Clustering-based spatial analysis (CluSA) framework through graph neural network for chronic kidney disease prediction using histopathology images.

Machine learning applied to digital pathology has been increasingly used to assess kidney function and diagnose the underlying cause of chronic kidney disease (CKD). We developed a novel computational framework, clustering-based spatial analysis (CluSA), that leverages unsupervised learning to learn spatial relationships between local visual patterns in kidney tissue. This framework minimizes the need for time-consuming and impractical expert annotations. 107,471 histopathology images obtained from 172 biopsy cores were used in the clustering and in the deep learning model. To incorporate spatial information over the clustered image patterns on the biopsy sample, we spatially encoded clustered patterns with colors and performed spatial analysis through graph neural network. A random forest classifier with various groups of features were used to predict CKD. For predicting eGFR at the biopsy, we achieved a sensitivity of 0.97, specificity of 0.90, and accuracy of 0.95. AUC was 0.96. For predicting eGFR changes in one-year, we achieved a sensitivity of 0.83, specificity of 0.85, and accuracy of 0.84. AUC was 0.85. This study presents the first spatial analysis based on unsupervised machine learning algorithms. Without expert annotation, CluSA framework can not only accurately classify and predict the degree of kidney function at the biopsy and in one year, but also identify novel predictors of kidney function and renal prognosis.

Unsupervised machine learning for identifying important visual features through bag-of-words using histopathology data from chronic kidney disease.

Pathologists use visual classification to assess patient kidney biopsy samples when diagnosing the underlying cause of kidney disease. However, the assessment is qualitative, or semi-quantitative at best, and reproducibility is challenging. To discover previously unknown features which predict patient outcomes and overcome substantial interobserver variability, we developed an unsupervised bag-of-words model. Our study applied to the C-PROBE cohort of patients with chronic kidney disease (CKD). 107,471 histopathology images were obtained from 161 biopsy cores and identified important morphological features in biopsy tissue that are highly predictive of the presence of CKD both at the time of biopsy and in one year. To evaluate the performance of our model, we estimated the AUC and its 95% confidence interval. We show that this method is reliable and reproducible and can achieve 0.93 AUC at predicting glomerular filtration rate at the time of biopsy as well as predicting a loss of function at one year. Additionally, with this method, we ranked the identified morphological features according to their importance as diagnostic markers for chronic kidney disease. In this study, we have demonstrated the feasibility of using an unsupervised machine learning method without human input in order to predict the level of kidney function in CKD. The results from our study indicate that the visual dictionary, or visual image pattern, obtained from unsupervised machine learning can predict outcomes using machine-derived values that correspond to both known and unknown clinically relevant features.

Continuous Renal Replacement Therapy among Patients with COVID-19 and Acute Kidney Injury.

BACKGROUND AND OBJECTIVES: Acute kidney injury (AKI) is a common complication among patients with COVID-19 and acute respiratory distress syndrome. Reports suggest that COVID-19 confers a pro-thrombotic state, which presents challenges in maintaining hemofilter patency and delivering continuous renal replacement therapy (CRRT). We present our initial experience with CRRT in critically ill patients with COVID-19, emphasizing circuit patency and the association between fluid balance during CRRT and respiratory parameters. DESIGN, SETTING, PARTICIPANTS, AND MEASUREMENTS: Retrospective chart review of 32 consecutive patients with COVID-19 and AKI managed with continuous venovenous hemodiafiltration with regional citrate anticoagulation (CVVHDF-RCA) according to the University of Michigan protocol. Primary outcome was mean CRRT circuit life per patient during the first 7 days of CRRT. We used simple linear regression to assess the relationship between patient characteristics and filter life. We also explored the relationship between fluid balance on CRRT and respiratory parameters using repeated measures modeling. RESULTS: Patients' mean age was 54.8 years and majority were Black (75%). Comorbidities included hypertension (90.6%), obesity (70.9%) diabetes (56.2%), and chronic kidney disease (40.6%). Median CRRT circuit life was 53.5 [interquartile range 39.1-77.6] hours. There was no association between circuit life and inflammatory or pro-thrombotic laboratory values (ferritin p = 0.92, C-reactive protein p = 0.29, D-dimer p = 0.24), or with systemic anticoagulation (p = 0.37). Net daily fluid removal during the first 7 days of CRRT was not associated with daily (closest recorded values to 20:00) PaO2/FIO2 ratio (p = 0.21) or positive end-expiratory pressure requirements (p = 0.47). CONCLUSIONS: We achieved adequate CRRT circuit life in COVID-19 patients using an established CVVHDF-RCA protocol. During the first 7 days of CRRT therapy, cumulative fluid balance was not associated with improvements in respiratory parameters, even after accounting for baseline fluid balance.

Management of Acute Kidney Injury/Renal Replacement Therapy in the Intensive Care Unit.

Common causes of acute kidney injury (AKI) in the ICU setting include acute tubular necrosis (due to shock, hemolysis, rhabdomyolysis, or procedures that compromise renal perfusion), abdominal compartment syndrome, urinary retention, and interstitial nephritis. Treatment is geared toward addressing the underlying cause. Dialysis may be required if renal injury does not resolve. Early initiation of dialysis based on the stage of AKI alone has not been shown to provide a mortality benefit. Dialysis modalities are based on the dialysis indication and the patient's clinical status. Providers should pay close attention to nutritional requirements and medication dosing according to renal function and dialysis modality.

Current Understanding of Clinical Manifestations of COVID-19 in Glomerular Disease.

Background: The novel coronavirus disease (COVID-19), also known as severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), is an evolving pandemic with significant mortality. Information about the impact of infection on glomerular disease patients in particular has been lacking. Understanding the virus's effect in glomerular disease is constantly changing. This review article summarizes the data published thus far on COVID-19 and its manifestations in pre-existing and de novo glomerular disease. Summary: While patients with glomerular disease may be at higher risk of severe COVID-19 due to their immunosuppressed status, some data suggest that a low amount of immunosuppression may be helpful in mitigating the systemic inflammatory response which is associated with high mortality rates in COVID-19. There have been a few case reports on COVID-19 causing glomerular disease relapse in patients. Multiple mechanisms have been proposed for kidney injury, proteinuria, and hematuria in the setting of COVID-19. More commonly, these are caused by direct tubular injury due to hemodynamic instability and hypoxic injury. However, the cytokine storm induced by COVID-19 may trigger common post-viral glomerular disease such as IgA nephropathy, anti-GBM, and ANCA vasculitis that have also been described in COVID-19 patients. Collapsing glomerulopathy, a hallmark of HIV-associated nephropathy, is being reported SARS-CoV-2 cases, particularly in patients with high-risk APOL1 alleles. Direct viral invasion of glomerular structures is hypothesized to cause a podocytopathy due to virus's affinity to ACE2, but evidence for this remains under study. Key Messages: Infection with SARS-CoV-2 may cause glomerular disease in certain patients. The mechanism of de novo glomerular disease in the setting of COVID-19 is under study. The management of patients with existing glomerular disease poses unique challenges, especially with regard to immunosuppression management. Further studies are needed to inform clinician decisions about the management of these patients during the COVID-19 pandemic.

Deployment of a New CRRT/PIRRT Device during the COVID-19 Pandemic Emergency: Organizational Challenges and Implementation Results.

INTRODUCTION: The coronavirus disease 2019 (COVID-19) pandemic led to increased demand nationwide for dialysis equipment, including supplies and machines. To meet the demand in our institution, our surge plan included rapid mobilization of a novel continuous renal replacement treatment (CRRT) machine named SAMI. The SAMI is a push-pull filtration enhanced dialysis machine that can conjugate extremely high single-pass solute removal efficiency with very precise fluid balance control. MATERIAL AND METHODS: Machine assembly was conducted on-site by local biomedical engineers with remote assistance by the vendor. One 3-h virtual training session of 3 dialysis nurses was conducted before SAMI deployment. The SAMI was deployed in prolonged intermittent replacement therapy (PIRRT) mode to maximize patients covered per machine per day. Live on-demand vendor support was provided to troubleshoot any issues for the first few cases. After 4 weeks of the SAMI implementation, data on treatments with the SAMI were collected, and a questionnaire was provided to the nurse trainees to assess device usability. RESULTS: On-site installation of the SAMI was accomplished with remote assistance. Delivery of remote training was successfully achieved. 23 PIRRT treatments were conducted in 10 patients. 7/10 of patients had CO-VID-19. The median PIRRT dose was 50 mL/kg/h (IQR [interquartile range] 44 - 62 mL/kg/h), and duration of the treatment was 8 h (IQR 6.3 - 8 h). Solute control was adequate. The user response was favorable to the set of usability questions involving user interface, on-screen instructions, machine setup, troubleshooting, and the ease of moving the machine. CONCLUSION: Assembly of the SAMI and training of nurses remotely are possible when access to vendor employees is restricted during states of emergency. The successful deployment of the SAMI in our institution during the pandemic with only 3-h virtual training supports that operating the SAMI is simple and safe.