The sparsity of labeled data is an obstacle to the development of Relation Extraction (RE) models and the completion of databases in various biomedical areas. While being of high interest in drug-discovery, the literature on natural products, reporting the identification of potential bioactive compounds from organisms, is a concrete example of such an overlooked topic. To mark the start of this new task, we created the first curated evaluation dataset and extracted literature items from the LOTUS database to build training sets. To this end, we developed a new sampler, inspired by diversity metrics in ecology, named Greedy Maximum Entropy sampler (https://github.com/idiap/gme-sampler). The strategic optimization of both balance and diversity of the selected items in the evaluation set is important given the resource-intensive nature of manual curation. After quantifying the noise in the training set, in the form of discrepancies between the text of input abstracts and the expected output labels, we explored different strategies accordingly. Framing the task as an end-to-end Relation Extraction, we evaluated the performance of standard fine-tuning (BioGPT, GPT-2, and Seq2rel) and few-shot learning with open Large Language Models (LLMs) (LLaMA 7B-65B). In addition to their evaluation in few-shot settings, we explore the potential of open LLMs as synthetic data generators and propose a new workflow for this purpose. All evaluated models exhibited substantial improvements when fine-tuned on synthetic abstracts rather than the original noisy data. We provide our best performing (F1-score = 59.0) BioGPT-Large model for end-to-end RE of natural products relationships along with all the training and evaluation datasets. See more details at https://github.com/idiap/abroad-re.

Background Brain metastases (BrM) affect up to 60% of patients with metastatic melanoma and are associated with poor prognosis. While combined immune checkpoint blockade of programmed death-1 (PD-1) and cytotoxic T-lymphocyte-associated protein 4 (CTLA-4) demonstrates intracranial efficacy in a proportion of patients with melanoma, the responses are rarely durable, particularly in patients with symptomatic BrM. The brain is an immune-specialized organ and immune responses are regulated differently to the periphery.Methods Using our previously established two-site model of melanoma BrM with concomitant intracranial and extracranial tumors, in which clinically observed efficacy of the combined PD-1/CTLA-4 (PC) blockade can be reproduced, we here explored the role of natural killer (NK) cells in BrM, using functional studies, immunophenotyping and molecular profiling.Results We demonstrate that NK cells are required for the intracranial efficacy of PC blockade. While both perforin and interferon gamma were necessary for the PC blockade-dependent control of intracranial tumor growth, NK cells isolated from intracranial tumors demonstrated only a limited cancer cell killing ability, and PC blockade did not alter the abundance of NK cells within tumors. However, the depletion of NK cells in PC blockade-treated mice led to tumor molecular profiles reminiscent of those observed in intracranial tumors that failed to respond to therapy. Furthermore, the depletion of NK cells resulted in a strikingly reduced abundance of CD8+ T cells within intracranial tumors, while the abundance of other immune cell populations including CD4+ T cells, macrophages and microglia remained unaltered. Adoptive T cell transfer experiments demonstrated that PC blockade-induced trafficking of CD8+ T cells to intracranial tumors was chemokine-dependent. In line with this, PC blockade enhanced intratumoral expression of several T cell-attracting chemokines and we observed high expression levels of cognate chemokine receptors on BrM-infiltrating CD8+ T cells in mice, as well as in human BrM. Importantly, the depletion of NK cells strikingly reduced the intratumoral expression levels of T cell attracting chemokines and vascular T cell entry receptors that were upregulated following PC blockade.Conclusion Our data demonstrate that NK cells underpin the efficacy of PC blockade in BrM by orchestrating the 'responder' molecular profile in tumors, and by controlling the intratumoral abundance of CD8+ T cells through regulation of multiple key molecular mediators of T cell trafficking.

Squamous cell carcinomas with two or more coexisting clearly different histological subtypes of penile carcinomas are designated as mixed carcinomas in current classification models. They represent about 10% of all penile carcinomas. The aim of this study was to detect HPV genotypes in these unusual tumors. Tumors were selected from an international series of 1010 patients with penile carcinomas. Mixed carcinomas were grouped, according to WHO recommendations, as follows: 1. Carcinomas with warty/basaloid features mixed with HPV-independent carcinomas and 2. HPV-independent subtypes mixed with each other. HPV detection and p16(INK4a) immunostaining were performed. For HPV detection, whole tissue section-PCR analyses were performed by SPF10-DEIA-LiPA25 (version 1). As expected, HPV was detected more frequently in HPV-associated mixed carcinomas than in HPV-independent mixed carcinomas. Carcinomas with basaloid or warty features mixed with other SCC subtypes showed an HPV positivity rate of 46% (33 of 72 tumors) compared with 7% found in tumors with nonwarty/basaloid morphology (2 of 29 tumors). Eleven high-risk HPV genotypes were identified and the most common was HPV16 (65%) usually associated with basaloid morphology. p16(INK4a) immunostaining was positive in 76% of HPV-positive tumors. As in nonmixed carcinomas, although in lower proportion, a variable array of HPV genotypes was detected in mixed carcinomas. Apparently, the presence of a non-HPV component in an otherwise typical HPV-associated type tumor does adversely affect the prevalence of HPV positivity. Any amount of HPV-associated morphology superior to 20% in a mixed tumor is sufficient to classify them as HPV-associated, a WHO requirement.

BACKGROUND: The epigenetic factors KAT6A (MOZ/MYST3) and KMT2A (MLL/MLL1) interact in normal hematopoiesis to regulate progenitors' self-renewal. Both proteins are recurrently translocated in AML, leading to impairment of critical differentiation pathways in these malignant cells. We evaluated the potential of different KAT6A therapeutic targeting strategies to alter the growth of KAT6A and KMT2A rearranged AMLs. METHODS: We investigated the action and potential mechanisms of the first-in-class KAT6A inhibitor, WM-1119 in KAT6A and KMT2A rearranged (KAT6Ar and KMT2Ar) AML using cellular (flow cytometry, colony assays, cell growth) and molecular (shRNA knock-down, CRISPR knock-out, bulk and single-cell RNA-seq, ChIP-seq) assays. We also used two novel genetic murine KAT6A models combined with the most common KMT2Ar AML, KMT2A

Ribosome quality control (RQC) resolves collided ribosomes, thus preventing their cytotoxic effects. The chemotherapeutic agent 5-Fluorouracil (5FU) is best known for its misincorporation into DNA and inhibition of thymidylate synthase. However, while a major determinant of 5FU's anticancer activity is its misincorporation into RNAs, the mechanisms by which cancer cells overcome the RNA-dependent 5FU toxicity remain ill-defined. Here, we report a role for RQC in mitigating the cytotoxic effects of 5FU. We show that 5FU treatment results in rapid induction of the mTOR signalling pathway, enhanced rate of mRNA translation initiation, and increased ribosome collisions. Consistently, a defective RQC exacerbates the 5FU-induced cell death, which is mitigated by blocking mTOR pathway or mRNA translation initiation. Furthermore, 5FU treatment enhances the expression of the key RQC factors ZNF598 and GIGYF2 via an mTOR-dependent post-translational mechanism. This adaptation likely mitigates the cytotoxic consequences of increased ribosome collisions upon 5FU treatment.

Prostate-specific antigen (PSA) is the most commonly used serum marker for prostate cancer. It plays a role in cancer detection, treatment monitoring, and more recently, in guiding adaptive therapy protocols, where treatment is alternated based on PSA levels. However, the relationship between PSA levels and tumor volume remains poorly understood. Empirical evidence suggests that different cancer cell types produce varying amounts of PSA. Despite this, current mathematical cancer models often assume either that all cell types contribute equally to PSA levels or that only certain subpopulations produce PSA at fixed rates. In this study, we compare Zhang et al.'s classical adaptive therapy protocol with the standard of care, which involves continuous maximum tolerable dose treatment, under different assumptions regarding PSA production. Specifically, we explore the possibility that testosterone-dependent, testosterone-producing, and testosterone-independent cells contribute to PSA production to varying degrees. We use the time to competitive release as a proxy for the time to disease progression. Our findings indicate that adaptive therapy consistently results in a longer time to competitive release compared to the standard of care, regardless of the assumptions about PSA production. However, when testosterone-independent cells are the sole PSA producers, Zhang et al.'s adaptive therapy protocol becomes inapplicable, as PSA levels never fall to half of their initial value, preventing therapy discontinuation. Additionally, we observe that the number and duration of treatment cycles in adaptive therapy are highly sensitive to assumptions about how much each cell type contributes to PSA production. Overall, our results emphasize the need for a deeper understanding of patient-specific PSA dynamics, which could enhance the effectiveness of adaptive therapy in prostate cancer treatment.

Treatments for cancer patients are becoming increasingly complex, and there is a growing desire from clinicians and patients for biomarkers that can account for this complexity to support informed decisions about clinical care. To achieve precision medicine, the new generation of biomarkers must reflect the spatial and temporal heterogeneity of cancer biology both between patients and within an individual patient. Mining the different layers of 'omics in a multi-modal way from a minimally invasive, easily repeatable, liquid biopsy has increasing potential in a range of clinical applications, and for improving our understanding of treatment response and resistance. Here, we detail the recent developments and methods allowing exploration of genomic, epigenomic, transcriptomic, and fragmentomic layers of 'omics from liquid biopsy, and their integration in a range of applications. We also consider the specific challenges that are posed by the clinical implementation of multi-omic liquid biopsies.

OBJECTIVE: The paper introduces a framework for the evaluation of the encoding of factual scientific knowledge, designed to streamline the manual evaluation process typically conducted by domain experts. Inferring over and extracting information from Large Language Models (LLMs) trained on a large corpus of scientific literature can potentially define a step change in biomedical discovery, reducing the barriers for accessing and integrating existing medical evidence. This work explores the potential of LLMs for dialoguing with biomedical background knowledge, using the context of antibiotic discovery. METHODS: The framework involves three evaluation steps, each assessing different aspects sequentially: fluency, prompt alignment, semantic coherence, factual knowledge, and specificity of the generated responses. By splitting these tasks between non-experts and experts, the framework reduces the effort required from the latter. The work provides a systematic assessment on the ability of eleven state-of-the-art LLMs, including ChatGPT, GPT-4 and Llama 2, in two prompting-based tasks: chemical compound definition generation and chemical compound-fungus relation determination. RESULTS: Although recent models have improved in fluency, factual accuracy is still low and models are biased towards over-represented entities. The ability of LLMs to serve as biomedical knowledge bases is questioned, and the need for additional systematic evaluation frameworks is highlighted. CONCLUSION: While LLMs are currently not fit for purpose to be used as biomedical factual knowledge bases in a zero-shot setting, there is a promising emerging property in the direction of factuality as the models become domain specialised, scale up in size and level of human feedback.

Monitoring melanoma incidence time trends by tumour thickness is essential to understanding the evolution of melanoma occurrence and guiding prevention strategies. To assess long-term incidence trends, tumour thickness was extracted from pathology reports in the Cancer Registry of Norway (1983-2007) and the Norwegian Melanoma Registry (2008-2019), n = 45,635 patients. Across all anatomic sites, T1 (≤ 1 mm) incidence increased most (men annual percentage change [AAPC] = 4.6, 95% confidence interval [95% CI] 4.2-5.0; women AAPC = 3.2, 95% CI 2.8-3.6); the increase was steep until 1989/90, followed by a plateau, and a further steep increase from 2004/05. Increased incidence was also observed for T2 (>1.0-2.0) melanoma (men AAPC = 2.8, 95% CI 2.4-3.2; women AAPC = 1.5, 95% CI 1.1-1.9), and T3 (>2.0-4.0) in men (AAPC = 1.4, 95% CI 0.9-1.9). T4 (>4.0) melanoma followed a similar overall pattern (men AAPC = 1.3, 95% CI 0.9-1.7, head/neck, upper limbs, and trunk; women AAPC = 0.9, 95% CI 0.4-1.4, upper limbs and trunk). Men had the highest T3 and T4 incidence and the sex difference increased with age. Regarding birth cohorts, age-specific incidence increased in all T categories in the oldest age groups, while stabilizing in younger patients born after 1950. Overall, the steep increase in T1 melanoma was not accompanied by a decrease in thick melanoma.