Background & Aims Metabolic dysfunction-associated steatohepatitis (MASH) is a prevalent chronic liver disease characterized by steatosis, inflammation, and hepatocyte damage, with or without fibrosis. Robust preclinical models that represent the clinical features of MASH are critical for drug development. We aimed to examine the effects of two widely used Western-type diets and thermoneutral housing on the induction of MASH and liver fibrosis in mice. Methods Male mice were fed either a Gubra-Amylin (GAN) MASH or Western non-trans-fat diet and were housed at standard room temperature (21°C) for a period of 25, 33, or 46 weeks or at thermoneutrality (30°C) for 25 weeks. Liver histology, transcriptome profiling, and in situ imaging cytometry were applied to analyze the differences between the models. Results Liver inflammation and hepatocyte damage were increased by housing mice at thermoneutrality. Remarkably, mice housed at thermoneutrality developed liver fibrosis after 25 weeks of diet feeding, reaching a degree of extracellular matrix deposition equivalent to the levels that required 46 weeks of diet feeding at standard temperature. Thermoneutral housing enhanced the expression of genes related to extracellular matrix organization and inflammatory pathways, demonstrating higher similarity to human MASH. This model was also characterized by distinctive hepatic immune cell infiltration and greater cell-cell interactions, compared to housing at standard temperatures. Conclusion Housing diet-induced MASH mice at thermoneutrality accelerated fibrosis onset and induced increased disease severity. Phenotypically and transcriptionally, this mouse model showed greater similarity to human MASH. In addition, MASH mice at thermoneutrality developed a distinct and enhanced hepatic innate and adaptive immune response.

INTRODUCTION:Kidney injury molecule-1 (KIM-1) expression reflects proximal renal tubular damage, but plasma and urine KIM-1 have not been jointly studied in a CKD cohort. METHODS:Plasma and urine KIM-1 were measured in 2,581 adults from the NURTuRE-CKD cohort, a multicentre, non-dialysis-dependent CKD cohort. Survival analyses, C-statistics, and net reclassification improvement were used to assess associations and predictive performance of plasma and urine KIM-1 for kidney failure (KF), all-cause mortality, and a secondary endpoint of combined CKD progression endpoint (CKE) (KF or >40% decline in eGFR) in the total cohort and in KDIGO albuminuria categories, early CKD (eGFR >45 mL/min/1.73 m2), and four plasma/urine KIM-1 groups, dichotomised above and below the median value. RESULTS:Median age was 65 years, baseline eGFR 34.8 mL/min/1.73 m2, and urine albumin-to-creatinine ratio (uACR) 22.3 mg/mmol. During median follow-up of 48.8 months, 616 (23.9%) participants developed KF, 817 (32%) experienced CKE, and 344 (13.3%) died. Plasma and urine KIM-1 levels increased with lower eGFR, higher uACR, and diabetes. Plasma KIM-1 was independently associated with KF, while urine KIM-1 was associated with pre-KF death. The combination of high plasma and high urine KIM-1 conferred the greatest hazards of KF and all-cause mortality. Combining plasma and urine KIM-1 led to a 24.1% improvement in net reclassification index for KF. In earlier stages of CKD, both biomarkers were associated with CKD progression and there were large improvements in risk prediction for plasma KIM-1 alone. Increased albuminuria amplified the relationship between plasma and urine KIM-1 and KF risk. CONCLUSION:This study highlights distinct prognostic associations of plasma and urine KIM-1 in CKD. Measuring both may be useful in improving risk stratification in people with CKD. For early-stage CKD, the need to use a combined CKE, including decline in eGFR, is emphasised as few of these people developed KF.

BACKGROUND AND AIMS:Chronic kidney disease (CKD) patients are particularly susceptible to coronary atherosclerosis, which can be assessed using computed tomography (CT)-based coronary artery calcium (CAC) score. However, such a costly examination might not always be required and cost-effective. This study investigates a novel screening approach utilizing pulse wave analysis combined with machine learning models to identify CKD patients at high risk for coronary atherosclerosis. METHODS:We analyzed retrospective data from 124 CKD stage 5 patients who underwent kidney transplantation. Pulse wave signals were collected using SphygmoCor system (AtCor Medical, Sydney, Australia), and CAC scores were determined via CT scans. Machine learning models were developed using either pulse wave features or traditional risk factors (TRF) to detect high CAC scores (≥100 Agatston units). RESULTS:The pulse wave-based model outperformed TRF-based model in identifying high CAC scores, particularly among younger patients. Specifically, the pulse wave-based classifier showed superior balanced accuracy in all analyzed age groups and superior sensitivity in patients under 60 years old, especially in those under 50 years old. The overall balanced accuracy of the pulse wave-based model exceeded 80 %, suggesting its potential as a reliable screening tool for detecting high risk of coronary atherosclerosis in CKD patients. CONCLUSIONS:Pulse wave analysis combined with machine learning offers a promising, non-invasive method for preliminary CAC screening in CKD patients. This approach could enhance early risk identification and improve clinical management, although further research is needed to validate and refine this method in larger, more diverse populations.

Background: AZD0233 is an orally administered, selective allosteric modulator of the C-X3-C motif chemokine receptor 1 (CX3CR1). CX3CR1 is a transmembrane receptor expressed by subsets of leukocytes, including monocytes, T-cells and NK cells. Upon binding to its ligand CX3CL1, CX3CR1 facilitates immune cell adhesion and migration into inflamed tissue. Increased expression of CX3CR1 has been observed in the myocardial biopsies from dilated cardiomyopathy (DCM) patients. Hypothesis: AZD0233 treatment will improve cardiac function by modulating cardiac inflammation in a preclinical DCM model. Methods: We assessed the efficacy of oral administration of AZD0233 on cardiac function and immune cell modulations using the muscle LIM protein knock-out (MLP-KO) mouse model of DCM. Results: After 4 weeks of AZD0233 treatment, a significant reduction in CD11b + Ly6C hi inflammatory monocytes was observed in the myocardial tissue by flow cytometry (0.90±0.10% at 100 mg/kg vs 1.31±0.19% in vehicle, p<0.05, n=9). Echocardiography indicated a dose-dependent improvement of systolic function as measured by an increase in left ventricular ejection fraction (percentage units 7.1 ± 1.5 higher at 100 mg/kg than vehicle treated p<0.001, n=11). After 8 weeks of treatment, histological analyses of myocardial sections showed a dose-dependent relative reduction in CX3CR1 + leukocytes by immunohistochemistry staining (48.6±21.3% at 100 mg/kg p<0.05 vs vehicle, n=11) and a decrease in myocardial fibrosis by Picrosirius Red staining (3.6±1.7% at 100 mg/kg vs 6.1±2.6% vehicle, p<0.01, n=11). Single nucleus RNA-seq analyses revealed a reduction in abundance of CD45 + leukocytes starting at 2 weeks and lasting up to 8 weeks during AZD0233 treatment. Subset-specific transcriptomic data indicated that macrophages differentially expressed genes involved in M2 polarization and inflammation resolution. Conversely, transcriptional changes in fibroblasts suggest reduced activation and fibrosis, along with increased cellular senescence. Conclusions/Perspectives: AZD0233 treatment exhibited cardiac protective effects, improving cardiac function and promoting reverse remodeling in MLP-KO mice. Histological and transcriptomic analyses indicated AZD0233 reduced CD45 + immune cell populations, shifted macrophage polarization towards M2 phenotype and inflammation resolution.

Background:The aim of the study was to investigate urinary and serum tumour necrosis factor (TNF)-like weak inducer of apoptosis (TWEAK) as potential biomarkers in a longitudinal cohort of patients with ANCA-associated vasculitis (AAV). Methods:Patients with active AAV were included in the study. The Birmingham Vasculitis Score 2003 (BVAS) was used for assessment of disease activity and C-reactive protein (CRP), creatinine, albuminuria, and serum (s) and urinary (u) TWEAK levels were measured at baseline and 6-month follow-up. sTWEAK was measured in population-based controls for comparison. Kidney biopsies from AAV patients were stained for TWEAK and its receptor fibroblast growth factor-inducible 14 (Fn14) using immunohistochemistry (IHC). Results:sTWEAK was measured in 74 patients and uTWEAK in 69 patients, 42 of whom had kidney involvement. uTWEAK-to-creatinine ratio (uTWEAK/Cr) was significantly higher at baseline compared with follow-up (median 7.21 vs 4.94 ng/mmol, P < .0001). Patients with kidney involvement had higher uTWEAK/Cr levels compared with those without (P = .03). A correlation was found between uTWEAK/Cr and BVAS (P = .006), albuminuria (P = .022) and crescentic changes (P = .03). sTWEAK levels were higher in patients at inclusion than at follow-up (P = .009) but no difference was found when comparing patients and controls, nor did sTWEAK correlate with BVAS. IHC staining showed a clear expression of TWEAK but a fainter pattern of Fn14 in kidney biopsies from AAV patients. Conclusions:uTWEAK/Cr correlated with BVAS, albuminuria and number of crescents in active AAV and may be a useful biomarker in assessing disease activity in patients with AAV, whereas sTWEAK level is not.

Machine learning algorithms that integrate multiple biomarkers are increasingly used in disease detection, yet economic considerations are often overlooked. Medial vascular calcification (mVC), a pathology associated with elevated cardiovascular risk in chronic kidney disease (CKD), requires cost-effective diagnostic approaches. This pilot study evaluated the cost-effectiveness of machine learning models for mVC detection using traditional risk markers and circulating biomarkers in 152 CKD patients undergoing living donor kidney transplantation. Patients were classified as having no/minimal (n = 93) or moderate/extensive (n = 59) mVC. Five classification frameworks with automatic variable selection identified predictors of mVC. Age and copeptin were selected by all algorithms, while diabetes, male sex, choline, and osteoprotegerin were chosen by four methods. The number of features selected ranged from 5 to 21. Although accuracy differences among classifiers were limited to 3%, models using more features nearly tripled the procedure's cost. By incorporating the incremental cost-effectiveness ratio, the study highlighted significant disparities in performance versus cost among classifiers. The present findings suggest that machine learning has the potential to complement imaging techniques for mVC detection and uncover novel biomarkers. However, modest performance improvements may not justify higher costs, underscoring the importance of considering cost-effectiveness when selecting classification models.

KEY POINTS:Artificial intelligence models effectively generalized across studies and animal models and reduced translational gaps when applied to human biopsies. Artificial intelligence assistance reduced study evaluation turnaround times by up to 90% versus manual whole slide imaging scoring, matching expert-level performance. Self-supervised learning captured diabetic kidney disease-relevant features and mitigated expert-specific bias. BACKGROUND:Assessment of pathology end points in animal models of diabetic kidney disease is time-consuming and prone to expert bias. In addition, the sparsity of human kidney biopsy data hinders the development of translational models from animals to humans. METHODS:We developed an artificial intelligence (AI)-driven workflow to streamline histopathologic assessments in animal models of diabetic nephropathy. Our approach ( 1 ) detected glomeruli in whole slide images, ( 2 ) enabled fast expert scoring through an annotation tool, and ( 3 ) automated scoring. By leveraging unlabeled preclinical data for self-supervised learning, we enhanced AI scoring performance, reduced expert bias, and enabled the translation of AI scoring from animal models to human biopsies. To translate AI models from preclinical studies to human biopsies, we introduced a method that adjusted the feature extractor to human-specific features during inference without the need for annotated examples. RESULTS:Our annotation tool streamlined glomerular scoring, reducing turnaround time by 80%. Supervised AI models outperformed expert agreement and further reduced turnaround time by 90%, demonstrating generalization across studies involving both the same and different animal models. Without supervision, the self-supervised model achieved a κ value of 0.78, effectively identifying glomerular changes without guidance. Incorporating self-supervised learning into supervised training improved performance to κ=0.84 and reduced bias compared with individual experts ( P < 0.001). Our translational approach achieved a κ value of 0.63 on human glomeruli, although the model was trained exclusively on mouse glomeruli scores, reducing the translational gap by 45%. CONCLUSIONS:In this study, we accelerated and enhanced pathology readouts in a real-life pharmaceutical industry setting. We show that AI-assisted scoring reduced pathologists' workload and expedited study assessments. Self-supervised learning captured intrinsic properties of kidney morphology without expert annotation and reduced expert bias and translational discrepancies, greatly facilitating translational activities in drug development for patients with diabetic kidney disease.

Current prognostic models of CKD lack tubular injury biomarkers. Kidney injury molecule-1 (KIM-1) is undetectable in healthy kidneys but upregulated in the proximal tubule following injury. Acutely, this functions to clear cell debris; however, prolonged injury can trigger a proinflammatory cascade, leading to inflammation and fibrosis. Both plasma and urine KIM-1 have been studied separately in CKD. Plasma KIM-1 has shown a stronger association with CKD progression, which has been attributed to a greater burden of tubular injury. However, no study in CKD has jointly studied the two in combination to determine if they exhibit differing prognostic associations. The National Unified Renal Translational Research Enterprise (NURTuRE)-CKD is a UK, prospective multicentre cohort study of adults with non-dialysis CKD. Both of these biomarkers were measured in 2, 581 participants. Plasma and urinary KIM-1 was assessed as continuous variables and dichotomised above and below the median for each to produce four groups. The primary outcomes were: Kidney failure (KF) defined as the first incidence of eGFR < 15ml/min/1.73m2 or the initiation of kidney replacement therapy or transplantation. All-cause mortality (ACM) pre-KF Cox proportional hazards (PH) regression models were constructed to analyse the association of plasma and urine KIM-1 with time to KF and ACM. C-statistics and net reclassification improvement were used to assess the predictive performance of plasma and urine KIM-1 individually and in combination. For KF, multivariable models controlled for age, sex, baseline eGFR, uACR, and both biomarkers. The same modelling approach was followed for ACM, with the addition of diabetes, hypertension, and prior vascular disease. When dichotomised above and below the median, the ‘low plasma, low urine KIM-1’ group was used as the reference. Median eGFR was 34.8 ml/min/1.73 m2 and uACR 22.3 mg/mmol. There were 616 (23.8%) KF and 344 (13.3%) ACM events before KF. The median value for plasma KIM-1 was 250 (IQR 292) pg/ml; for urinary KIM-1, this was 178 (IQR 178) pg/mmol. Both negatively correlated with eGFR: plasma (Rho = −0.289, P < 0.001), and urine (Rho = −0.143, P < 0.001) and positively correlated with uACR: plasma (Rho = 0.426, P < 0.001) urine (Rho = 0.357, P < 0.001). In univariate analysis, both biomarkers were significantly associated with KF; however, in fully adjusted models, only pKIM-1 remained significant with an HR of 1.32 (1.16 – 1.50), while uKIM-1 was no longer significant with an HR of 0.92 (0.82–1.03). Individually, neither biomarker improved risk prediction for KF over established risk factors (eGFR, uACR, age, sex) however, when combined, there was a 24.1% net reclassification improvement. For ACM, both were significant in univariate analysis; however, in fully adjusted models, only uKIM-1 was significant with a HR of 1.57 (1.32–1.87), while pKIM-1 was not, HR 1.01 (0.85–1.19). The largest improvement in risk prediction over established risk factors was with urinary KIM-1, with a net reclassification improvement of 39.5% (no additional improvement was seen with the addition of pKIM-1 to uKIM-1 for ACM). KM survival curves for KF and ACM can be seen for the dichotomised KIM-1 groups (Fig. 1). Participants with both high plasma and urine KIM-1 experienced both events most frequently. These persisted after adjustments for covariates in the CoxPH model. In this study, elevated plasma KIM-1 was independently associated with an increased risk of kidney failure. In contrast, urinary KIM-1 was associated with an increased all-cause mortality risk, possibly suggesting distinct mechanistic associations. The combination of elevated plasma and urine KIM-1 levels together conferred the greatest risk of ACM and KF and led to improved reclassification. Although the mechanism by which urinary KIM-1 is associated with increased mortality, rather than KF, remains speculative, their combined measurement can enhance risk stratification in individuals with CKD, improve reclassification for kidney failure beyond established markers, and provide valuable insights into tubular health.

Background CKD carries a variable risk for multiple adverse outcomes, highlighting the need for a personalized approach. This study evaluated several novel biomarkers linked to key disease mechanisms to predict the risk of kidney failure (first event of eGFR <15 ml/min per 1.73 m(2) or KRT), all-cause mortality, and a composite of both. Methods We included 2884 adults with nondialysis CKD from 16 nephrology centers across the United Kingdom. Twenty-one biomarkers associated with kidney damage, fibrosis, inflammation, and cardiovascular disease were analyzed in urine, plasma, or serum. Cox proportional hazards models were used to assess biomarker associations and develop risk prediction models. Results Participants had mean age 63 (15) years; 58% were male and 87% White. Median eGFR was 35 (25-47) ml/min per 1.73 m2, and the median urinary albumin-to-creatinine ratio was 197 (32-895) mg/g. During median 48 (33-55) months of follow-up, 680 kidney failure events and 414 all-cause mortality events occurred. For kidney failure, a model combining three biomarkers (soluble TNF receptor 1, soluble cluster of differentiation 40, and urinary collagen type 1 alpha 1 chain) showed good discrimination (C-index, 0.86; 95% confidence interval [CI], 0.83 to 0.89) but was outperformed by a model using established risk factors (age, sex, ethnicity, eGFR, and urinary albumin-to-creatinine ratio; C-index, 0.90; 95% CI, 0.88 to 0.92). For all-cause mortality, a model using three biomarkers (high-sensitivity cardiac troponin T, N-terminal pro-brain natriuretic peptide, and soluble urokinase plasminogen activator receptor) demonstrated equivalent discrimination (C-index, 0.80; 95% CI, 0.75 to 0.84) to an established risk factor model (C-index, 0.80; 95% CI, 0.76 to 0.84). For the composite outcome, the biomarker model discrimination (C-index, 0.78; 95% CI, 0.76 to 0.81) was numerically higher than for established risk factors (C-index, 0.77; 95% CI, 0.74 to 0.80), and the addition of biomarkers to the established risk factors led to a small but statistically significant improvement in discrimination (C-index, 0.80; 95% CI, 0.77 to 0.82; P value <0.01). Conclusions Risk prediction models incorporating novel biomarkers showed comparable discrimination to established risk factors of kidney failure and all-cause mortality.Clinical Trial registry name and registration number:ClinicalTrials.gov, NCT04084145.

In renal histopathology, the routine clinical use of several histological stains presents challenges for the direct application of stain-specific deep learning-based analysis tools to whole-slide images. We present an approach to the in silico histological staining of kidney tissue where samples stained with hematoxylin and eosin (H&E) are virtually restained with periodic acid-Schiff (PAS). Our approach is underpinned by cycle-consistent generative adversarial neural networks trained on the National Unified Renal Translational Research Enterprise data set-the first UK-wide Biobank for chronic kidney disease-which features diverse data from 16 nephrology centers. Our work is divided into the following 4 main components: (1) we developed a virtual staining model, which infers PAS staining from H&E; (2) 2 board-certified pathologists assessed the virtual staining by attempting to distinguish it from real examples; (3) we trained a glomerular segmentation model using 3 independent renal segmentation data sets (Kidney Precision Medicine Project, Human BioMolecular Atlas Program [Kidney], and data by Jayapandian et al); and (4) we demonstrated the utility of virtual staining by inferring PAS staining from previously unseen H&E test images and applying our PAS-specific glomerular segmentation model. Each pathologist was able to identify 52.5% and 75.8% of the virtually stained images, respectively, showing an overlap in the variability of the authentic and synthetic staining. We discussed the utility of virtual staining in digital pathology, the need for pathology-specific testing with respect to chronic damage, and minimal changes and steps for incorporating more stains. Furthermore, alongside this article, we included complete glomerular annotations for 20 Kidney Precision Medicine Project H&E-stained slides.

Accurate segmentation of glomerulus instances attains high clinical significance in the automated analysis of renal biopsies to aid in diagnosing and monitoring kidney disease. Analyzing real-world histopathology images often encompasses inter-observer variability and requires a labor-intensive process of data annotation. Therefore, conventional supervised learning approaches generally achieve sub-optimal performance when applied to external datasets. Considering these challenges, we present a semi-supervised learning approach for glomeruli segmentation based on the weak-to-strong consistency framework validated on multiple real-world datasets. Our experimental results on 3 independent datasets indicate superior performance of our approach as compared with existing supervised baseline models such as U-Net and SegFormer.

Despite the recent advances in our understanding of the role of lipids, metabolites and related enzymes in mediating kidney injury, there is limited integrated multi-omics data identifying potential metabolic pathways driving impaired kidney function. The limited availability of kidney biopsies from living donors with acute kidney injury has remained a major constraint. Here, we validated the use of deceased transplant donor kidneys as a good model to study acute kidney injury in humans and characterized these kidneys using imaging and multi-omics approaches. We noted consistent changes in kidney injury and inflammatory markers in donors with reduced kidney function. Neighborhood and correlation analyses of imaging mass cytometry data showed that subsets of kidney cells (proximal tubular cells and fibroblasts) are associated with the expression profile of kidney immune cells, potentially linking these cells to kidney inflammation. Integrated transcriptomic and metabolomic analysis of human kidneys showed that kidney arachidonic acid metabolism and seven other metabolic pathways were upregulated following diminished kidney function. To validate the arachidonic acid pathway in impaired kidney function we demonstrated increased levels of cytosolic phospholipase A2 protein and related lipid mediators (prostaglandin E2) in the injured kidneys. Further, inhibition of cytosolic phospholipase A2 reduced injury and inflammation in human kidney proximal tubular epithelial cells in vitro. Thus, our study identified cell types and metabolic pathways that may be critical for controlling inflammation associated with impaired kidney function in humans.