A peer-reviewed open access journal published by the National Institute of Environmental Health Sciences

Prediction of kidney function and chronic kidney disease (CKD) through kidney ultrasound imaging has long been considered desirable in clinical practice because of its safety, convenience, and affordability. However, this highly desirable approach is beyond the capability of human vision. We developed a deep learning approach for automatically determining the estimated glomerular filtration rate (eGFR) and CKD status. We exploited the transfer learning technique, integrating the powerful ResNet model pretrained on an ImageNet dataset in our neural network architecture, to predict kidney function based on 4,505 kidney ultrasound images labeled using eGFRs derived from serum creatinine concentrations. To further extract the information from ultrasound images, we leveraged kidney length annotations to remove the peripheral region of the kidneys and applied various data augmentation schemes to produce additional data with variations. Bootstrap aggregation was also applied to avoid overfitting and improve the model’s generalization. Moreover, the kidney function features obtained by our deep neural network were used to identify the CKD status defined by an eGFR of <60 ml/min/1.73 m². A Pearson correlation coefficient of 0.741 indicated the strong relationship between artificial intelligence (AI)- and creatinine-based GFR estimations. Overall CKD status classification accuracy of our model was 85.6% —higher than that of experienced nephrologists (60.3%–80.1%). Our model is the first fundamental step toward realizing the potential of transforming kidney ultrasound imaging into an effective, real-time, distant screening tool. AI-GFR estimation offers the possibility of noninvasive assessment of kidney function, a key goal of AI-powered functional automation in clinical practice.

Current evidence supports the notion that environmental exposures are associated with DNA-methylation and expression changes that can impact human health. Our objective was to conduct a systematic review of epidemiologic studies evaluating the association between environmental chemicals with DNA methylation levels in adults. After excluding arsenic, recently evaluated in a systematic review, we identified a total of 17 articles (6 on cadmium, 4 on lead, 2 on mercury, 1 on nickel, 1 on antimony, 1 on tungsten, 5 on persistent organic pollutants and perfluorinated compounds, 1 on bisphenol A, and 3 on polycyclic aromatic hydrocarbons). The selected articles reported quantitative methods to determine DNA methylation including immunocolorimetric assays for total content of genomic DNA methylation, and microarray technologies, methylation-specific quantitative PCR, Luminometric Methylation Assay (LUMA), and bisulfite pyrosequencing for DNA methylation content of genomic sites such as gene promoters, LINE-1, Alu elements, and others. Considering consistency, temporality, strength, dose-response relationship, and biological plausibility, we concluded that the current evidence is not sufficient to provide inference because differences across studies and limited samples sizes make it difficult to compare across studies and to evaluate sources of heterogeneity. Important questions for future research include the need for larger and longitudinal studies, the validation of findings, and the systematic evaluation of the dose-response relationships. Future studies should also consider the evaluation of epigenetic marks recently in the research spotlight such as DNA hydroxymethylation and the role of underlying genetic variants.

Introduction: Abdominal surgery is probably associated with more likelihood to cause acute kidney injury (AKI). The aim of this study was to evaluate whether early or late start of renal replacement therapy (RRT) defined by simplified RIFLE (sRIFLE) classification in AKI patients after major abdominal surgery will affect outcome. Methods: A multicenter prospective observational study based on the NSARF (N ational Taiwan University S urgical ICU A ssociated R enal F ailure) Study Group database. 98 patients (41 female, mean age 66.4 ± 13.9 years) who underwent acute RRT according to local indications for post-major abdominal surgery AKI between 1 January, 2002 and 31 December, 2005 were enrolled The demographic data, comorbid diseases, types of surgery and RRT, as well as the indications for RRT were documented. The patients were divided into early dialysis (sRIFLE-0 or Risk) and late dialysis (LD, sRIFLE -Injury or Failure) groups. Then we measured and recorded patients' outcome including in-hospital mortality and RRT wean-off until 30 June, 2006. Results: The in-hospital mortality was compared as endpoint. Fifty-seven patients (58.2%) died during hospitalization. LD (hazard ratio (HR) 1.846; P = 0.027), old age (HR 2.090; P = 0.010), cardiac failure (HR 4.620; P < 0.001), pre-RRT SOFA score (HR 1.152; P < 0.001) were independent indicators for in-hospital mortality. Conclusions: The findings of this study support earlier initiation of acute RRT, and also underscore the importance of predicting prognoses of major abdominal surgical patients with AKI by using RIFLE classification.

OBJECTIVE: Little is known about arsenic metabolism in diabetes development. We investigated the prospective associations of low-moderate arsenic exposure and arsenic metabolism with diabetes incidence in the Strong Heart Study. RESEARCH DESIGN AND METHODS: A total of 1,694 diabetes-free participants aged 45–75 years were recruited in 1989–1991 and followed through 1998–1999. We used the proportions of urine inorganic arsenic (iAs), monomethylarsonate (MMA), and dimethylarsinate (DMA) over their sum (expressed as iAs%, MMA%, and DMA%) as the biomarkers of arsenic metabolism. Diabetes was defined as fasting glucose ≥126 mg/dL, 2-h glucose ≥200 mg/dL, self-reported diabetes history, or self-reported use of antidiabetic medications. RESULTS: Over 11,263.2 person-years of follow-up, 396 participants developed diabetes. Using the leave-one-out approach to model the dynamics of arsenic metabolism, we found that lower MMA% was associated with higher diabetes incidence. The hazard ratios (95% CI) of diabetes incidence for a 5% increase in MMA% were 0.77 (0.63–0.93) and 0.82 (0.73–0.92) when iAs% and DMA%, respectively, were left out of the model. DMA% was associated with higher diabetes incidence only when MMA% decreased (left out of the model) but not when iAs% decreased. iAs% was also associated with higher diabetes incidence when MMA% decreased. The association between MMA% and diabetes incidence was similar by age, sex, study site, obesity, and urine iAs concentrations. CONCLUSIONS: Arsenic metabolism, particularly lower MMA%, was prospectively associated with increased incidence of diabetes. Research is needed to evaluate whether arsenic metabolism is related to diabetes incidence per se or through its close connections with one-carbon metabolism.