Editors’ Selections from Relevant Scientific Publications

PROX1 promotes hepatocyte cell fate (by Lim et al. via Nat Genet)Cell fate plasticity is essential for development, while unlocked plasticity is a cancer hallmark. Using a computational approach, Lim and colleagues identified “safeguard repressors” for 18 cell types that block phenotypic plasticity lifelong. They further studied a protein called PROX1, which stably establishes liver cells as hepatocytes by suppressing genetic programs that give rise to other cell fates. This function also serves as a safeguard against tumorigenesis—the researchers showed that overexpression of the Prox1 gene prevented both the onset and progression of hepatocellular carcinoma (HCC) in cancer-prone mouse models. They also analyzed tumor specimens from HCC patients and found that elevated Prox1 expression was associated with increased overall survival. These results indicate that a better understanding of PROX1 and other cell-type-specific safeguard repressors could reveal new opportunities to stall or prevent cancer growth.Lim B, … Mall M. Nat Genet.2025 Feb 13.Single-nucleotide variants (by Genomics Education via Genomelink)A new study nominated single-nucleotide variants (SNVs) in regulatory DNA and their putative target genes associated with inherited cancer risk. Kellman and colleagues analyzed data from 4041 SNVs using massively parallel reporter assays (MPRA) across 13 common human cancers in diploid human cells. They identified 380 MPRA-functional, differentially active SNVs (daSNVs). By integrating multiple orthogonal data sources, the researchers nominated 1,169 putative gene targets of those daSNVs linked to cancer risks. Further investigation of one specific SNV, rs10411210 suggested that its risk allele may increase cancer risk by upregulating RHPN2 expression, thereby enhancing RhoA activation. This study provides a biological basis for variant prioritization and further examination of carcinogenic mechanisms.Kellman LN, … Khavari PA. Nat Genet.2025 Feb 17.Differential Methylation between cancerous and non-cancerous samples (by Zhang et al. via J Clin Invest)A recent study developed a method for detecting early-stage esophageal cancer (EC) using methylated DNA markers (MDM) from various genomic regions in blood samples. Zhang and colleagues identified 31 methylated regions associated with EC by analyzing whole-genome bisulfite sequencing (WGBS) data from plasma samples of 56 EC patients and 107 healthy controls, as well as methylation data from public databases. The researchers subsequently constructed a three-MDM model by optimizing the probes in a small cohort (n = 297). This model was validated in a multicenter, prospective cohort (n = 1429) and distinguished EC patients from non-EC volunteers with a sensitivity of 85.5% and a specificity of 95.3%. It also detected early-stage tumors with sensitivities of 56% for Stage 0 and 77% for Stage I. These findings suggest that measuring tumor-associated methylation status in blood samples may serve as a non-invasive and reliable method for early-stage EC detection.Zhang R, … Li Y. J Clin Invest.2025 Feb 25:e186816.Mammography (by Hernström et al via Lancet Digit Health)In Europe, mammograms are reviewed by two radiologists to ensure more accurate breast cancer screening, but this creates a substantial workload for specialists. Hernström and colleagues recently published results from the Mammography Screening with Artificial Intelligence (MASAI) trial, which evaluated whether an AI trained on radiological data could be used to triage screening examinations and as detection support by highlighting suspicious findings. Over 53,000 women had their mammograms initially analyzed by AI; only those designated “high risk” were referred for double reading, while all others were reviewed by a single human expert. The algorithm performed well, detecting 29% more cancers relative to the control cohort of women who received standard double-reading, achieving roughly the same false-positive rate while reducing radiologists’ workload by 44%. Pending further validation, this approach could improve the efficiency of screening without compromising quality of care.Hernström V, … Lång K. Lancet Digit Health. 2025 Feb 3:S2589-7500(24)00267-X.