The Discordance Between the Renal Histopathology and Clinical Presentation of Diabetic Nephropathy Calls for Novel Approaches for the Prediction and Monitoring of Kidney Failure in Diabetes

Circulating Activin A, Kidney Fibrosis, and Adverse Events

BACKGROUND

Identification of reliable biomarkers to assess kidney fibrosis severity is necessary for patients with CKD. Activin A, a member of the TGF- β superfamily, has been suggested as a biomarker for kidney fibrosis. However, its precise utility in this regard remains to be established.

METHODS

We investigated the correlation between plasma activin A levels, kidney fibrosis severity, and the incidence of major adverse kidney events in patients who underwent native kidney biopsies at a tertiary medical center. We performed RNA sequencing and histological analyses on kidney biopsy specimens to assess activin A expression. In vitro experiments were also conducted to explore the potential attenuation of TGF- β -induced fibroblast activation through activin A inhibition.

RESULTS

A total of 339 patients with biopsy-confirmed kidney diseases were enrolled. Baseline eGFR was 36 ml/min per 1.73 m 2 , and the urine protein/creatinine ratio was 2.9 mg/mg. Multivariable logistic regression analysis revealed a significant association between plasma activin A levels and the extent of tubulointerstitial fibrosis. Our RNA sequencing data demonstrated a positive correlation between kidney INHBA expression and plasma activin A levels. Furthermore, the histological analysis showed that myofibroblasts were the primary activin A-positive interstitial cells in diseased kidneys. During a median follow-up of 22 months, 113 participants experienced major adverse kidney events. Cox proportional hazards analysis initially found a positive association between plasma activin A levels and kidney event risk, but it became insignificant after adjusting for confounders. In cultured fibroblasts, knockdown of activin A significantly attenuated TGF- β -induced fibroblast-myofibroblast conversion.

CONCLUSIONS

Plasma activin A levels correlate with kidney fibrosis severity and adverse outcomes in various kidney disorders.

Spatially Aware Transformer Networks for Contextual Prediction of Diabetic Nephropathy Progression from Whole Slide Images

Diabetic nephropathy (DN) in the context of type 2 diabetes is the leading cause of end-stage renal disease (ESRD) in the United States. DN is graded based on glomerular morphology and has a spatially heterogeneous presentation in kidney biopsies that complicates pathologists' predictions of disease progression. Artificial intelligence and deep learning methods for pathology have shown promise for quantitative pathological evaluation and clinical trajectory estimation; but, they often fail to capture large-scale spatial anatomy and relationships found in whole slide images (WSIs). In this study, we present a transformer-based, multi-stage ESRD prediction framework built upon nonlinear dimensionality reduction, relative Euclidean pixel distance embeddings between every pair of observable glomeruli, and a corresponding spatial self-attention mechanism for a robust contextual representation. We developed a deep transformer network for encoding WSI and predicting future ESRD using a dataset of 56 kidney biopsy WSIs from DN patients at Seoul National University Hospital. Using a leave-one-out cross-validation scheme, our modified transformer framework outperformed RNNs, XGBoost, and logistic regression baseline models, and resulted in an area under the receiver operating characteristic curve (AUC) of 0.97 (95% CI: 0.90-1.00) for predicting two-year ESRD, compared with an AUC of 0.86 (95% CI: 0.66-0.99) without our relative distance embedding, and an AUC of 0.76 (95% CI: 0.59-0.92) without a denoising autoencoder module. While the variability and generalizability induced by smaller sample sizes are challenging, our distance-based embedding approach and overfitting mitigation techniques yielded results that sugest opportunities for future spatially aware WSI research using limited pathology datasets.

Spatially Aware Transformer Networks for Contextual Prediction of Diabetic Nephropathy Progression from Whole Slide Images

Diabetic nephropathy (DN) in the context of type 2 diabetes is the leading cause of end-stage renal disease (ESRD) in the United States. DN is graded based on glomerular morphology and has a spatially heterogeneous presentation in kidney biopsies that complicates pathologists' predictions of disease progression. Artificial intelligence and deep learning methods for pathology have shown promise for quantitative pathological evaluation and clinical trajectory estimation; but, they often fail to capture large-scale spatial anatomy and relationships found in whole slide images (WSIs). In this study, we present a transformer-based, multi-stage ESRD prediction framework built upon nonlinear dimensionality reduction, relative Euclidean pixel distance embeddings between every pair of observable glomeruli, and a corresponding spatial self-attention mechanism for a robust contextual representation. We developed a deep transformer network for encoding WSI and predicting future ESRD using a dataset of 56 kidney biopsy WSIs from DN patients at Seoul National University Hospital. Using a leave-one-out cross-validation scheme, our modified transformer framework outperformed RNNs, XGBoost, and logistic regression baseline models, and resulted in an area under the receiver operating characteristic curve (AUC) of 0.97 (95% CI: 0.90-1.00) for predicting two-year ESRD, compared with an AUC of 0.86 (95% CI: 0.66-0.99) without our relative distance embedding, and an AUC of 0.76 (95% CI: 0.59-0.92) without a denoising autoencoder module. While the variability and generalizability induced by smaller sample sizes are challenging, our distance-based embedding approach and overfitting mitigation techniques yielded results that suggest opportunities for future spatially aware WSI research using limited pathology datasets.

Correlation between Glucose/C-Peptide Ratio and the Risk of Disease Progression in Diabetic Nephropathy Patients: A Clinical Retrospective Analysis

The aim of this study is to analyze the correlation between the glucose/C-peptide ratio and the risk of disease progression in patients with diabetic nephropathy. Ninety-three patients with diabetic nephropathy, who were treated in the Chun'an Branch of Zhejiang Provincial People's Hospital, China, from January 2016 to January 2019, were recruited as subjects. In accordance with the disease progression, the patients were divided into a progression group (n = 59) and a nonprogression group (n = 34). Clinical data were compared between the two groups. Pearson's correlation was applied to analyze the correlation of age, postprandial glucose/C-peptide, glycosylated hemoglobin, insulin resistance index, serum cystatin C, uric acid, 24 h urinary albumin excretion rate (24 hUAER), and estimated glomerular filtration rate (eGFR). Univariate and multivariate logistic regression models were utilized to analyze the influencing factors for the risk of disease progression in patients with diabetic nephropathy. The receiver operating characteristic (ROC) curve was employed to assess the predictive value of postprandial glucose/C-peptide on the risk of disease progression in patients with diabetic nephropathy. Results. The age differences, postprandial glucose/C-peptide, glycosylated hemoglobin, insulin resistance index, serum cystatin C, uric acid, 24 hUAER, and eGFR were significantly different between the two groups (P < 0.05). Pearson's linear correlation analysis exhibited that postprandial glucose/C-peptide, insulin resistance index, serum cystatin C, and uric acid were positively correlated with 24 hUAER (r = 0.514, 0.345, 0.311, 0.279, P < 0.05). Age, postprandial glucose/C-peptide, insulin resistance index, serum cystatin C, and uric acid were negatively correlated with eGFR (r = -0.210, -0.610, -0.351, -0.347, and -0.274, P < 0.05). Univariate logistic regression analysis displayed that age (OR = 0.938; P=0.043), postprandial glucose/C-peptide (OR = 0.851; p ≤ 0.001), insulin resistance index (OR = 0.219; p ≤ 0.001), serum cystatin C (OR = 0.113; p ≤ 0.001), and uric acid (OR = 0.989; P=0.001) were risk factors for the risk of disease progression in patients with diabetic nephropathy. Multivariate logistic regression analysis exhibited that postprandial glucose/C-peptide (OR = 0.747; P=0.004), insulin resistance index (OR = 0.072; P=0.012), serum cystatin C (OR = 0.023; P=0.020), and uric acid (OR = 0.967; P=0.039) were independent risk factors for the risk of disease progression in patients with diabetic nephropathy. The ROC curve results demonstrated that the AUC of postprandial glucose/C-peptide predicting the risk of disease progression in patients with diabetic nephropathy was 0.931. Postprandial glucose/C-peptide, insulin resistance index, serum cystatin C, and uric acid are correlated with 24hUAER and eGFR. Postprandial glucose/C-peptide, insulin resistance index, serum cystatin C, and uric acid are independent risk factors for the risk of disease progression in patients with diabetic nephropathy. Among them, postprandial glucose/C-peptide can be employed as a crucial indicator to predict the risk of disease progression in diabetic nephropathy patients.