The genetically predicted causal associations between circulating 3-hydroxybutyrate levels and malignant neoplasms: A pan-cancer Mendelian randomization study.

OBJECTIVE: The ketogenic diet or exogenous supplementation with 3-hydroxybutyrate (3HB) is progressively gaining recognition as a valuable therapeutic or health intervention strategy. However, the effects of 3HB on cancers have been inconsistent in previous studies. This study aimed to comprehensively investigate the causal effects of circulating 3HB levels on 120 cancer phenotypes, and explore the 3HB mediation effect between liver fat accumulation and cancers. METHODS: Univariate Mendelian randomization (UVMR) was used in this study to investigate the causal impact of circulating 3HB levels on cancers. We conducted meta-analyses for 3HB-cancer associations sourced from different exposure data. In multivariate MR(MVMR), the body mass index, alcohol frequency and diabetes were included as covariates to investigate the independent effect of 3HB on cancer risk. Additionally, utilizing mediation MR analysis, we checked the potential mediating role of 3HB in the association between liver fat and cancer. RESULTS: Integrating findings from UVMR and MVMR, we observed that elevated circulating 3HB levels were associated with reduced risk of developing diffuse large B-cell lymphoma(DLBCL) (OR[95%CI] = 0.28[0.14-0.57] p = 3.92e-04), biliary malignancies (OR[95%CI] = 0.30[0.15-0.60], p = 7.67e-04), hepatocellular carcinoma(HCC) (OR[95%CI] = 0.25[0.09-0.71], p = 9.33e-03), primary lymphoid and hematopoietic malignancies (OR[95%CI] = 0.76[0.58-0.99], p = 0.045). Further UVMR analysis revealed that an increase in the percent liver fat was associated with reduced 3HB levels (Beta[95%CI] = -0.073[-0.122∼-0.024], p = 0.0034) and enhanced susceptibility to HCC (OR[95%CI] = 13.9[9.76-19.79], p = 3.14e-48), biliary malignancies (OR[95%CI] = 4.04[3.22-5.07], p = 1.64e-33), nasopharyngeal cancer (OR[95%CI] = 3.26[1.10-9.67], p = 0.03), and primary lymphoid and hematopoietic malignancies (OR[95%CI] = 1.27[1.13-1.44], p = 1.04e-4). Furthermore, 3HB fully mediated the effect of liver fat on susceptibility to DLBCL (OR[95%CI] = 1.076[1.01-1.15], p = 0.034). CONCLUSIONS: Circulating 3HB is associated with a reduced susceptibility to developing DLBCL, HCC, biliary malignancies, and primary lymphoid and hematopoietic malignancies. The impaired ketogenesis induced by metabolic-dysfunction associated fatty liver disease (MAFLD) contributes to risk of DLBCL.

Improving the Generalizability of Infantile Cataracts Detection via Deep Learning-Based Lens Partition Strategy and Multicenter Datasets.

Infantile cataract is the main cause of infant blindness worldwide. Although previous studies developed artificial intelligence (AI) diagnostic systems for detecting infantile cataracts in a single center, its generalizability is not ideal because of the complicated noises and heterogeneity of multicenter slit-lamp images, which impedes the application of these AI systems in real-world clinics. In this study, we developed two lens partition strategies (LPSs) based on deep learning Faster R-CNN and Hough transform for improving the generalizability of infantile cataracts detection. A total of 1,643 multicenter slit-lamp images collected from five ophthalmic clinics were used to evaluate the performance of LPSs. The generalizability of Faster R-CNN for screening and grading was explored by sequentially adding multicenter images to the training dataset. For the normal and abnormal lenses partition, the Faster R-CNN achieved the average intersection over union of 0.9419 and 0.9107, respectively, and their average precisions are both > 95%. Compared with the Hough transform, the accuracy, specificity, and sensitivity of Faster R-CNN for opacity area grading were improved by 5.31, 8.09, and 3.29%, respectively. Similar improvements were presented on the other grading of opacity density and location. The minimal training sample size required by Faster R-CNN is determined on multicenter slit-lamp images. Furthermore, the Faster R-CNN achieved real-time lens partition with only 0.25 s for a single image, whereas the Hough transform needs 34.46 s. Finally, using Grad-Cam and t-SNE techniques, the most relevant lesion regions were highlighted in heatmaps, and the high-level features were discriminated. This study provides an effective LPS for improving the generalizability of infantile cataracts detection. This system has the potential to be applied to multicenter slit-lamp images.