American Journal of Nephrology最新文献

Background: Blood clot formation and capillary fiber blockage in dialyzers remain critical challenges for patients with end-stage kidney disease (ESKD) undergoing hemodialysis. This study aimed to develop a machine learning model that effectively quantifies residual blood clots in dialyzer images captured using bedside smartphone cameras.

Methods: Dialyzer images were collected using mobile phones, and preprocessing techniques-such as background noise removal and image segmentation-were applied to focus on relevant regions. Data augmentation was used to increase model robustness. Composite images were created by combining views from both ends of the dialyzer, enhancing the model's ability to detect residual clots. We developed a binary classification model to distinguish between <10% and ~30% blood clot levels using a pre-trained ConvNeXt architecture. Explainable AI (LIME) was incorporated to ensure the model focused on clinically relevant areas in its predictions.

Results: The dataset was split into training (60%), validation (20%), and testing (20%) sets, with 10 random trials for robustness. The ConvNeXt model achieved an accuracy of 0.6971 without pre-training or data augmentation, which increased to 0.7572 with pre-trained weights. Our combined framework yielded the highest accuracy (0.7672) and reduced standard deviation, indicating greater robustness. For comparison, two nephrology nurses achieved accuracies of 0.6271 and 0.6005 when manually classifying clot levels based solely on end images.

Conclusions: Our approach effectively detects residual blood clots in dialyzers using ConvNeXt by leveraging image data from both ends. The use of explainable AI tools confirmed the model's ability to accurately identify blood clots by focusing on relevant regions. Our study emphasizes the need to balance model complexity with computational efficiency. The ConvNeXt base model successfully avoided overfitting while maintaining practical performance, which could lead to improved clinical decision-making by minimizing circuit downtime and optimizing anemia management.

Introduction: Fibroblast growth factor 23 (FGF23) levels are markedly elevated in patients with kidney failure, but the mechanisms are incompletely understood. Recent evidence suggests that kidney-derived glycerol-3-phosphate (G-3-P), a glycolytic byproduct, mediates FGF23 production in response to dietary phosphate loading. However, the role of G-3-P is unknown in patients with kidney failure.

Methods: Serum G-3-P levels were quantified by LC/MS in 35 healthy individuals and 650 patients undergoing hemodialysis. Association between serum phosphorus and G-3-P was examined using unadjusted and multivariable linear regression models. We next analyzed the associations of G-3-P and known regulators of FGF23 production with serum FGF23.

Results: Median serum G-3-P level in patients undergoing hemodialysis was 220 ng/mL (IQR, 118-325), 2.2-fold higher than that of healthy individuals (98 ng/mL; IQR, 80-129). In patients undergoing hemodialysis, higher serum phosphorus was strongly associated with increased G-3-P in the unadjusted model; this association persisted after multivariate adjustment and when restricted to patients undergoing hemodialysis for over 10 years. In univariate analyses, higher serum phosphorus, calcium, intact PTH, and G-3-P, and active vitamin D use were each significantly associated with higher FGF23. Multivariate analysis identified G-3-P as an independent predictor of FGF23. Further adjustment for transferrin saturation, ferritin, and C-reactive protein did not change these findings.

Conclusion: Even in patients with kidney failure, G-3-P may rise in response to phosphate retention and act as a regulator of FGF23 production. Further studies are needed to test these hypotheses and determine whether the apparently non-functioning kidney retains the capacity to produce G-3-P.

Background: Previous studies on the effects of urate-lowering therapy (ULT) in patients with chronic kidney disease (CKD) and asymptomatic hyperuricemia have yielded conflicting results regarding renal outcomes. This study aimed to investigate the impact of initiating febuxostat on kidney prognosis in patients with stage 3-4 CKD and asymptomatic hyperuricemia using real-world data.

Methods: Using data from Zhejiang Provincial People's Hospital, we conducted a target trial emulation study involving 3232 patients newly diagnosed with stage 3-4 CKD and asymptomatic hyperuricemia between January 1, 2018, and December 31, 2024. Using a clone-censor-weight approach, we compared the strategy of initiating febuxostat within one year of the first detected serum uric acid (UA) level >420 μmol/L versus to no initiation. Patients were followed for up to 5 years after hyperuricemia was diagnosed. The primary outcome was a composite kidney outcome consisting of a ≥40% decline in estimated glomerular filtration rate (eGFR), end-stage kidney disease (ESKD) (eGFR <15 mL/min/1.73 m²), or initiation of kidney replacement therapy (KRT). Secondary outcomes included all-cause mortality and major adverse cardiovascular events (MACE).

Results: Among patients newly diagnosed with CKD and hyperuricemia (mean age 71.8 years, 64% male, mean eGFR 42.8 mL/min/1.73 m², mean serum UA level 499.4 μmol/L), 631 individuals (20%) initiated febuxostat therapy. Compared with non-initiation, febuxostat initiation was not significantly associated with the primary composite kidney outcome (HR, 1.07; 95% CI, 0.91 to 1.20), all-cause mortality (HR, 1.00; 95% CI, 0.66 to 1.35), or MACE (HR, 1.03; 95% CI, 0.95 to 1.11).

Conclusion: In patients with stage 3-4 CKD and asymptomatic hyperuricemia, initiation of febuxostat was not associated with kidney outcomes, all-cause mortality, or MACE. Further studies are warranted to validate these findings.

Introduction: The global incidence of chronic kidney disease (CKD) continues to rise, but delayed epidemiological data pose challenges to public health policy. Traditional surveillance methods often suffer from reporting delays. Recent advances in artificial intelligence (AI) offer novel opportunities for enhancing disease burden predictions.

Methods: We collected CKD incidence data from 21 Global Burden of Disease (GBD) regions spanning from 1990 to 2021. Using five advanced AI models (GPT-4o, Claude-3.7, DeepSeek-R1, Grok-3, and Gemini 2.5) and two traditional forecasting methods (autoregressive integrated moving average and Bayesian age-period-cohort), we predicted CKD incidence for 2023. The performance of the models was evaluated by comparing the predicted values to the actual observed data. All models were trained using the same data and instructions.

Results: The AI models and traditional models performed similarly, with near-perfect accuracy in predicting incidence rates in regions such as the Americas, Central Europe, East Asia, high-income Asia Pacific, Southeast Asia, and tropical Latin America. Among the models, GPT-4o demonstrated the highest mean accuracy of 0.722, with all models achieving average accuracies above 0.65. No statistically significant difference in accuracy was observed between AI-based and traditional models (ANOVA p = 0.27).

Conclusion: State-of-the-art AI models, when systematically prompted and standardized, can predict global CKD incidence with accuracy comparable to traditional statistical models. AI-driven epidemiological forecasting holds promise for enhancing real-time public health planning and resource allocation, particularly in regions with stable historical data.

Objective: To explore the clinical value of diffusion-weighted imaging (DWI) combined with T2 mapping for the early evaluation of delayed graft function (DGF) after renal transplantation.

Method: A total of 92 patients who underwent allogeneic renal transplantation were prospectively recruited. All patients underwent magnetic resonance imaging of the transplanted kidneys after the operation. Diffusion-weighted imaging and mapping of the T2-relaxation time (T2 mapping) were measured and analysed. According to the recovery of renal function within one week after surgery, the patients were divided into a normal graft function (NGF) group and a DGF group.

Results: The apparent diffusion coefficient (ADC) values of the cortex and medulla in the DGF group were significantly lower than in the NGF group (P < 0.05). The T2 values of the cortex and medulla in the DGF group were significantly higher than in the NGF group (P < 0.05). The eGFR was positively correlated with the cortical ADC value (P < 0001) and medullary ADC value (P < 0001) and negatively correlated with the medulla T2 value (P = 0.027). The results of the binary logistic regression analysis indicated that age, creatinine level, eGFR, cortical ADC value, medulla ADC value, cortical T2 value and medulla T2 value were independent factors related to DGF. Using these 7 indicators for joint prediction, the AUC was 0.895, and the prediction effect was good.

Conclusion: Diffusion-weighted imaging combined with T2 mapping has important differential diagnostic value for DGF after renal transplantation.

Introduction: Renin-angiotensin system inhibitors (RASi) are critical for cardiovascular diseases (CVD), but adverse effects sometimes lead to discontinuation, raising concerns about impacts on major outcomes. Although the observational studies have suggested continuation or restarting of RASi, the evidence from randomized controlled trials (RCTs) and systematic reviews based on RCTs are not sufficient.

Method: We performed a systematic review and meta-analysis including only RCTs. We searched MEDLINE, EMBASE, CENTRAL, ClinicalTrials.gov, and EU Clinical Trials Register for the full text review analysis. Primary outcomes included all-cause death and CVD events. Risk of bias was assessed using version 2 of the Cochrane Risk of bias tool (RoB2), and the certainty of evidence was evaluated using the Grading of Recommendations Assessment, Development, and Evaluation approach.

Results: Among the seven included RCTs (n = 928), three studies (n = 745) reported all-cause mortality and two studies (n = 697) reported CVD events. The meta-analysis did not show difference in all-cause mortality between intervention and control groups (RR 0.95, 95% CI: 0.54 to 1.65, I2 = 0%) and CVD events (RR 1.22, 95% CI: 1.00 to 1.50, I2 = 0%) between intervention and control groups. The certainty of evidence was rated as very low for both outcomes due to RoB, imprecision, and clinical heterogeneity.

Conclusion: In this systematic review and meta-analysis, there might not be deference in the risk of all-cause mortality or CVD events following RASi discontinuation compared with continuation. The number of enrolled studies was limited, and the certainty of evidence was very low, thus our results should be interpreted carefully.

Introduction: Cardiovascular-kidney-metabolic (CKM) syndrome significantly impacts clinical outcomes, though evidence linking integrated cardiometabolic-kidney biomarkers to prognosis remains sparse. This study evaluated prognostic associations of these biomarkers and developed machine learning (ML)-based mortality prediction models for CKM patients.

Methods: Using NHANES data (1999-2018) and death records from 10,616 stage 0-3 CKM patients, we analyzed cardiometabolic-kidney indices: cardiometabolic index (CMI), atherogenic index of plasma (AIP), estimated glomerular filtration rate (eGFR), and urinary albumin-creatinine ratio (uACR). Survival analysis incorporated the Kaplan-Meier curves, Cox regression, and restricted cubic splines to evaluate nonlinear associations. Risk reclassification was quantified via net reclassification index (NRI) and integrated discrimination improvement (IDI). Optimal mortality thresholds were determined using survival cut-point analysis, and inflammation's mediating role was explored. Seven ML models were trained, with performance assessed by area under the receiver operating characteristic curve (AUC-ROC), Brier score, and net clinical benefit.

Results: Over a median 96-month follow-up, 847 deaths occurred. Elevated CMI, AIP, and uACR, along with reduced eGFR, independently predicted mortality (all p < 0.05), with nonlinear trends for CMI, eGFR, and uACR (p-nonlinearity < 0.05). High-risk thresholds for these indices increased mortality risk by 1.19-1.91-fold. Combining all indices improved risk stratification (NRI = 15.8%, IDI = 3.4%). Inflammation mediated 1.1-5.0% of biomarker-mortality associations. Among ML models, XGBoost achieved optimal performance (AUC = 0.852, 95% CI: 0.829-0.877), with Brier score of 0.063 (95% CI: 0.056-0.069) and provided clinical net benefits across risk thresholds from 0 to 0.6.

Conclusion: Cardiometabolic-kidney indices significantly associated with prognosis in CKM patients, highlighting the importance of heart-kidney-metabolism crosstalk. Combining easily accessible biomarkers with the XGBoost model may facilitate risk stratification.

Introduction: Urgent dialysis is labor-intensive and expensive because it requires specialized nursing staff. Most hospitals schedule a fixed number of nurses daily for urgent dialysis needs, but daily dialysis demand fluctuates, leading to inefficiencies.

Methods: We developed statistical, machine learning, and deep learning models to predict the next 7 days' dialysis needs. Our study included a retrospective (April 1, 2018, to March 31, 2023) and prospective component (November 1 to 30, 2023, and May 31 to June 27, 2024) across four hospitals (hospital A for one hospital and hospital B for three hospitals combined). To avoid model over-fitting, we divided our data into three sets: training, testing, and validation. The latter was performed prospectively during two silent deployment periods. The primary outcome measure was the mean absolute error (MAE).

Results: The mean daily dialysis volume in the retrospective data was 16.0 (standard deviation [SD], 5.7) for hospital A and 4.5 (SD, 2.3) for hospital B. The best performing models were autoregressive integrated moving average (ARIMA) and temporal convolutional network; both resulted in an MAE of 3.0 procedures for hospital A and 1.5 procedures for hospital B, compared to 4.4 and 1.9, respectively, for the benchmark. During our two prospective evaluations, the mean daily dialysis volume was 16.8 (SD, 4.5) for hospital A and 4.2 (SD, 2.5) for hospital B. The ARIMA model resulted in the lowest MAE at 2.2 and 1.5 procedures, respectively.

Conclusions: Our multicenter, 6-year study demonstrated that urgent in-hospital dialysis needs can be accurately forecasted.

Introduction: BK polyomavirus (BKPyV) in kidney transplant associated with adverse graft outcome. The aim of this study was to examine graft loss risk of BKPyV associated nephropathy (BKPyVAN) and BKPyV-DNAemia in relation with de novo donor-specific antibody (DSA) and rejection status.

Methods: Two hundred and forty patients from a multicentre cohort who had regular BKPyV and donor-DSA surveillance were retrospectively reviewed and stratified according to the presence of BKPyV-DNAemia and rejection.

Results: BKPyV-DNAemia did not associate with de novo DSA development (hazard ratio [HR] 1.15, 95% confidence interval [CI] 0.50-2.67, p = 0.74) but de novo DSA was more commonly observed in patients who developed rejection (BKV+/rejection- 4.3% (n = 2) vs. BKV+/rejection+ 57.1% (n = 4), p < 0.001). BKPyV-DNAemia (adjusted HR 4.02, 95% CI: 1.30-12.43, p = 0.016) and de novo DSA (adjusted HR 6.76, 95% CI: 2.51-18.24, p < 0.001) were independent factors associated with antibody-mediated rejection. Patients with BKPyV-DNAemia who were further complicated with rejection had approximately 6-fold risk of graft loss (adjusted HR 6.24, 95% CI: 2.04-19.09, p = 0.001), whereas patient with BKPyV-DNAemia alone did not experience significant increase graft loss risk (adjusted HR 1.76, 95% CI: 0.64-4.81, p = 0.27).

Conclusions: Our study suggested that DSA monitoring would be warranted during immunosuppressant reduction for BKPyV-DNAemia and less aggressive reduction of immunosuppressant when DSA emerges might be a reasonable strategy to avoid overzealous reduction of immunosuppressant that could precipitate allograft rejection.