome.

cancer biology↗

bioRxiv2026

HetNetEX: Exact Asymptotic Inference in Heterogeneous Biomedical Knowledge Graphs

Heterogeneous biomedical knowledge networks (hetnets) integrate disparate data types, drugs, genes, diseases, and pathways, across independent sources; Hetionet (https://het.io) is a widely used example. A standard approach for assessing connectivity significance is XSwap, which permutes the hetnet P times and fits a gamma-hurdle null model to the degree-weighted path count (DWPC), pooling permuted values across pairs with matching source and target degrees to increase the effective sample size. This permutation approach has been highly successful in practice, but it faces four practical constraints in large graphs: (1) a finite resolution for the smallest reportable p-values, (2) computational cost that grows prohibitive at path lengths L [≥] 4 or 5, (3) a variance model (Var {propto} {micro}2) that departs from the configuration-model form (1 +{kappa} ){micro}, and (4) O(P 10m L) runtime. To complement this approach, we present HetNetEX (Heterogeneous Network EXact inference), which computes the null DWPC distribution analytically from degree sequences using the configuration model in O(Ln) time. In simulations at P = 200 across L = 1-4, HetNetEX achieves Spearman{rho} > 0.96 concordance with XSwap rankings while being >10,000x faster and providing analytical p-values without a resolution ceiling. High-degree pairs show larger XSwap sampling error than low-degree pairs, reflecting the finite-sample nature of permutation that analytical computation avoids.

systems biology↗

bioRxiv2025

eIF3d and eIF3e mediate selective translational control of hypoxia that can be inhibited by novel small molecules

Exposure to hypoxia is linked to increased cellular plasticity and enhanced metastasis; effects which are primarily attributed to the transcriptional activation of large gene programs downstream of hypoxia inducible factors (HIFs). However, translational effects in hypoxia, that likely precede transcriptional effects, have remained largely unexplored. Using ribosome-profiling, we uncovered a selective translational response in acute hypoxia that is eIF3d/eIF3e-dependent and controls downstream hypoxic responses including HIF1a accumulation and cellular invasion. We further demonstrated that eIF3e copy number and an eIF3e-expression signature are associated with worsened outcomes for breast cancer patients. Finally, we identified a class of novel small molecules that target eIF3e specifically, reducing the translational response to hypoxia and to ER stress, another stressor that is dependent on eIF3d/eIF3e-mediated translation. Our data uncover critical functions for eIF3d/eIF3e in the hypoxic response and identify a potential means to inhibit stress-induced translation, and potentially plasticity and metastasis, mediated by eIF3e.

cancer biology↗

bioRxiv2024

An EYA3/NF-κB/CCL2 signaling axis suppresses cytotoxic NK cells in the pre-metastatic niche to promote triple negative breast cancer metastasis

Patients with Triple Negative Breast Cancer (TNBC) exhibit high rates of metastases and poor prognoses. The Eyes absent (EYA) family of proteins are developmental transcriptional cofactors/phosphatases that are re-expressed and/or upregulated in numerous cancers. Herein, we demonstrate that EYA3 correlates with decreased survival in breast cancer, and that it strongly, and specifically, regulates metastasis via a novel mechanism that involves NF-kB signaling and an altered innate immune profile at the pre-metastatic niche (PMN). Remarkably, restoration of NF-kB signaling downstream of Eya3 knockdown (KD) restores metastasis without restoring primary tumor growth, isolating EYA3/NF-kB effects to the metastatic site. We show that secreted CCL2, regulated downstream of EYA3/NF-kB, specifically decreases cytotoxic NK cells in the PMN and that re-expression of Ccl2 in Eya3-KD cells is sufficient to rescue activation/levels of cytotoxic NK cells in vitro and at the PMN, where EYA3-mediated decreases in cytotoxic NK cells are required for metastatic outgrowth. Importantly, analysis of public breast cancer datasets uncovers a significant correlation of EYA3 with NF-kB/CCL2, underscoring the relevance of EYA3/NF-kB/CCL2 to human disease. Our findings suggest that inhibition of EYA3 could be a powerful means to re-activate the innate immune response at the PMN, inhibiting TNBC metastasis. SignificanceEYA3 promotes metastasis of TNBC cells by promoting NF-kB-mediated CCL2 expression and inhibiting cytotoxic NK cells at the pre-metastatic niche, highlighting a potential therapeutic target in this subset of breast cancer.

cancer biology↗

bioRxiv2023

curatedPCaData: Integration of clinical, genomic, and signature features in a curated and harmonized prostate cancer data resource

Genomic and transcriptomic data have been generated across a wide range of prostate cancer (PCa) study cohorts. These data can be used to better characterize the molecular features associated with clinical outcomes and to test hypotheses across multiple, independent patient cohorts. In addition, derived features, such as estimates of cell composition, risk scores, and androgen receptor (AR) scores, can be used to develop novel hypotheses leveraging existing multi-omic datasets. The full potential of such data is yet to be realized as independent datasets exist in different repositories, have been processed using different pipelines, and derived and clinical features are often not provided or unstandardized. Here, we present the curatedPCaData R package, a harmonized data resource representing >2900 primary tumor, >200 normal tissue, and >500 metastatic PCa samples across 19 datasets processed using standardized pipelines with updated gene annotations. We show that meta-analysis across harmonized studies has great potential for robust and clinically meaningful insights. curatedPCaData is an open and accessible community resource with code made available for reproducibility.

bioinformatics↗

bioRxiv2022

Microparticle-delivered Cxcl9 delays the relapse of Braf inhibitor-treated melanoma

Patients with BRAF-mutant melanoma show significant responses to combined BRAF and MEK inhibition, but most relapse within 2 years. A major reservoir for drug resistance is minimal residual disease (MRD), comprised of drug-tolerant tumor cells laying in a dormant state. Towards exploiting potential therapeutic vulnerabilities of MRD, we established a genetically engineered mouse model of BrafV600E-driven melanoma MRD wherein genetic BrafV600E extinction leads to strong but incomplete tumor regression. Transcriptional time-course analysis after BrafV600E extinction revealed that after an initial surge of immune activation, tumors later became immunologically "cold" after MRD establishment. Computational analysis identified candidate T-cell recruiting chemokines that may be central players in the process, being strongly upregulated initially and steeply decreasing as the immune response faded. Therefore, we hypothesized that sustaining the chemokine signaling could impair MRD maintenance through increased recruitment of effector T-cells. We show that intratumoral administration of recombinant Cxcl9, either naked or loaded in microparticles, significantly impaired MRD relapse in BRAF-inhibited tumors, including several complete responses after microparticle-delivered rCxcl9 combined with BRAF and MEK-inhibition. Our experiments constitute a proof of concept that chemokine-based microparticle delivery systems are a potential strategy to forestall tumor relapse and thus improve the clinical success of frontline treatment methods.

cancer biology↗

bioRxiv2021

Generative neural networks separate common and specific transcriptional responses

Genome-wide transcriptome profiling identifies genes that are prone to differential expression across contexts ("common DEGs"), as well as genes with changes specific to the experimental manipulation. Distinguishing common DEGs from those that are specifically changed in a context of interest allows more efficient prediction of which genes are specific to a given biological process under scrutiny. Currently, commonly differentially expressed genes or pathways can only be identified through the laborious manual curation of highly controlled experiments, an inordinately time-consuming and impractical endeavor. Here we pioneer an approach for identifying common patterns using generative neural networks. This approach produces a background set of transcriptomic experiments from which a null distribution of gene and pathway changes can be generated. By comparing the set of differentially expressed genes found in a target experiment against the generated background set, common results can be easily separated from specific ones. This "Specific cOntext Pattern Highlighting In Expression data" (SOPHIE) approach is broadly applicable to new platforms or any species with a large collection of gene expression data. We apply SOPHIE to diverse datasets including those from human, human cancer, and the bacterial pathogen Pseudomonas aeruginosa. SOPHIE identifies common DEGs in concordance with previously described, manually and systematically determined common DEGs. Further, molecular validation indicates that SOPHIE detects highly specific, but low magnitude, biologically relevant, transcriptional changes. SOPHIEs measure of specificity can complement log fold change values generated from traditional differential expression analyses. For example, by filtering the set of differentially expressed genes, one can identify those genes that are specifically relevant to the experimental condition of interest. Consequently, these results can inform future research directions.

bioinformatics↗

bioRxiv2021

PRDM proteins control Wnt/β-catenin activity to regulate craniofacial chondrocyte differentiation

Cranial neural crest (NCC)-derived chondrocyte precursors undergo a dynamic differentiation and maturation process to establish a scaffold for subsequent bone formation, alterations in which contribute to congenital birth defects. Here, we demonstrate that transcription factor and histone methyltransferase proteins Prdm3 and Prdm16 control the differentiation switch of cranial NCCs to craniofacial cartilage. Loss of either results in hypoplastic and unorganized chondrocytes due to impaired cellular orientation and polarity. We show that PRDMs regulate cartilage differentiation by controlling the timing of Wnt/{beta}-catenin activity in strikingly different ways: prdm3 represses while prdm16 activates global gene expression, though both by regulating Wnt enhanceosome activity and chromatin accessibility. Finally, we show that manipulating Wnt/{beta}-catenin signaling pharmacologically or generating prdm3-/-;prdm16-/- double mutants rescues craniofacial cartilage defects. Our findings reveal upstream regulatory roles for Prdm3 and Prdm16 in cranial NCCs to control Wnt/{beta}-catenin transcriptional activity during chondrocyte differentiation to ensure proper development of the craniofacial skeleton. HighlightsO_LIPrdm3 and Prdm16 are required for chondrocyte organization in vertebrate craniofacial cartilage C_LIO_LILoss of Prdm3 and Prdm16 alters expression of Wnt/{beta}-catenin signaling components C_LIO_LIPrdm3 and Prdm16 oppositely control global chromatin accessibility C_LIO_LIPrdm3 and Prdm16 cartilage defects can be rescued pharmacologically or genetically C_LI

developmental biology↗

bioRxiv2021

SIX1 is a master regulator of the Rhabdomyosarcoma undifferentiated state

Rhabdomyosarcoma (RMS) is a pediatric skeletal muscle sarcoma characterized by the expression of the myogenic-lineage transcription factors (TF) MYOD1 and MYOG. Despite high expression of these TFs, RMS cells fail to terminally differentiate, suggesting the presence of factors that alter their function. Here, we demonstrate that the developmental TF, SIX1, is highly expressed in RMS and is critical to maintain a muscle progenitor-like state. SIX1 loss induces terminal differentiation of RMS cells into myotube-like cells and dramatically impedes tumor growth in vivo. We show that SIX1 maintains the RMS undifferentiated state by controlling enhancer activity and MYOD1 occupancy at loci more permissive to tumor growth over terminal muscle differentiation. Finally, we demonstrate that a gene signature derived from SIX1 loss correlates with differentiation status in RMS and predicts RMS progression in human disease. Our findings demonstrate a master regulatory role for SIX1 in the repression of RMS differentiation via genome-wide alterations in MYOD1-mediated transcription. HighlightsO_LISIX1 prevents differentiation in RMS while it promotes differentiation during normal development C_LIO_LIFN-RMS are highly dependent on SIX1 for growth in both zebrafish and mouse xenograft models C_LIO_LILoss of SIX1 alters the transcriptional landscape of RMS cells, inducing a growth to differentiation switch C_LIO_LISIX1 knockdown in FN-RMS causes reduced super enhancer-based activity at stem-related genes and enhanced MYOD1 binding to differentiation loci, resulting in the activation of a myogenic differentiation program C_LIO_LIA gene signature derived from SIX1 loss strongly correlates with myogenic differentiation status and is predictive of advanced RMS. C_LI

cancer biology↗

bioRxiv2021

The M1 aminopeptidase NPEPPS is a novel regulator of cisplatin sensitivity

There is an unmet need to improve efficacy of platinum-based cancer chemotherapy. Using multi-omic assessment of cisplatin-responsive and -resistant human bladder cancer cell lines and whole-genome CRISPR screens, we identified Puromycin-Sensitive Aminopeptidase, NPEPPS, as a novel driver of cisplatin resistance. NPEPPS depletion sensitizes resistant bladder cancer cells to cisplatin in vitro and in vivo. Conversely, overexpression of NPEPPS in sensitive cells increased cisplatin resistance. We show that NPEPPS affects treatment response by regulating intracellular cisplatin concentrations. Patient-derived organoids (PDOs) generated from bladder cancer samples before and after cisplatin-based treatment, and from patients who did not receive cisplatin, were evaluated for sensitivity to cisplatin and they were found to be concordant with clinical response. In PDOs, shRNA depletion or pharmacologic inhibition of NPEPPS led to increased cisplatin sensitivity, while NPEPPS overexpression had the opposite effect. Our data present NPEPPS as a novel and druggable driver of cisplatin resistance by regulating intracellular cisplatin concentrations, along with providing the preclinical data to support clinical trials combining NPEPPS inhibition with cisplatin.

cancer biology↗

bioRxiv2020

Mediator of DNA damage checkpoint 1 (MDC1) is a novel estrogen receptor co-regulator in invasive lobular carcinoma of the breast

Invasive lobular carcinoma (ILC) is the most common histological subtype of breast cancer, and nearly all ILC tumors express estrogen receptor alpha (ER). However, clinical and laboratory data suggest ILC are strongly estrogen-driven but not equally sensitive to anti-estrogen therapies. We hypothesized that ILC-specific ER transcriptional co-regulators mediate ER functions in ILC and anti-estrogen resistance, and profiled ER-associated proteins by mass spectrometry. Three ER+ ILC cell lines, MDA MB 134VI, SUM44PE, and BCK4, were compared to published data from ER+ invasive ductal carcinoma (IDC) cell lines, and we examined whether siRNA knockdown of identified proteins suppressed ER-driven proliferation in ILC cells. This approach found mediator of DNA damage checkpoint 1 (MDC1), a key tumor suppressor in DNA damage response (DDR), as a putative novel ER co-regulator in ILC. We confirmed ER:MDC1 interaction was specific to ILC cell lines versus IDC cells, and found MDC1 knockdown suppressed ILC cell proliferation and suppressed tamoxifen resistance in MDA MB 134VI. Using RNA-sequencing, we found that in ILC cells, MDC1 knockdown broadly dysregulates the estrogen-driven ER transcriptome, with ER:MDC1 target genes enriched for hormone-response-elements in their promoter regions. Importantly, our data are inconsistent with MDC1 regulating ER via MDC1 DDR and tumor suppressor functions, but instead suggest a novel oncogenic role for MDC1 in mediating ER transcriptional activity as a co-regulator. Supporting this, in breast tumor tissue microarrays MDC1 protein was frequently low or absent in IDC or ER-ILC, but MDC1 loss is rare in ER+ ILC. ER:MDC1 interaction and MDC1 co-regulator functions may underlie cell type-specific ER functions in ILC, and serve as important biomarkers and therapeutic targets to overcome anti-estrogen resistance in ILC.

cancer biology↗

bioRxiv2020

Crowdsourcing assessment of maternal blood multi-omics for predicting gestational age and preterm birth

Identification of pregnancies at risk of preterm birth (PTB), the leading cause of newborn deaths, remains challenging given the syndromic nature of the disease. We report a longitudinal multi-omics study coupled with a DREAM challenge to develop predictive models of PTB. We found that whole blood gene expression predicts ultrasound-based gestational ages in normal and complicated pregnancies (r=0.83), as well as the delivery date in normal pregnancies (r=0.86), with an accuracy comparable to ultrasound. However, unlike the latter, transcriptomic data collected at <37 weeks of gestation predicted the delivery date of one third of spontaneous (sPTB) cases within 2 weeks of the actual date. Based on samples collected before 33 weeks in asymptomatic women we found expression changes preceding preterm prelabor rupture of the membranes that were consistent across time points and cohorts, involving, among others, leukocyte-mediated immunity. Plasma proteomic random forests predicted sPTB with higher accuracy and earlier in pregnancy than whole blood transcriptomic models (e.g. AUROC=0.76 vs. AUROC=0.6 at 27-33 weeks of gestation).

bioinformatics↗

bioRxiv2020

The capacity of the ovarian cancer tumor microenvironment to integrate inflammation signaling conveys a shorter disease-free interval

Ovarian cancer has one of the highest deaths to incidence ratios across all cancers. Initial chemotherapy is typically effective, but most patients will develop chemo-resistant disease. Mechanisms driving clinical chemo-response and -resistance in ovarian cancer are not well understood. However, achieving optimal surgical cytoreduction improves survival, and cytoreduction is improved by neoadjuvant platinum/taxane-based chemotherapy (NACT). NACT offers a window to profile pre-versus post-therapy tumor specimens, which we used to identify chemotherapy-induced changes to the tumor microenvironment. We hypothesized changes in the immune microenvironment correlate with tumor chemo-response and disease progression. We obtained matched pre- and post-NACT archival tumor tissues from patients with high-grade serous ovarian cancer (patient n=6). We measured mRNA levels of 770 genes (NanoString), and performed reverse phase protein array (RPPA) on a subset of matched tumors. We examined cytokine levels in additional pre-NACT ascites samples (n=39) by multiplex ELISA. A tissue microarray with 128 annotated ovarian tumors expanded the transcriptional, RPPA, and cytokine data by multi-spectral immunohistochemistry. In NanoString analyses, transcriptional profiles segregated based on pre- and post-NACT status. The most upregulated gene post-NACT was IL6 (17.1-fold, adjusted p = 0.045). RPPA data were highly concordant with mRNA, consistent with elevated immune infiltration. Elevated IL-6 in pre-NACT ascites specimens correlated with a shorter time to recurrence. Integrating NanoString, RPPA, and cytokine studies identified an activated inflammatory signaling network and induced IL6 and IER3 (Immediate Early Response 3) post-NACT, associated with poor chemo-response and decreased time to recurrence. Taken together, multi-omic profiling of ovarian tumor samples before and after NACT provides unique insight into chemo-induced changes to the tumor and microenvironment. We integrated transcriptional, proteomic, and cytokine data and identified a novel IL-6/IER3 signaling axis through increased inflammatory signaling which may drive ovarian cancer chemo-resistance.

cancer biology↗

bioRxiv2018

Improved inference of chromosome conformation from images of labeled loci

We previously published a method that infers chromosome conformation from images of fluorescently-tagged genomic loci, for the case when there are many loci labeled with each distinguishable color. Here we build on our previous work and improve the reconstruction algorithm to address previous limitations. We show that these improvements 1) increase the reconstruction accuracy and 2) allow the method to be used on large-scale problems involving several hundred labeled loci. Simulations indicate that full-chromosome reconstructions at 1/2 Mb resolution are possible using existing labeling and imaging technologies. The updated reconstruction code and the script files used for this paper are available at: https://github.com/heltilda/align3d.

bioinformatics↗

bioRxiv2018

Trisomy 21 drives production of neurotoxic tryptophan catabolites via the interferon-inducible kynurenine pathway

Trisomy 21 (T21) causes Down syndrome (DS), affecting immune and neurological function by unknown mechanisms. We report here the results of a large metabolomics study showing that people with DS produce elevated levels of kynurenine and quinolinic acid, two tryptophan catabolites with potent immunosuppressive and neurotoxic properties, respectively. We found that immune cells of people with DS overexpress IDO1, the rate-limiting enzyme in the kynurenine pathway (KP) and a known interferon (IFN)-stimulated gene. Furthermore, we found a positive correlation between levels of specific inflammatory cytokines and KP dysregulation. Using metabolic flux assays, we found that IFN stimulation causes IDO1 overexpression and kynurenine overproduction in cells with T21, dependent on overexpression of IFN receptors encoded on chromosome 21. Finally, KP dysregulation is conserved in a mouse model of DS carrying triplication of the IFN receptors. Altogether, these results reveal a mechanism by which T21 could drive immunosuppression and neurotoxicity in DS.

physiology↗

bioRxiv2018

Low MSH2 protein levels identify muscle-invasive bladder cancer resistant to cisplatin

BackgroundThe response to first-line, platinum-based treatment of muscle-invasive bladder cancer has not improved in three decades.\n\nObjectiveThe objective of this study is to identify genes that predict cisplatin resistance in bladder cancer.\n\nDesignWe performed a whole-genome, CRISPR-based screen in a bladder cancer cell line treated with cisplatin to identify genes that mediate response to cisplatin. Targeted validation was performed in vitro across two bladder cancer cell lines. The top gene candidate was validated in a publicly available bladder cancer dataset containing 340 bladder cancer patients with treatment, protein, and survival information.\n\nResults and limitationsThe cisplatin resistance screen suggested the mismatch repair pathway through the loss of MSH2 and MLH1 contribute to cisplatin resistance. Bladder cancer cells depleted of MSH2 are resistant to cisplatin in vitro, in part due to a reduction in apoptosis. These cells maintain sensitivity to the cisplatin-analog, oxaliplatin. Bladder tumors with low protein levels of MSH2 have poorer overall survival when treated with cisplatin- or carboplatin-based therapy.\n\nConclusionsWe generated in vitro and clinical support that bladder cancer cell lines and tumors with low levels of MSH2 are more resistant to cisplatin-based therapy. Further studies are warranted to determine the ability of MSH2 protein levels to serve as a prospective biomarker of chemotherapy response in bladder cancer.\n\nPatient summaryWe report the first evidence that the protein level of MSH2 may contribute to chemotherapy resistance observed in bladder cancer. MSH2 levels has the potential to serve as a biomarker of treatment response.

cancer biology↗

bioRxiv2016

A community-based collaboration to build prediction models for short-term discontinuation of docetaxel in metastatic castration-resistant prostate cancer patients

BackgroundDocetaxel has a demonstrated survival benefit for metastatic castration-resistant prostate cancer (mCRPC). However, 10-20% of patients discontinue docetaxel prematurely because of toxicity-induced adverse events, and managing risk factors for toxicity remains an ongoing challenge for health care providers and patients. Prospective identification of high-risk patients for early discontinuation has the potential to assist clinical decision-making and can improve the design of more efficient clinical trials. In partnership with Project Data Sphere (PDS), a non-profit initiative facilitating clinical trial data-sharing, we designed an open-data, crowdsourced DREAM (Dialogue for Reverse Engineering Assessments and Methods) Challenge for developing models to predict early discontinuation of docetaxel\n\nMethodsData from the comparator arms of four phase III clinical trials in first-line mCRPC were obtained from PDS, including 476 patients treated with docetaxel and prednisone from the ASCENT2 trial, 598 patients treated with docetaxel, prednisone/prednisolone, and placebo in the VENICE trial, 526 patients treated with docetaxel, prednisone, and placebo in the MAINSAIL trial, and 528 patients treated with docetaxel and placebo in the ENTHUSE 33 trial. Early discontinuation was defined as treatment stoppage within three months due to adverse treatment effects. Over 150 clinical features including laboratory values, medical history, lesion measures, prior treatment, and demographic variables were curated and made freely available for model building for all four trials. The ASCENT2, VENICE, and MAINSAIL trial data sets formed the training set that also included patient discontinuation status. The ENTHUSE 33 trial, with patient discontinuation status hidden, was used as an independent validation set to evaluate model performance. Prediction performance was assessed using area under the precision-recall curve (AUPRC) and the Bayes factor was used to compare the performance between prediction models.\n\nResultsThe frequency of early discontinuation was similar between training (ASCENT2, VENICE, and MAINSAIL) and validation (ENTHUSE 33) sets, 12.3% versus 10.4% of docetaxel-treated patients, respectively. In total, 34 independent teams submitted predictions from 61 different models. AUPRC ranged from 0.088 to 0.178 across submissions with a random model performance of 0.104. Seven models with comparable AUPRC scores (Bayes factor [≤]; 3) were observed to outperform all other models. A post-challenge analysis of risk predictions generated by these seven models revealed three distinct patient subgroups: patients consistently predicted to be at high-risk or low-risk for early discontinuation and those with discordant risk predictions. Early discontinuation events were two-times higher in the high-versus low-risk subgroup and baseline clinical features such as presence/absence of metastatic liver lesions, and prior treatment with analgesics and ACE inhibitors exhibited statistically significant differences between the high- and low-risk subgroups (adjusted P < 0.05). An ensemble-based model constructed from a post-Challenge community collaboration resulted in the best overall prediction performance (AUPRC = 0.230) and represented a marked improvement over any individual Challenge submission. A\n\nFindingsOur results demonstrate that routinely collected clinical features can be used to prospectively inform clinicians of mCRPC patients risk to discontinue docetaxel treatment early due to adverse events and to the best of our knowledge is the first to establish performance benchmarks in this area. This work also underscores the \"wisdom of crowds\" approach by demonstrating that improved prediction of patient outcomes is obtainable by combining methods across an extended community. These findings were made possible because data from separate trials were made publicly available and centrally compiled through PDS.

bioinformatics↗