Akkermansia muciniphila is a prominent member of the intestinal microbiota that has the unique ability to subsist on the mucin O-glycans that form a protective barrier between the host and the gut microbiome. Numerous studies highlight its positive role in metabolic regulation and mucosal barrier maintenance, leading to propositions that A. muciniphila could be used as a next-generation probiotic. However, other work indicates that the effects of A. muciniphila vary depending on nutrition, host genetics and the interaction with surrounding microbes. Furthermore, strain-specific differences in the ability to modulate intestinal barrier function and antimicrobial resistance profiles remain underexplored. Here, by focusing on potential sources of this variation, we provide a nuanced discussion on the complex role of A. muciniphila in human health. With A. muciniphila as an example, we argue that a microbe’s specific environment must be considered to enable critical evaluation of next-generation probiotics. This Review discusses the complex role of the gut microbe Akkermansia muciniphila in human health.
Recent tabular Foundational Models (FM) such as TabPFN and TabICL, leverage in-context learning to achieve strong performance without gradient updates or fine-tuning. However, their robustness to adversarial manipulation remains largely unexplored. In this work, we present a comprehensive study of the adversarial vulnerabilities of tabular FM, focusing on both their fragility to targeted test-time attacks and their potential misuse as adversarial tools. We show on three benchmarks in finance, cybersecurity and healthcare, that small, structured perturbations to test inputs can significantly degrade prediction accuracy, even when training context remain fixed. Additionally, we demonstrate that tabular FM can be repurposed to generate transferable evasion to conventional models such as random forests and XGBoost, and on a lesser extent to deep tabular models. To improve tabular FM, we formulate the robustification problem as an optimization of the weights (adversarial fine-tuning), or the context (adversarial in-context learning). We introduce an in-context adversarial training strategy that incrementally replaces the context with adversarial perturbed instances, without updating model weights. Our approach improves robustness across multiple tabular benchmarks. Together, these findings position tabular FM as both a target and a source of adversarial threats, highlighting the urgent need for robust training and evaluation practices in this emerging paradigm.
BACKGROUND:Global disparities exist in cancer incidence, mortality, and survival. We aimed to provide estimates of avoidable deaths among people diagnosed with cancer to inform the prioritisation of interventions and narrow cancer inequalities. METHODS:National incidence estimates for 35 cancer sites in 2022 for 185 countries were extracted from the GLOBOCAN database. We estimated numbers of avoidable deaths within 5 years of diagnosis for patients diagnosed with cancer in 2022, consisting of those deaths avoidable through primary prevention (preventable deaths) and those avoidable through early detection and improved access to treatment (treatable deaths) by cancer site, country, region, and human development index (HDI) group. Preventable deaths were estimated using population attributable fractions for tobacco use, alcohol consumption, excess body weight, infectious agents, and ultraviolet radiation obtained from the literature. Treatable deaths were estimated by eliminating survival differences using 5-year net survival from the SURVCAN-3 project and additional sources. Preventable, treatable, and overall avoidable deaths as proportions of the total expected deaths within 5 years of cancer diagnosis were also calculated. FINDINGS:5 years after cancer diagnosis, 4·5 million (47·6% [95% uncertainty interval 47·5-47·8]) of the 9·4 million expected deaths were avoidable. Of these avoidable deaths, 3·1 million (3·1-3·1; 33·2% [33·1-33·3] of total expected deaths) were preventable and 1·4 million (1·4-1·4; 14·4% [14·4-14·5]) were treatable. Lung, liver, stomach, colorectal, and cervical cancers contributed the greatest burden, collectively accounting for 59·1% of all avoidable deaths. Lung cancer was responsible for the most preventable deaths (1·1 million; 34·6% of all preventable deaths), while female breast cancer was responsible for the most treatable deaths (0·2 million; 14·8% of all treatable deaths). Disproportionately large proportions of avoidable deaths from cervical and breast cancer were observed in countries with a low or medium HDI. INTERPRETATION:Nearly half of deaths among people diagnosed with cancer globally could be avoided through primary prevention and improvements in early detection and curative cancer treatment. Global efforts are needed to tailor prevention, early diagnosis, and treatment of cancer to address inequities in avoidable deaths, especially in low and medium HDI countries. FUNDING:Erasmus Mundus Exchange Programme and French National Cancer Institute (INCa).
Dendritic cells (DCs) are crucial in processing and cross-presenting tumor-associated antigens via MHC I molecules to prime CD8+ T lymphocytes (CTLs) for antitumor immunity. However, the moderate regulation of antigen endocytic escape from lysosomes into the cytoplasm hinders this cross-presentation process due to the lack of finely tunable tools at the organelle-specific level. Magneto-mechanical modulation technology provides a promising approach to regulate lysosomal membrane permeability (LMP) and promote antigen leakage for efficient cross-presentation. This approach relies on magnetic nanorobots within lysosomes that can exert adjustable intracellular forces in response to external rotating magnetic fields. This chapter presents a comprehensive protocol for evaluating the antitumor response of DCs induced by magneto-mechanical regulation. This protocol includes a step-by-step assessment of cytotoxicity on DCs caused by magneto-mechanical regulation, analysis of nanorobot colocalization with lysosomes, and their responses (such as LMP, cytoskeleton morphology, and mitochondria function) through fluorescence imaging. Additionally, we quantify DC mechano-activation and cross-presentation efficiency using flow cytometry.
OBJECTIVES:Using network analysis, which takes a holistic approach to health systems, we aimed to identify which psychosocial burden dimensions are the most central and, thus, critical to prioritising to improve the overall health of people with type 1 diabetes (PwT1D). DESIGN:A cross-sectional network analysis. SETTING:We used data from participants attending 44 diabetes centres in France, who were enrolled in the SFDT1 cohort study between June 2020 and February 2024. PARTICIPANTS:We included 1430 PwT1D (52% women, median age (IQR) 41 (31-52.8) years) who had completed questionnaires on diabetes burden. OUTCOME MEASURES:The items from questionnaires on diabetes distress, fear of hypoglycaemia, quality of life, treatment burden and the impact of diabetes on education and work. RESULTS:The network was highly stable (correlation stability coefficient=0.75). We observed nine domains within the network; 'Loneliness, Worrying & Burnout' was the most influential. We further grouped the domains into three distinct syndromes labelled 'Diabetes Distress', 'Treatment Burden' and 'Impact of Diabetes on Life'. These syndromes reflect the most relevant pillars of the psychosocial burden in PwT1D. CONCLUSIONS:We observed that 'Loneliness, Worrying & Burnout' is the most influential psychosocial burden network domain to prioritise for type 1 diabetes care. This new network-based approach opens the path to defining personalised interventions targeting the most critical burden parameters to expect the most significant overall beneficial impact on PwT1D's health. TRIAL REGISTRATION NUMBER:NCT04657783.
