Comparison of three machine learning algorithms for classification of B-cell neoplasms using clinical flow cytometry data.

Multiparameter flow cytometry data is visually inspected by expert personnel as part of standard clinical disease diagnosis practice. This is a demanding and costly process, and recent research has demonstrated that it is possible to utilize artificial intelligence (AI) algorithms to assist in the interpretive process. Here we report our examination of three previously published machine learning methods for classification of flow cytometry data and apply these to a B-cell neoplasm dataset to obtain predicted disease subtypes. Each of the examined methods classifies samples according to specific disease categories using ungated flow cytometry data. We compare and contrast the three algorithms with respect to their architectures, and we report the multiclass classification accuracies and relative required computation times. Despite different architectures, two of the methods, flowCat and EnsembleCNN, had similarly good accuracies with relatively fast computational times. We note a speed advantage for EnsembleCNN, particularly in the case of addition of training data and retraining of the classifier.

Semi-Supervised, Attention-Based Deep Learning for Predicting TMPRSS2:ERG Fusion Status in Prostate Cancer Using Whole Slide Images.

IMPLICATIONS: Our study illuminates the potential of deep learning in effectively inferring key prostate cancer genetic alterations from the tissue morphology depicted in routinely available histology slides, offering a cost-effective method that could revolutionize diagnostic strategies in oncology.

Distinct mesenchymal cell states mediate prostate cancer progression.

In the complex tumor microenvironment (TME), mesenchymal cells are key players, yet their specific roles in prostate cancer (PCa) progression remain to be fully deciphered. This study employs single-cell RNA sequencing to delineate molecular changes in tumor stroma that influence PCa progression and metastasis. Analyzing mesenchymal cells from four genetically engineered mouse models (GEMMs) and correlating these findings with human tumors, we identify eight stromal cell populations with distinct transcriptional identities consistent across both species. Notably, stromal signatures in advanced mouse disease reflect those in human bone metastases, highlighting periostin's role in invasion and differentiation. From these insights, we derive a gene signature that predicts metastatic progression in localized disease beyond traditional Gleason scores. Our results illuminate the critical influence of stromal dynamics on PCa progression, suggesting new prognostic tools and therapeutic targets.

Divergent immune microenvironments in two tumor nodules from a patient with mismatch repair-deficient prostate cancer.

Patients with prostate cancer (PC) generally do not respond favorably to immune checkpoint inhibitors, which may be due to a low abundance of tumor-infiltrating lymphocytes even when mutational load is high. Here, we identified a patient who presented with high-grade primary prostate cancer with two adjacent tumor nodules. While both nodules were mismatch repair-deficient (MMRd), exhibited pathogenic MSH2 and MSH6 alterations, had a high tumor mutational burden (TMB), and demonstrated high microsatellite instability (MSI), they had markedly distinct immune phenotypes. The first displayed a dense infiltrate of lymphocytes ("hot nodule"), while the second displayed significantly fewer infiltrating lymphocytes ("cold nodule"). Whole-exome DNA analysis found that both nodules shared many identical mutations, indicating that they were derived from a single clone. However, the cold nodule appeared to be sub-clonal relative to the hot nodule, suggesting divergent evolution of the cold nodule from the hot nodule. Whole-transcriptome RNA analysis found that the cold nodule demonstrated lower expression of genes related to antigen presentation (HLA) and, paradoxically, classical tumor immune tolerance markers such as PD-L1 (CD274) and CTLA-4. Immune cell deconvolution suggested that the hot nodule was enriched not only in CD8+ and CD4 + T lymphocytes, but also in M1 macrophages, activated NK cells, and γδ T cells compared to the cold nodule. This case highlights that MMRd/TMB-high PC can evolve to minimize an anti-tumor immune response, and nominates downregulation of antigen presentation machinery (HLA loss) as a potential mechanism of adaptive immune evasion in PC.

Concomitant assessment of PD-1 and CD56 expression identifies subsets of resting cord blood Vδ2 T cells with disparate cytotoxic potential.

Vγ9Vδ2 T lymphocytes are programmed for broad antimicrobial responses with rapid production of Th1 cytokines even before birth, and thus thought to play key roles against pathogens in infants. The process regulating Vδ2 cell acquisition of cytotoxic potential shortly after birth remains understudied. We observed that perforin production in cord blood Vδ2 cells correlates with phenotypes defined by the concomitant assessment of PD-1 and CD56. Bulk RNA sequencing of sorted Vδ2 cell fractions indicated that transcripts related to cytotoxic activity and NK function are enriched in the subset with the highest proportion of perforin+ cells. Among differentially expressed transcripts, IRF8, previously linked to CD8 T cell effector differentiation and NK maturation, has the potential to mediate Vδ2 cell differentiation towards cytotoxic effectors. Our current and past results support the hypothesis that distinct mechanisms regulate Vδ2 cell cytotoxic function before and after birth, possibly linked to different levels of microbial exposure.

Notch-based gene signature for predicting the response to neoadjuvant chemotherapy in triple-negative breast cancer.

BACKGROUND: While the efficacy of neoadjuvant chemotherapy (NACT) in treating triple-negative breast cancer (TNBC) is generally accepted, not all patients derive benefit from this preoperative treatment. Presently, there are no validated biomarkers to predict the NACT response, and previous attempts to develop predictive classifiers based on gene expression data have not demonstrated clinical utility. However, predictive models incorporating biological constraints have shown increased robustness and improved performance compared to agnostic classifiers. METHODS: We used the preoperative transcriptomic profiles from 298 patients with TNBC to train and test a rank-based classifier, k-top scoring pairs, to predict whether the patient will have pathological complete response (pCR) or residual disease (RD) following NACT. To reduce overfitting and enhance the signature's interpretability, we constrained the training process to genes involved in the Notch signaling pathway. Subsequently, we evaluated the signature performance on two independent cohorts with 75 and 71 patients. Finally, we assessed the prognostic value of the signature by examining its association with relapse-free survival (RFS) using Kaplan‒Meier (KM) survival estimates and a multivariate Cox proportional hazards model. RESULTS: The final signature consists of five gene pairs, whose relative ordering can be predictive of the NACT response. The signature has a robust performance at predicting pCR in TNBC patients with an area under the ROC curve (AUC) of 0.76 and 0.85 in the first and second testing cohorts, respectively, outperforming other gene signatures developed for the same purpose. Additionally, the signature was significantly associated with RFS in an independent TNBC patient cohort even after adjusting for T stage, patient age at the time of diagnosis, type of breast surgery, and menopausal status. CONCLUSION: We introduce a robust gene signature to predict pathological complete response (pCR) in patients with TNBC. This signature applies easily interpretable, rank-based decision rules to genes regulated by the Notch signaling pathway, a known determinant in breast cancer chemoresistance. The robust predictive and prognostic performance of the signature make it a strong candidate for clinical implementation, aiding in the stratification of TNBC patients undergoing NACT.

PhenoSV: interpretable phenotype-aware model for the prioritization of genes affected by structural variants.

Structural variants (SVs) represent a major source of genetic variation associated with phenotypic diversity and disease susceptibility. While long-read sequencing can discover over 20,000 SVs per human genome, interpreting their functional consequences remains challenging. Existing methods for identifying disease-related SVs focus on deletion/duplication only and cannot prioritize individual genes affected by SVs, especially for noncoding SVs. Here, we introduce PhenoSV, a phenotype-aware machine-learning model that interprets all major types of SVs and genes affected. PhenoSV segments and annotates SVs with diverse genomic features and employs a transformer-based architecture to predict their impacts under a multiple-instance learning framework. With phenotype information, PhenoSV further utilizes gene-phenotype associations to prioritize phenotype-related SVs. Evaluation on extensive human SV datasets covering all SV types demonstrates PhenoSV's superior performance over competing methods. Applications in diseases suggest that PhenoSV can determine disease-related genes from SVs. A web server and a command-line tool for PhenoSV are available at https://phenosv.wglab.org .

Tumor-Specific CD8+ T Cells from the Bone Marrow Resist Exhaustion and Exhibit Increased Persistence in Tumor-Bearing Hosts as Compared to Tumor Infiltrating Lymphocytes.

Immunotherapy is now an integral aspect of cancer therapy. Strategies employing adoptive cell therapy (ACT) have seen the establishment of chimeric antigen receptor (CAR)-T cells using peripheral blood lymphocytes as well as tumor infiltrating lymphocytes (TILs) with significant clinical results. Despite these successes, the limitations of the current strategies are also emerging and novel approaches are needed. The bone marrow (BM) is an immunological niche that houses T cells with specificity for previously encountered antigens, including tumor-associated antigens from certain solid cancers. This study sought to improve our understanding of tumor-specific BM T cells in the context of solid tumors by comparing them with TILs, and to assess whether there is a rationale for using the BM as a source of T cells for ACT against solid malignancies. Herein, we demonstrate that T cells from the BM appear superior to TILs as a source of cells for cellular therapy. Specifically, they possess a memory-enriched phenotype and exhibit improved effector function, greater persistence within a tumor-bearing host, and the capacity for increased tumor infiltration. Taken together, these data provide a foundation for further exploring the BM as a source of tumor-specific T cells for ACT in solid malignancies.

CALANGO: A phylogeny-aware comparative genomics tool for discovering quantitative genotype-phenotype associations across species.

Living species vary significantly in phenotype and genomic content. Sophisticated statistical methods linking genes with phenotypes within a species have led to breakthroughs in complex genetic diseases and genetic breeding. Despite the abundance of genomic and phenotypic data available for thousands of species, finding genotype-phenotype associations across species is challenging due to the non-independence of species data resulting from common ancestry. To address this, we present CALANGO (comparative analysis with annotation-based genomic components), a phylogeny-aware comparative genomics tool to find homologous regions and biological roles associated with quantitative phenotypes across species. In two case studies, CALANGO identified both known and previously unidentified genotype-phenotype associations. The first study revealed unknown aspects of the ecological interaction between Escherichia coli, its integrated bacteriophages, and the pathogenicity phenotype. The second identified an association between maximum height in angiosperms and the expansion of a reproductive mechanism that prevents inbreeding and increases genetic diversity, with implications for conservation biology and agriculture.

Optimal fluid volume for detecting malignancy in serous effusions: a single institution experience.

INTRODUCTION: Detection of malignant cells in serous fluids is an indicator of advanced stage of malignancy and is critical in clinical management decisions and prompt treatment initiation. The minimum volume which is ideal for detecting malignancy in serous fluid is not well established. In this study, we aim to identify optimal volume that will be ideal for adequate cytopathological diagnosis. MATERIALS AND METHODS: A total of 1597 samples of serous fluids from 1134 patients were included in the study. Samples were diagnosed based on International System for Reporting Serous Fluid Cytopathology (ISRSFC). Clinicopathologic results from different diagnostic groups were compared and statistically analyzed. RESULTS: Pleural fluids comprised 890 (55.7%) specimens, followed by 456 (28.6%) peritoneal, 128 (8%) ascites, and 123 (7.7%) pericardial fluid specimens. The majority were negative for malignancy (1138, 71.3%), followed by malignant (376, 23.5%), atypical (59, 3.7%), and suspicious for malignancy (24, 1.5%). Malignancy was identified in sample with volumes from 5 mL to 5000 mL. Rate of detection of malignant cells increased significantly with higher sample volumes. For malignancy detection the optimal volume for overall serous fluid is 70 mL. Pericardial fluid is an exception, with lower mean volume and significantly lower proportion of cases with malignant diagnosis. CONCLUSIONS: Our study indicates that higher fluid volumes have a higher rate of malignancy detection and a low false-negative rate. We recommend a minimum of 70 mL of serous fluid for optimal cytopathologic examination and malignancy detection. Pericardial fluid is an exception, with lower mean volume and thus lower requirement.

MultiNEP: a multi-omics network enhancement framework for prioritizing disease genes and metabolites simultaneously.

MOTIVATION: Many studies have successfully used network information to prioritize candidate omics profiles associated with diseases. The metabolome, as the link between genotypes and phenotypes, has accumulated growing attention. Using a "multi-omics" network constructed with a gene-gene network, a metabolite-metabolite network, and a gene-metabolite network to simultaneously prioritize candidate disease-associated metabolites and gene expressions could further utilize gene-metabolite interactions that are not used when prioritizing them separately. However, the number of metabolites is usually 100 times fewer than that of genes. Without accounting for this imbalance issue, we cannot effectively use gene-metabolite interactions when simultaneously prioritizing disease-associated metabolites and genes. RESULTS: Here, we developed a Multi-omics Network Enhancement Prioritization (MultiNEP) framework with a weighting scheme to reweight contributions of different sub-networks in a multi-omics network to effectively prioritize candidate disease-associated metabolites and genes simultaneously. In simulation studies, MultiNEP outperforms competing methods that do not address network imbalances and identifies more true signal genes and metabolites simultaneously when we down-weight relative contributions of the gene-gene network and up-weight that of the metabolite-metabolite network to the gene-metabolite network. Applications to two human cancer cohorts show that MultiNEP prioritizes more cancer-related genes by effectively using both within- and between-omics interactions after handling network imbalance. AVAILABILITY AND IMPLEMENTATION: The developed MultiNEP framework is implemented in an R package and available at: https://github.com/Karenxzr/MultiNep.

Distinct mesenchymal cell states mediate prostate cancer progression.

Alterations in tumor stroma influence prostate cancer progression and metastatic potential. However, the molecular underpinnings of this stromal-epithelial crosstalk are largely unknown. Here, we compare mesenchymal cells from four genetically engineered mouse models (GEMMs) of prostate cancer representing different stages of the disease to their wild-type (WT) counterparts by single-cell RNA sequencing (scRNA-seq) and, ultimately, to human tumors with comparable genotypes. We identified 8 transcriptionally and functionally distinct stromal populations responsible for common and GEMM-specific transcriptional programs. We show that stromal responses are conserved in mouse models and human prostate cancers with the same genomic alterations. We noted striking similarities between the transcriptional profiles of the stroma of murine models of advanced disease and those of of human prostate cancer bone metastases. These profiles were then used to build a robust gene signature that can predict metastatic progression in prostate cancer patients with localized disease and is also associated with progression-free survival independent of Gleason score. Taken together, this offers new evidence that stromal microenvironment mediates prostate cancer progression, further identifying tissue-based biomarkers and potential therapeutic targets of aggressive and metastatic disease.

REPAC: analysis of alternative polyadenylation from RNA-sequencing data.

Alternative polyadenylation (APA) is an important post-transcriptional mechanism that has major implications in biological processes and diseases. Although specialized sequencing methods for polyadenylation exist, availability of these data are limited compared to RNA-sequencing data. We developed REPAC, a framework for the analysis of APA from RNA-sequencing data. Using REPAC, we investigate the landscape of APA caused by activation of B cells. We also show that REPAC is faster than alternative methods by at least 7-fold and that it scales well to hundreds of samples. Overall, the REPAC method offers an accurate, easy, and convenient solution for the exploration of APA.

Using biological constraints to improve prediction in precision oncology.

Many gene signatures have been developed by applying machine learning (ML) on omics profiles, however, their clinical utility is often hindered by limited interpretability and unstable performance. Here, we show the importance of embedding prior biological knowledge in the decision rules yielded by ML approaches to build robust classifiers. We tested this by applying different ML algorithms on gene expression data to predict three difficult cancer phenotypes: bladder cancer progression to muscle-invasive disease, response to neoadjuvant chemotherapy in triple-negative breast cancer, and prostate cancer metastatic progression. We developed two sets of classifiers: mechanistic, by restricting the training to features capturing specific biological mechanisms; and agnostic, in which the training did not use any a priori biological information. Mechanistic models had a similar or better testing performance than their agnostic counterparts, with enhanced interpretability. Our findings support the use of biological constraints to develop robust gene signatures with high translational potential.

Association of self-identified race and genetic ancestry with the immunogenomic landscape of primary prostate cancer.

The genomic and immune landscapes of prostate cancer differ by self-identified race. However, few studies have examined the genome-wide copy number landscape and immune content of matched cohorts with genetic ancestry data and clinical outcomes. Here, we assessed prostate cancer somatic copy number alterations (sCNA) and tumor immune content of a grade-matched, surgically treated cohort of 145 self-identified Black (BL) and 145 self-identified White (WH) patients with genetic ancestry estimation. A generalized linear model adjusted with age, preoperative prostate-specific antigen (PSA), and Gleason Grade Group and filtered for germline copy number variations (gCNV) identified 143 loci where copy number varied significantly by percent African ancestry, clustering on chromosomes 6p, 10q, 11p, 12p, and 17p. Multivariable Cox regression models adjusted for age, preoperative PSA levels, and Gleason Grade Group revealed that chromosome 8q gains (including MYC) were significantly associated with biochemical recurrence and metastasis, independent of genetic ancestry. Finally, Treg density in BL and WH patients was significantly correlated with percent genome altered, and these findings were validated in the TCGA cohort. Taken together, our findings identify specific sCNA linked to genetic ancestry and outcome in primary prostate cancer and demonstrate that Treg infiltration varies by global sCNA burden in primary disease.

A Phase 1 Trial of Durvalumab in Combination with Bacillus Calmette-Guerin (BCG) or External Beam Radiation Therapy in Patients with BCG-unresponsive Non-muscle-Invasive Bladder Cancer: The Hoosier Cancer Research Network GU16-243 ADAPT-BLADDER Study.

BACKGROUND: Novel treatments and trial designs remain a high priority for bacillus Calmette-Guerin (BCG)-unresponsive non-muscle-invasive bladder cancer (NMIBC) patients. OBJECTIVE: To evaluate the safety and preliminary efficacy of anti-PD-L1 directed therapy with durvalumab (D), durvalumab plus BCG (D + BCG), and durvalumab plus external beam radiation therapy (D + EBRT). DESIGN, SETTING, AND PARTICIPANTS: A multicenter phase 1 trial was conducted at community and academic sites. INTERVENTION: Patients received 1120 mg of D intravenously every 3 wk for eight cycles. D + BCG patients also received full-dose intravesical BCG weekly for 6 wk with BCG maintenance recommended. D + EBRT patients received concurrent EBRT (6 Gy × 3 in cycle 1 only). OUTCOME MEASUREMENTS AND STATISTICAL ANALYSIS: Post-treatment cystoscopy and urine cytology were performed at 3 and 6 -mo, with bladder biopsies required at the 6-mo evaluation. The recommended phase 2 dose (RP2D) for each regimen was the primary endpoint. Secondary endpoints included toxicity profiles and complete response (CR) rates. RESULTS AND LIMITATIONS: Twenty-eight patients were treated in the D (n = 3), D + BCG (n = 13), and D + EBRT (n = 12) cohorts. Full-dose D, full-dose BCG, and 6 Gy fractions × 3 were determined as the RP2Ds. One patient (4%) experienced a grade 3 dose limiting toxicity event of autoimmune hepatitis. The 3-mo CR occurred in 64% of all patients and in 33%, 85%, and 50% within the D, D + BCG, and D + EBRT cohorts, respectively. Twelve-month CRs were achieved in 46% of all patients and in 73% of D + BCG and 33% of D + EBRT patients. CONCLUSIONS: D combined with intravesical BCG or EBRT proved feasible and safe in BCG-unresponsive NMIBC patients. Encouraging preliminary efficacy justifies further study of combination therapy approaches. PATIENT SUMMARY: Durvalumab combination therapy can be safely administered to non-muscle-invasive bladder cancer patients with the goal of increasing durable response rates.

Hyaluronan driven by epithelial aPKC deficiency remodels the microenvironment and creates a vulnerability in mesenchymal colorectal cancer.

Mesenchymal colorectal cancer (mCRC) is microsatellite stable (MSS), highly desmoplastic, with CD8+ T cells excluded to the stromal periphery, resistant to immunotherapy, and driven by low levels of the atypical protein kinase Cs (aPKCs) in the intestinal epithelium. We show here that a salient feature of these tumors is the accumulation of hyaluronan (HA) which, along with reduced aPKC levels, predicts poor survival. HA promotes epithelial heterogeneity and the emergence of a tumor fetal metaplastic cell (TFMC) population endowed with invasive cancer features through a network of interactions with activated fibroblasts. TFMCs are sensitive to HA deposition, and their metaplastic markers have prognostic value. We demonstrate that in vivo HA degradation with a clinical dose of hyaluronidase impairs mCRC tumorigenesis and liver metastasis and enables immune checkpoint blockade therapy by promoting the recruitment of B and CD8+ T cells, including a proportion with resident memory features, and by blocking immunosuppression.

Recurrent germline variant in ATM associated with familial myeloproliferative neoplasms.

Genetic predisposition (familial risk) in the myeloproliferative neoplasms (MPNs) is more common than the risk observed in most other cancers, including breast, prostate, and colon. Up to 10% of MPNs are considered to be familial. Recent genome-wide association studies have identified genomic loci associated with an MPN diagnosis. However, the identification of variants with functional contributions to the development of MPN remains limited. In this study, we have included 630 MPN patients and whole genome sequencing was performed in 64 individuals with familial MPN to uncover recurrent germline predisposition variants. Both targeted and unbiased filtering of single nucleotide variants (SNVs) was performed, with a comparison to 218 individuals with MPN unselected for familial status. This approach identified an ATM L2307F SNV occurring in nearly 8% of individuals with familial MPN. Structural protein modeling of this variant suggested stabilization of inactive ATM dimer, and alteration of the endogenous ATM locus in a human myeloid cell line resulted in decreased phosphorylation of the downstream tumor suppressor CHEK2. These results implicate ATM, and the DNA-damage response pathway, in predisposition to MPN.

A FOXO1-dependent transcription network is a targetable vulnerability of mantle cell lymphomas.

Targeting lineage-defined transcriptional dependencies has emerged as an effective therapeutic strategy in cancer treatment. Through screening for molecular vulnerabilities of mantle cell lymphoma (MCL), we identified a set of transcription factors (TFs) including FOXO1, EBF1, PAX5, and IRF4 that are essential for MCL propagation. Integrated chromatin immunoprecipitation and sequencing (ChIP-Seq) with transcriptional network reconstruction analysis revealed FOXO1 as a master regulator that acts upstream in the regulatory TF hierarchy. FOXO1 is both necessary and sufficient to drive MCL lineage commitment through supporting the lineage-specific transcription programs. We further show that FOXO1, but not its close paralog FOXO3, can reprogram myeloid leukemia cells and induce B-lineage gene expression. Finally, we demonstrate that cpd10, a small molecule identified from an enriched FOXO1 inhibitor library, induces a robust cytotoxic response in MCL cells in vitro and suppresses MCL progression in vivo. Our findings establish FOXO1 inhibition as a therapeutic strategy targeting lineage-driven transcriptional addiction in MCL.

Rapid detection of SARS-CoV-2 variants of concern by single nucleotide polymorphism genotyping using TaqMan assays.

We evaluated the performance of SARS-CoV-2 TaqMan real-time reverse-transcription PCR (RT-qPCR) assays (ThermoFisher) for detecting 2 nonsynonymous spike protein mutations, E484K and N501Y. Assay accuracy was evaluated by whole genome sequencing (WGS). Residual nasopharyngeal SARS-CoV-2 positive samples (N = 510) from a diverse patient population in New York City submitted for routine SARS-CoV-2 testing during January-April 2020 were used. We detected 91 (18%) N501Y and 101 (20%) E484K variants. Four samples (0.8%) were positive for both variants. The assay had nearly perfect concordance with WGS in the validation subset, detecting B.1.1.7 and B.1.526 variants among others. Sensitivity and specificity ranged from 0.95 to 1.00. Positive and negative predictive values were 0.98-1.00. TaqMan genotyping successfully predicted the presence of B.1.1.7, but had significantly lower sensitivity, 62% (95% CI, 0.53, 0.71), for predicting B.1.526 sub-lineages lacking E484K. This approach is rapid and accurate for detecting SARS-CoV-2 variants and can be rapidly implemented in routine clinical setting.