Editors’ Selections from Relevant Scientific Publications

Lactobacillus johnsonii encodes bile salt hydrolase (by Liu et al. via Cancer Research)A new study reveals that individual susceptibility to high-fat diet (HFD)-induced colorectal cancer (CRC) is influenced by gut microbiota composition. Using the ApcMin/+ mouse model, Liu and colleagues identified Lactobacillus johnsonii (L. johnsonii) as a protective bacterium enriched in animals resistant to HFD-induced colorectal tumorigenesis. Administration of L. johnsonii abrogated the pro-tumorigenic effects of HFD in both ApcMin/+ and azoxymethane/dextran sulfate-induced mouse models by elevating intestinal levels of chenodeoxycholic acid (CDCA). L. johnsonii converted conjugated bile acids into CDCA via bacterial bile salt hydrolase activity, and CDCA induced mitochondrial dysfunction and oxidative stress, leading to apoptosis in cancer cells. These findings highlight the gut microbiota’s role in modulating cancer susceptibility and suggest that probiotic interventions targeting bile acid metabolism may represent a precision prevention strategy for HFD-associated CRC.Liu C, … He Z. Cancer Res. 2025 Sep 15.Microbiota (via Easy-Peasy.AI)A new study indicates that specific oral bacteria and fungi are associated with an increased risk of pancreatic cancer. Analyzing data from 122,000 participants in two large epidemiological cohorts (CPS-II and PLCO), Meng and colleagues characterized the oral bacteriome and mycobiome through whole-genome shotgun sequencing and internal transcribed spacer sequencing, respectively. Three periodontal pathogens (Porphyromonas gingivalis, Eubacterium nodatum, and Parvimonas micra) along with 13 additional oral bacteria and the fungi Candida tropicalis and unspecified Candida (Candida spp.), were associated with an increased risk of pancreatic cancer, whereas 8 oral bacteria and the fungi Candida albicans and Malassezia globosa associated with a decreased risk. A microbial risk score (MRS) based on 27 oral species was calculated, with each standard deviation increase in MRS associated with more than a threefold higher risk of pancreatic cancer. These findings implicate the oral microbiome in pancreatic carcinogenesis and suggest its potential as a noninvasive biomarker for risk stratification and targeted prevention.Meng Y, … Ahn J. JAMA Oncol. 2025 Sep 18.Commonly used GLP-1RAs (by Benff via Wikimedia Commons)A recent study investigated the association between glucagon-like peptide-1 receptor agonists (GLP-1RAs), commonly prescribed for type-2 diabetes and weight management, and cancer risks among adults with obesity. Using a target trial emulation design with electronic health records from the multicenter OneFlorida+ health research network, Dai and colleagues conducted a retrospective cohort study of 86,632 matched participants including 43,317 GLP-1RA users and 43,315 nonusers to compare the incidence of 14 cancer types. GLP-1RA use was significantly associated with a reduced overall cancer risk, particularly for endometrial, ovarian, and meningioma cancers, while a marginally nonsignificant increased risk of kidney cancer was observed. These results suggest that GLP-1RAs may confer protective effects against certain cancers, although potential risks remain. The study underscores the need for long-term follow-up to better understand these associations and ensure patient safety as GLP-1RAs continue to be widely used for obesity treatment.Dai H, … Bian J. JAMA Oncol. 2025 Aug 21:e252681.Compact microscope (by Bai et al. via Nat Commun)A recent study demonstrates that an AI-assisted, ultra-low-cost microscope system can effectively detect high-risk cervical precancer in resource-limited settings. Bai and colleagues built a compact microscope using consumer-grade electronics and aspherical lenses. To overcome the limited resolution that hampers accurate identification of abnormal cells in cervical samples, the researchers trained a coarse instance classifier to extract feature sequences from the top 200 instances most likely to contain lesions. They further developed the Att-Transformer model to selectively attend to and integrate sparse lesion information from these sequences, enabling accurate slide grading. Trained on 3,510 slides from four hospitals, the model was validated across four independent datasets, achieving area under the curve (AUC) values of 0.87 and 0.89 on slides from two external hospitals, 0.89 on newly collected slides from the original hospitals, and 0.85 on HPV-positive slides. These results highlight the potential of this low-cost, AI-powered approach to expand access to cervical cancer screening and improve early detection in underserved communities where conventional cytology remains impractical due to cost and workforce limitations.Bai J, … Liu X. Nat Commun. 2025 Aug 11;16(1):7429.