Modeling the Simultaneous Dynamics of Proteins in Blood Plasma and the Cerebrospinal Fluid in Human In Vivo.

The analysis of protein dynamics or turnover in patients has the potential to reveal altered protein recycling, such as in Alzheimer's disease, and to provide informative data regarding drug efficacy or certain biological processes. The observed protein dynamics in a solid tissue or a fluid is the net result of not only protein synthesis and degradation but also transport across biological compartments. We report an accurate 3-biological compartment model able to simultaneously account for the protein dynamics observed in blood plasma and the cerebrospinal fluid (CSF) including a hidden central nervous system (CNS) compartment. We successfully applied this model to 69 proteins of a single individual displaying similar or very different dynamics in plasma and CSF. This study puts a strong emphasis on the methods and tools needed to develop this type of model. We believe that it will be useful to any researcher dealing with protein dynamics data modeling.

Farnesyltransferase inhibition overcomes oncogene-addicted non-small cell lung cancer adaptive resistance to targeted therapies.

Drug-tolerance has emerged as one of the major non-genetic adaptive processes driving resistance to targeted therapy (TT) in non-small cell lung cancer (NSCLC). However, the kinetics and sequence of molecular events governing this adaptive response remain poorly understood. Here, we combine real-time monitoring of the cell-cycle dynamics and single-cell RNA sequencing in a broad panel of oncogenic addiction such as EGFR-, ALK-, BRAF- and KRAS-mutant NSCLC, treated with their corresponding TT. We identify a common path of drug adaptation, which invariably involves alveolar type 1 (AT1) differentiation and Rho-associated protein kinase (ROCK)-mediated cytoskeletal remodeling. We also isolate and characterize a rare population of early escapers, which represent the earliest resistance-initiating cells that emerge in the first hours of treatment from the AT1-like population. A phenotypic drug screen identify farnesyltransferase inhibitors (FTI) such as tipifarnib as the most effective drugs in preventing relapse to TT in vitro and in vivo in several models of oncogenic addiction, which is confirmed by genetic depletion of the farnesyltransferase. These findings pave the way for the development of treatments combining TT and FTI to effectively prevent tumor relapse in oncogene-addicted NSCLC patients.

Prediction of response to promising first-line chemotherapy in ovarian cancer patients with residual peritoneal tumors: practical biomarkers and robust multiplex models.

BACKGROUND: Platinum/taxane (TC) chemotherapy with debulking surgery stays the mainstay of the treatment in ovarian cancer patients with peritoneal metastasis, and recently its novel modality, intraperitoneal carboplatin with dose-dense paclitaxel (ddTCip), was shown to have greater therapeutic impact. Nevertheless, the response varies among patients and consequent recurrence, or relapse often occurs. Discovery of therapeutic response predictor to ddTCip and/or TC therapy is eagerly awaited to improve the treatment outcome. METHODS: Using datasets in 76 participants in our ddTCip study and published databases on patients received TC therapy, we first validated a total of 75 previously suggested markers, sought out more active biomarkers through the association analyses of genome-wide transcriptome and genotyping data with progression-free survival (PFS) and adverse events, and then developed multiplex statistical prediction models for PFS and toxicity by mainly using multiple regression analysis and the classification and regression tree (CART) algorithm. RESULTS: The association analyses revealed that SPINK1 could be a possible biomarker of ddTCip efficacy, while ABCB1 rs1045642 and ERCC1 rs11615 would be a predictor of hematologic toxicity and peripheral neuropathy, respectively. Multiple regression analyses and CART algorithm finally provided a potent efficacy prediction model using 5 gene expression data and robust multiplex toxicity prediction models-CART models using a total of 4 genotype combinations and multiple regression models using 15 polymorphisms on 12 genes. CONCLUSION: Biomarkers and multiplex models composed here could work well in the response prediction of ddTCip and/or TC therapy, which might contribute to realize optimal selection of the key therapy.

Cancer-associated fibroblast spatial heterogeneity and EMILIN1 expression in the tumor microenvironment modulate TGF-ß activity and CD8+ T-cell infiltration in breast cancer.

Rationale: The tumor microenvironment (TME) and its multifaceted interactions with cancer cells are major targets for cancer treatment. Single-cell technologies have brought major insights into the TME, but the resulting complexity often precludes conclusions on function. Methods: We combined single-cell RNA sequencing and spatial transcriptomic data to explore the relationship between different cancer-associated fibroblast (CAF) populations and immune cell exclusion in breast tumors. The significance of the findings was then evaluated in a cohort of tumors (N=75) from breast cancer patients using immunohistochemistry analysis. Results: Our data show for the first time the degree of spatial organization of different CAF populations in breast cancer. We found that IL-iCAFs, Detox-iCAFs, and IFNγ-iCAFs tended to cluster together, while Wound-myCAFs, TGFß-myCAFs, and ECM-myCAFs formed another group that overlapped with elevated TGF-ß signaling. Differential gene expression analysis of areas with CD8+ T-cell infiltration/exclusion within the TGF-ß signaling-rich zones identified elastin microfibrillar interface protein 1 (EMILIN1) as a top modulated gene. EMILIN1, a TGF-ß inhibitor, was upregulated in IFNγ-iCAFs directly modulating TGFß immunosuppressive function. Histological analysis of 75 breast cancer samples confirmed that high EMILIN1 expression in the tumor margins was related to high CD8+ T-cell infiltration, consistent with our spatial gene expression analysis. High EMILIN1 expression was also associated with better prognosis of patients with breast cancer, underscoring its functional significance for the recruitment of cytotoxic T cells into the tumor area. Conclusion: Our data show that correlating TGF-ß signaling to a CAF subpopulation is not enough because proteins with TGF-ß-modulating activity originating from other CAF subpopulations can alter its activity. Therefore, therapeutic targeting should remain focused on biological processes rather than on specific CAF subtypes.

Circulating DNA reveals a specific and higher fragmentation of the Y chromosome.

Chromosome stability is a key point in genome evolution, particularly that of the Y chromosome. The Y chromosome loss in blood and tumor cells is well established. Through processes that are common to other chromosomes too, the Y chromosome undergoes degradation and fragmentation in the blood stream before elimination. This process gives rise to circulating DNA (cirDNA) fragments, whose examination may provide potential insight into the role of DNA fragmentation in blood for the Y chromosome elimination. In this study, we employed shallow whole genome sequencing (sWGS) to comprehensively assess the total cirDNA and the individual chromosome fragment size profiles in the plasma of healthy male individuals. Here, we show that (i) the fragment size profiles of total circulating DNA (cirDNA) and DNA fragments originating from autosomes and the X chromosome in blood plasma are homogeneous, and have a remarkably low variability (mean CV = 7%) among healthy individuals, (ii) the Y chromosome has a distinct fragment size profile with the accumulation of the fragment < 145 bp and depletion of the dinucleosome-associated fragments (290-390 bp), and its fragment fraction in blood decreases with age. These results indicate a higher fragmentation of the Y chromosome compared to other chromosomes and this in turn might be due to its increased susceptibility to degradation. Our findings pave the way for an elucidation of the impact of chromosomal origin on DNA degradation and the Y chromosome biology.

Design and selection of optimal ErbB-targeting bispecific antibodies in pancreatic cancer.

The ErbB family of receptor tyrosine kinases is a primary target for small molecules and antibodies for pancreatic cancer treatment. Nonetheless, the current treatments for this tumor are not optimal due to lack of efficacy, resistance, or toxicity. Here, using the novel BiXAb™ tetravalent format platform, we generated bispecific antibodies against EGFR, HER2, or HER3 by considering rational epitope combinations. We then screened these bispecific antibodies and compared them with the parental single antibodies and antibody pair combinations. The screen readouts included measuring binding to the cognate receptors (mono and bispecificity), intracellular phosphorylation signaling, cell proliferation, apoptosis and receptor expression, and also immune system engagement assays (antibody-dependent cell-mediated cytotoxicity and complement-dependent cytotoxicity). Among the 30 BiXAbs™ tested, we selected 3Patri-1Cetu-Fc, 3Patri-1Matu-Fc and 3Patri-2Trastu-Fc as lead candidates. The in vivo testing of these three highly efficient bispecific antibodies against EGFR and HER2 or HER3 in pre-clinical mouse models of pancreatic cancer showed deep antibody penetration in these dense tumors and robust tumor growth reduction. Application of such semi-rational/semi-empirical approach, which includes various immunological assays to compare pre-selected antibodies and their combinations with bispecific antibodies, represents the first attempt to identify potent bispecific antibodies against ErbB family members in pancreatic cancer.

Inferring ligand-receptor cellular networks from bulk and spatial transcriptomic datasets with BulkSignalR.

The study of cellular networks mediated by ligand-receptor interactions has attracted much attention recently owing to single-cell omics. However, rich collections of bulk data accompanied with clinical information exists and continue to be generated with no equivalent in single-cell so far. In parallel, spatial transcriptomic (ST) analyses represent a revolutionary tool in biology. A large number of ST projects rely on multicellular resolution, for instance the Visium™ platform, where several cells are analyzed at each location, thus producing localized bulk data. Here, we describe BulkSignalR, a R package to infer ligand-receptor networks from bulk data. BulkSignalR integrates ligand-receptor interactions with downstream pathways to estimate statistical significance. A range of visualization methods complement the statistics, including functions dedicated to spatial data. We demonstrate BulkSignalR relevance using different datasets, including new Visium liver metastasis ST data, with experimental validation of protein colocalization. A comparison with other ST packages shows the significantly higher quality of BulkSignalR inferences. BulkSignalR can be applied to any species thanks to its built-in generic ortholog mapping functionality.

Modular response analysis reformulated as a multilinear regression problem.

MOTIVATION: Modular response analysis (MRA) is a well-established method to infer biological networks from perturbation data. Classically, MRA requires the solution of a linear system, and results are sensitive to noise in the data and perturbation intensities. Due to noise propagation, applications to networks of 10 nodes or more are difficult. RESULTS: We propose a new formulation of MRA as a multilinear regression problem. This enables to integrate all the replicates and potential additional perturbations in a larger, over-determined, and more stable system of equations. More relevant confidence intervals on network parameters can be obtained, and we show competitive performance for networks of size up to 1000. Prior knowledge integration in the form of known null edges further improves these results. AVAILABILITY AND IMPLEMENTATION: The R code used to obtain the presented results is available from GitHub: https://github.com/J-P-Borg/BioInformatics.

FT-GPI, a highly sensitive and accurate predictor of GPI-anchored proteins, reveals the composition and evolution of the GPI proteome in Plasmodium species.

BACKGROUND: Protozoan parasites are known to attach specific and diverse group of proteins to their plasma membrane via a GPI anchor. In malaria parasites, GPI-anchored proteins (GPI-APs) have been shown to play an important role in host-pathogen interactions and a key function in host cell invasion and immune evasion. Because of their immunogenic properties, some of these proteins have been considered as malaria vaccine candidates. However, identification of all possible GPI-APs encoded by these parasites remains challenging due to their sequence diversity and limitations of the tools used for their characterization. METHODS: The FT-GPI software was developed to detect GPI-APs based on the presence of a hydrophobic helix at both ends of the premature peptide. FT-GPI was implemented in C ++and applied to study the GPI-proteome of 46 isolates of the order Haemosporida. Using the GPI proteome of Plasmodium falciparum strain 3D7 and Plasmodium vivax strain Sal-1, a heuristic method was defined to select the most sensitive and specific FT-GPI software parameters. RESULTS: FT-GPI enabled revision of the GPI-proteome of P. falciparum and P. vivax, including the identification of novel GPI-APs. Orthology- and synteny-based analyses showed that 19 of the 37 GPI-APs found in the order Haemosporida are conserved among Plasmodium species. Our analyses suggest that gene duplication and deletion events may have contributed significantly to the evolution of the GPI proteome, and its composition correlates with speciation. CONCLUSION: FT-GPI-based prediction is a useful tool for mining GPI-APs and gaining further insights into their evolution and sequence diversity. This resource may also help identify new protein candidates for the development of vaccines for malaria and other parasitic diseases.

The landscape of cancer-associated fibroblasts in colorectal cancer liver metastases.

Rationale: Patients with colorectal cancer die mainly due to liver metastases (CRC-LM). Although the tumor microenvironment (TME) plays an important role in tumor development and therapeutic response, our understanding of the individual TME components, especially cancer-associated fibroblasts (CAFs), remains limited. Methods: We analyzed CRC-LM CAFs and cancer cells by single-cell transcriptomics and used bioinformatics for data analysis and integration with related available single-cell and bulk transcriptomic datasets. We validated key findings by RT-qPCR, western blotting, and immunofluorescence. Results: By single-cell transcriptomic analysis of 4,397 CAFs from six CRC-LM samples, we identified two main CAF populations, contractile CAFs and extracellular matrix (ECM)-remodeling/pro-angiogenic CAFs, and four subpopulations with distinct phenotypes. We found that ECM-remodeling/pro-angiogenic CAFs derive from portal resident fibroblasts. They associate with areas of strong desmoplastic reaction and Wnt signaling in low-proliferating tumor cells engulfed in a stiff extracellular matrix. By integrating public single-cell primary liver tumor data, we propose a model to explain how different liver malignancies recruit CAFs of different origins to this organ. Lastly, we found that LTBP2 plays an important role in modulating collagen biosynthesis, ECM organization, and adhesion pathways. We developed fully human antibodies against LTBP2 that depleted LTBP2+ CAFs in vitro. Conclusion: This study complements recent reports on CRC-LM CAF heterogeneity at the single-cell resolution. The number of sequenced CAFs was more than one order of magnitude larger compared to existing data. LTBP2 targeting by antibodies might create opportunities to deplete ECM-remodeling CAFs in CRC-LMs. This might be combined with other therapies, e.g., anti-angiogenic compounds as already done in CRC. Moreover, we showed that in intrahepatic cholangiocarcinoma, in which ECM-remodeling CAF proportion is similar to that of CRC-LM, several genes expressed by ECM-remodeling CAFs, such as LTBP2, were associated with survival.

Mechanical signatures of human colon cancers.

Besides the standard parameters used for colorectal cancer (CRC) management, new features are needed in clinical practice to improve progression-free and overall survival. In some cancers, the microenvironment mechanical properties can contribute to cancer progression and metastasis formation, or constitute a physical barrier for drug penetration or immune cell infiltration. These mechanical properties remain poorly known for colon tissues. Using a multidisciplinary approach including clinical data, physics and geostatistics, we characterized the stiffness of healthy and malignant colon specimens. For this purpose, we analyzed a prospective cohort of 18 patients with untreated colon adenocarcinoma using atomic force microscopy to generate micrometer-scale mechanical maps. We characterized the stiffness of normal epithelium samples taken far away or close to the tumor area and selected tumor tissue areas. These data showed that normal epithelium was softer than tumors. In tumors, stroma areas were stiffer than malignant epithelial cell areas. Among the clinical parameters, tumor left location, higher stage, and RAS mutations were associated with increased tissue stiffness. Thus, in patients with CRC, measuring tumor tissue rigidity may have a translational value and an impact on patient care.

Application of modular response analysis to medium- to large-size biological systems.

The development of high-throughput genomic technologies associated with recent genetic perturbation techniques such as short hairpin RNA (shRNA), gene trapping, or gene editing (CRISPR/Cas9) has made it possible to obtain large perturbation data sets. These data sets are invaluable sources of information regarding the function of genes, and they offer unique opportunities to reverse engineer gene regulatory networks in specific cell types. Modular response analysis (MRA) is a well-accepted mathematical modeling method that is precisely aimed at such network inference tasks, but its use has been limited to rather small biological systems so far. In this study, we show that MRA can be employed on large systems with almost 1,000 network components. In particular, we show that MRA performance surpasses general-purpose mutual information-based algorithms. Part of these competitive results was obtained by the application of a novel heuristic that pruned MRA-inferred interactions a posteriori. We also exploited a block structure in MRA linear algebra to parallelize large system resolutions.

Mechanisms underlying the cooperation between loss of epithelial polarity and Notch signaling during neoplastic growth in Drosophila.

Aggressive neoplastic growth can be initiated by a limited number of genetic alterations, such as the well-established cooperation between loss of cell architecture and hyperactive signaling pathways. However, our understanding of how these different alterations interact and influence each other remains very incomplete. Using Drosophila paradigms of imaginal wing disc epithelial growth, we have monitored the changes in Notch pathway activity according to the polarity status of cells (scrib mutant). We show that the scrib mutation impacts the direct transcriptional output of the Notch pathway, without altering the global distribution of Su(H), the Notch-dedicated transcription factor. The Notch-dependent neoplasms require, however, the action of a group of transcription factors, similar to those previously identified for Ras/scrib neoplasm (namely AP-1, Stat92E, Ftz-F1 and basic leucine zipper factors), further suggesting the importance of this transcription factor network during neoplastic growth. Finally, our work highlights some Notch/scrib specificities, in particular the role of the PAR domain-containing basic leucine zipper transcription factor and Notch direct target Pdp1 for neoplastic growth.

Detection of soluble biomarkers of pancreatic cancer in endoscopic ultrasound-guided fine-needle aspiration samples.

BACKGROUND: Biomarkers are urgently needed for pancreatic ductal adenocarcinoma (PDAC). Endoscopic ultrasound-guided fine-needle aspiration (EUS-FNA) is the cornerstone for diagnosing PDAC. We developed a method for discovery of PDAC biomarkers using the discarded EUS-FNA liquid. METHODS: This retrospective study included 58 patients with suspected pancreatic lesions who underwent EUS-FNA. Protein extracts from EUS-FNA liquid were analyzed by mass spectrometry. Proteomic and clinical data were modeled by supervised statistical learning to identify protein markers and clinical variables that distinguish PDAC. RESULTS: Statistical modeling revealed a protein signature for PDAC screening that achieved high sensitivity and specificity (0.92, 95 % confidence interval [CI] 0.79-0.98, and 0.85, 95 %CI 0.67-0.93, respectively). We also developed a protein signature score (PSS) to guide PDAC diagnosis. In combination with patient age, the PSS achieved 100 % certainty in correctly identifying PDAC patients > 54 years. In addition, 3 /4 inconclusive EUS-FNA biopsies were correctly identified using PSS. CONCLUSIONS: EUS-FNA-derived fluid is a rich source of PDAC proteins with biomarker potential. The PSS requires further validation and verification of the feasibility of measuring these proteins in patient sera.

An R package for generic modular response analysis and its application to estrogen and retinoic acid receptor crosstalk.

Modular response analysis (MRA) is a widely used inference technique developed to uncover directions and strengths of connections in molecular networks under a steady-state condition by means of perturbation experiments. We devised several extensions of this methodology to search genomic data for new associations with a biological network inferred by MRA, to improve the predictive accuracy of MRA-inferred networks, and to estimate confidence intervals of MRA parameters from datasets with low numbers of replicates. The classical MRA computations and their extensions were implemented in a freely available R package called aiMeRA ( https://github.com/bioinfo-ircm/aiMeRA/ ). We illustrated the application of our package by assessing the crosstalk between estrogen and retinoic acid receptors, two nuclear receptors implicated in several hormone-driven cancers, such as breast cancer. Based on new data generated for this study, our analysis revealed potential cross-inhibition mediated by the shared corepressors NRIP1 and LCoR. We designed aiMeRA for non-specialists and to allow biologists to perform their own analyses.

MAGI1 inhibits the AMOTL2/p38 stress pathway and prevents luminal breast tumorigenesis.

Alterations to cell polarization or to intercellular junctions are often associated with epithelial cancer progression, including breast cancers (BCa). We show here that the loss of the junctional scaffold protein MAGI1 is associated with bad prognosis in luminal BCa, and promotes tumorigenesis. E-cadherin and the actin binding scaffold AMOTL2 accumulate in MAGI1 deficient cells which are subjected to increased stiffness. These alterations are associated with low YAP activity, the terminal Hippo-pathway effector, but with an elevated ROCK and p38 Stress Activated Protein Kinase activities. Blocking ROCK prevented p38 activation, suggesting that MAGI1 limits p38 activity in part through releasing actin strength. Importantly, the increased tumorigenicity of MAGI1 deficient cells is rescued in the absence of AMOTL2 or after inhibition of p38, demonstrating that MAGI1 acts as a tumor-suppressor in luminal BCa by inhibiting an AMOTL2/p38 stress pathway.

The immune contexture of primary central nervous system diffuse large B cell lymphoma associates with patient survival and specific cell signaling.

Rationale: Primary central nervous system diffuse large B-cell lymphoma (PCNSL) is a rare and aggressive entity that resides in an immune-privileged site. The tumor microenvironment (TME) and the disruption of the immune surveillance influence lymphoma pathogenesis and immunotherapy resistance. Despite growing knowledge on heterogeneous therapeutic responses, no comprehensive description of the PCNSL TME is available. We hence investigated the immune subtypes of PCNSL and their association with molecular signaling and survival. Methods: Analysis of PCNSL transcriptomes (sequencing, n = 20; microarrays, n = 34). Integrated correlation analysis and signaling pathway topology enabled us to infer intercellular interactions. Immunohistopathology and digital imaging were used to validate bioinformatic results. Results: Transcriptomics revealed three immune subtypes: immune-rich, poor, and intermediate. The immune-rich subtype was associated to better survival and characterized by hyper-activation of STAT3 signaling and inflammatory signaling, e.g., IFNγ and TNF-α, resembling the hot subtype described in primary testicular lymphoma and solid cancer. WNT/ß-catenin, HIPPO, and NOTCH signaling were hyper-activated in the immune-poor subtype. HLA down-modulation was clearly associated with a low or intermediate immune infiltration and the absence of T-cell activation. Moreover, HLA class I down-regulation was also correlated with worse survival with implications on immune-intermediate PCNSL that frequently feature reduced HLA expression. A ligand-receptor intercellular network revealed high expression of two immune checkpoints, i.e., CTLA-4/CD86 and TIM-3/LAGLS9. TIM-3 and galectin-9 proteins were clearly upregulated in PCNSL. Conclusion: Altogether, our study reveals that patient stratification according to immune subtypes, HLA status, and immune checkpoint molecule quantification should be considered prior to immune checkpoint inhibitor therapy. Moreover, TIM-3 protein should be considered an axis for future therapeutic development.

Transcriptomic and genomic heterogeneity in blastic plasmacytoid dendritic cell neoplasms: from ontogeny to oncogenesis.

Oncogenesis and ontogeny of blastic plasmacytoid dendritic cell neoplasm (BPDCN) remain uncertain, between canonical plasmacytoid dendritic cells (pDCs) and AXL+ SIGLEC6+ DCs (AS-DCs). We compared 12 BPDCN to 164 acute leukemia by Affymetrix HG-U133 Plus 2.0 arrays: BPDCN were closer to B-cell acute lymphoblastic leukemia (ALL), with enrichment in pDC, B-cell signatures, vesicular transport, deubiquitination pathways, and AS-DC signatures, but only in some cases. Importantly, 1 T-cell ALL clustered with BPDCN, with compatible morphology, immunophenotype (cCD3+ sCD3- CD123+ cTCL1+ CD304+), and genetics. Many oncogenetic pathways are deregulated in BPDCN compared with normal pDC, such as cell-cycle kinases, and importantly, the transcription factor SOX4, involved in B ontogeny, pDC ontogeny, and cancer cell invasion. High-throughput sequencing (HaloPlex) showed myeloid mutations (TET2, 62%; ASXL1, 46%; ZRSR2, 31%) associated with lymphoid mutations (IKZF1), whereas single-nucleotide polymorphism (SNP) array (Affymetrix SNP array 6.0) revealed frequent losses (mean: 9 per patient) involving key hematological oncogenes (RB1, IKZF1/2/3, ETV6, NR3C1, CDKN2A/B, TP53) and immune response genes (IFNGR, TGFB, CLEC4C, IFNA cluster). Various markers suggest an AS-DC origin, but not in all patients, and some of these abnormalities are related to the leukemogenesis process, such as the 9p deletion, leading to decreased expression of genes encoding type I interferons. In addition, the AS-DC profile is only found in a subgroup of patients. Overall, the cellular ontogenic origin of BPDCN remains to be characterized, and these results highlight the heterogeneity of BPDCN, with a risk of a diagnostic trap.

Machine Learning-Assisted Evaluation of Circulating DNA Quantitative Analysis for Cancer Screening.

While the utility of circulating cell-free DNA (cfDNA) in cancer screening and early detection have recently been investigated by testing genetic and epigenetic alterations, here, an original approach by examining cfDNA quantitative and structural features is developed. First, the potential of cfDNA quantitative and structural parameters is independently demonstrated in cell culture, murine, and human plasma models. Subsequently, these variables are evaluated in a large retrospective cohort of 289 healthy individuals and 983 patients with various cancer types; after age resampling, this evaluation is done independently and the variables are combined using a machine learning approach. Implementation of a decision tree prediction model for the detection and classification of healthy and cancer patients shows unprecedented performance for 0, I, and II colorectal cancer stages (specificity, 0.89 and sensitivity, 0.72). Consequently, the methodological proof of concept of using both quantitative and structural biomarkers, and classification with a machine learning method are highlighted, as an efficient strategy for cancer screening. It is foreseen that the classification rate may even be improved by the addition of such biomarkers to fragmentomics, methylation, or the detection of genetic alterations. The optimization of such a multianalyte strategy with this machine learning method is therefore warranted.

The molecular landscape and microenvironment of salivary duct carcinoma reveal new therapeutic opportunities.

Purpose: Salivary duct carcinoma (SDC) is a rare and aggressive salivary gland cancer subtype with poor prognosis. The mutational landscape of SDC has already been the object of several studies, however little is known regarding the functional genomics and the tumor microenvironment despite their importance in oncology. Our investigation aimed at describing both the functional genomics of SDC and the SDC microenvironment, along with their clinical relevance. Methods: RNA-sequencing (24 tumors), proteomics (17 tumors), immunohistochemistry (22 tumors), and multiplexed immunofluorescence (3 tumors) data were obtained from three different patient cohorts and analyzed by digital imaging and bioinformatics. Adjacent non-tumoral tissue from patients in two cohorts were used in transcriptomic and proteomic analyses. Results: Transcriptomic and proteomic data revealed the importance of Notch, TGF-ß, and interferon-γ signaling for all SDCs. We confirmed an overall strong desmoplastic reaction by measuring α-SMA abundance, the level of which was associated with recurrence-free survival (RFS). Two distinct immune phenotypes were observed: immune-poor SDCs (36%) and immune-infiltrated SDCs (64%). Advanced bioinformatics analysis of the transcriptomic data suggested 72 ligand-receptor interactions occurred in the microenvironment and correlated with the immune phenotype. Among these interactions, three immune checkpoints were validated by immunofluorescence, including CTLA-4/DC86 and TIM-3/galectin-9 interactions, previously unidentified in SDC. Immunofluorescence analysis also confirmed an important immunosuppressive role of macrophages and NK cells, also supported by the transcriptomic data. Conclusions: Together our data significantly increase the understanding of SDC biology and open new perspectives for SDC tumor treatment. Before applying immunotherapy, patient stratification according to the immune infiltrate should be taken into account. Immune-infiltrated SDC could benefit from immune checkpoint-targeting therapy, with novel options such as anti-CTLA-4. Macrophages or NK cells could also be targeted. The dense stroma, i.e., fibroblasts or hyaluronic acid, may also be the focus for immune-poor SDC therapies, e.g. in combination with Notch or TGF-ß inhibitors, or molecules targeting SDC mutations.