Interpretable machine learning model based on clinical factors for predicting muscle radiodensity loss after treatment in ovarian cancer.

PURPOSE: Muscle radiodensity loss after surgery and adjuvant chemotherapy is associated with poor outcomes in ovarian cancer. Assessing muscle radiodensity is a real-world clinical challenge owing to the requirement for computed tomography (CT) with consistent protocols and labor-intensive processes. This study aimed to use interpretable machine learning (ML) to predict muscle radiodensity loss. METHODS: This study included 723 patients with ovarian cancer who underwent primary debulking surgery and platinum-based chemotherapy between 2010 and 2019 at two tertiary centers (579 in cohort 1 and 144 in cohort 2). Muscle radiodensity was assessed from pre- and post-treatment CT acquired with consistent protocols, and a decrease in radiodensity ≥ 5% was defined as loss. Six ML models were trained, and their performances were evaluated using the area under the curve (AUC) and F1-score. The SHapley Additive exPlanations (SHAP) method was applied to interpret the ML models. RESULTS: The CatBoost model achieved the highest AUC of 0.871 (95% confidence interval, 0.870-0.874) and F1-score of 0.688 (95% confidence interval, 0.685-0.691) among the models in the training set and outperformed in the external validation set, with an AUC of 0.839 and F1-score of 0.673. Albumin change, ascites, and residual disease were the most important features associated with a higher likelihood of muscle radiodensity loss. The SHAP force plot provided an individualized interpretation of model predictions. CONCLUSION: An interpretable ML model can assist clinicians in identifying ovarian cancer patients at risk of muscle radiodensity loss after treatment and understanding the contributors of muscle radiodensity loss.

What Makes a Bacterial Species Pathogenic?: Comparative Genomic Analysis of the Genus Leptospira

Leptospirosis, caused by spirochetes of the genus Leptospira, is a globally widespread, neglected and emerging zoonotic disease. While whole genome analysis of individual pathogenic, intermediately pathogenic and saprophytic Leptospira species has been reported, comprehensive cross-species genomic comparison of all known species of infectious and non-infectious Leptospira, with the goal of identifying genes related to pathogenesis and mammalian host adaptation, remains a key gap in the field. Infectious Leptospira, comprised of pathogenic and intermediately pathogenic Leptospira, evolutionarily diverged from non-infectious, saprophytic Leptospira, as demonstrated by the following computational biology analyses: 1) the definitive taxonomy and evolutionary relatedness among all known Leptospira species; 2) genomically-predicted metabolic reconstructions that indicate novel adaptation of infectious Leptospira to mammals, including sialic acid biosynthesis, pathogen-specific porphyrin metabolism and the first-time demonstration of cobalamin (B12) autotrophy as a bacterial virulence factor; 3) CRISPR/Cas systems demonstrated only to be present in pathogenic Leptospira, suggesting a potential mechanism for this clade's refractoriness to gene targeting; 4) finding Leptospira pathogen-specific specialized protein secretion systems; 5) novel virulence-related genes/gene families such as the Virulence Modifying (VM) (PF07598 paralogs) proteins and pathogen-specific adhesins; 6) discovery of novel, pathogen-specific protein modification and secretion mechanisms including unique lipoprotein signal peptide motifs, Sec-independent twin arginine protein secretion motifs, and the absence of certain canonical signal recognition particle proteins from all Leptospira; and 7) and demonstration of infectious Leptospira-specific signal-responsive gene expression, motility and chemotaxis systems. By identifying large scale changes in infectious (pathogenic and intermediately pathogenic) vs. non-infectious Leptospira, this work provides new insights into the evolution of a genus of bacterial pathogens. This work will be a comprehensive roadmap for understanding leptospirosis pathogenesis. More generally, it provides new insights into mechanisms by which bacterial pathogens adapt to mammalian hosts

Acquisition of negative complement regulators by the saprophyte Leptospira biflexa expressing LigA or LigB confers enhanced survival in human serum

Leptospiral immunoglobulin-like (Lig) proteins are surface exposed molecules present in pathogenic but not in saprophytic Leptospira species. We have previously shown that Lig proteins interact with the soluble complement regulators Factor H (FH), FH like-1 (FHL-1), FH related-1 (FHR-1) and C4b Binding Protein (C4BP). In this study, we used the saprophyte L. biflexa serovar Patoc as a surrogate host to address the specific role of LigA and LigB proteins in leptospiral complement evasion. L biflexa expressing LigA or LigB was able to acquire FH and C4BP. Bound complement regulators retained their cofactor activities of Fl in the proteolytic cleavage of C3b and C4b. Moreover, heterologous expression of ligA and ligB genes in the saprophyte L. biflexa enhanced bacterial survival in human serum. Complement deposition on lig-transformed L biflexa was assessed by flow cytometry analysis. With regard to MAC deposition, L. biflexa expressing LigA or LigB presented an intermediate profile: MAC deposition levels were greater than those found in the pathogenic L. interrogans, but lower than those observed for L. biflexa wildtype. In conclusion, Lig proteins contribute to in vitro control of complement activation on the leptospiral surface, promoting an increased bacterial survival in human serum. (C) 2016 European Federation of Immunological Societies. Published by Elsevier B.V. All rights reserved.