Eiichiro Kanda

Chronic kidney disease (CKD) is a complex disease that is related not only to dialysis but also to the onset of cardiovascular disease and life prognosis. As renal function declines with age and depending on lifestyle, the number of patients with CKD is rapidly increasing in Japan. Accurate prognosis prediction for patients with CKD in clinical settings is important for selecting treatment methods and screening patients with high-risk. In recent years, big databases on CKD and dialysis have been constructed through the use of data science technology, and the pathology of CKD is being elucidated. Therefore, we developed an artificial intelligence (AI) system that can accurately predict the prognosis of CKD such as its progression, the timing of dialysis introduction, and death. Aiming for its social implementation, the prognosis prediction system developed for patients with CKD was released on the website. We then developed a clinical practice guideline creation support system called Doctor K as an AI system. When creating clinical practice guidelines, huge amounts of manpower and time are required to conduct a systematic review of thousands of papers. Therefore, we developed a natural language processing (NLP) AI system to significantly improve work efficiency. Doctor K was used in the preparation of the guidelines of the Japanese Society of Nephrology. Furthermore, by comparing and analyzing the medical word virtual space constructed by the NLP AI system based on patient big data, we proved using the latest mathematical theory (category theory) that this system reflects the pathology of CKD. This suggests the possibility that the NLP AI system can predict patient prognosis. We hope that, through these studies, the use of AI based on big data will lead to the development of new treatments and improvement in patient prognosis.

Introduction: The optimal time for vascular access (VA) creation remains controversial. Methods: We conducted a cohort study using data from the Japanese Society for Dialysis Therapy Renal Data Registry. Adult patients who started receiving hemodialysis in 2007 and had a permanent VA created were included. The exposure of interest was the timing of VA creation, categorized into three groups: early VA creation (defined as creation at least 4 months before hemodialysis initiation), just prior VA creation (creation between 1 and 3 months before hemodialysis initiation), and late VA creation (creation within 1 month of or after hemodialysis initiation). Cox regression analyses were used to compare 1-year all-cause mortality, with late VA creation as the reference group. Owing to the violations of the proportional hazards assumptions, the follow-up period was divided into “early” (1–4 months follow-up) and “late” (5–12 months follow-up) periods. Results: Overall, 1,280 (15.4%) of 8,322 patients died. Both early creation and just prior creation were associated with lower all-cause mortality in the early period compared with late creation. In the late period, the hazard ratios (HRs) for all-cause mortality decreased with earlier VA creation (adjusted HRs [95% confidence intervals]: 0.49 [0.35–0.67] for the early creation group and 0.63 [0.51–0.79] for the just prior creation group). Conclusion: Our study suggests that VA creation at least 1 month before hemodialysis initiation is associated with lower all-cause mortality in the early period, with earlier VA creation resulting in further mortality reduction in the late period.

Introduction: Chronic kidney disease (CKD) can have a profound impact on patients’ lives. However, multinational data on patients’ lived experience with CKD are scarce. Methods: Individuals from the prospective cohort of DISCOVER CKD (NCT04034992), an observational cohort study, were recruited to participate in one-to-one telephone interviews to explore their lived experience with CKD. A target of 100 participant interviews was planned across four countries (Japan, Spain, the UK, and the USA). These qualitative interviews, lasting ~60–90 min, were conducted in the local language by trained interviewers with specific experience in CKD, between January and June 2023. Transcribed interviews were translated into English for coding and analysis. Data were coded using qualitative research software. Results: Of the 105 participants interviewed, 103 were included in the final analysis. The average time since CKD diagnosis was 9.5 years, and at least half (50.5%) of participants had CKD Stage 3A or 3B. CKD diagnosis was an emotional experience, driven by worry (n = 29/103; 28.2%) and shock (n = 26/103; 25.2%), and participants often reported feeling inadequately informed. Additional information was frequently sought, either online or via other healthcare providers. The proportion of participants reporting no impacts of CKD on their lives was highest in those with CKD Stage 1 and 2 (64.3%). Conversely, every participant in the CKD Stage 5 on dialysis group reported some impact of CKD on their lives. Across all participants, the most reported impacts were anxiety or depression (37.9%) or ability to sleep (37.9%). The frequency of the reported impacts appeared to increase with disease severity, with the highest rates observed in the dialysis group. In that group, the most frequently reported impact was on the ability to work (80.0%). Conclusion: Findings from this multinational qualitative study suggest that patients may experience symptoms and signs of disease prior to diagnosis; however, these are often non-specific and may not be directly associated with CKD. Once diagnosed, the burden of CKD can have a diverse, negative impact on various aspects of patients’ lives. This highlights the need for early identification of at-risk individuals, and the importance of early CKD diagnosis and management with guideline-directed therapies to either prevent further deterioration of CKD or slow its progression, thus reducing symptom burden and improving quality of life.

There are cases in which CKD progression is difficult to evaluate, because the changes in estimated glomerular filtration rate (eGFR) and proteinuria sometimes show opposite directions as CKD progresses. Indices and models that enable the easy and accurate risk prediction of end-stage-kidney disease (ESKD) are indispensable to CKD therapy. In this study, we investigated whether a CKD stage coordinate transformed into a vector field (CKD potential model) accurately predicts ESKD risk. Meta-analysis of large-scale cohort studies of CKD patients in PubMed was conducted to develop the model. The distance from CKD stage G2 A1 to a patient’s data on eGFR and proteinuria was defined as r. We developed the CKD potential model on the basis of the data from the meta-analysis of three previous cohort studies: ESKD risk = exp(r). Then, the model was validated using data from a cohort study of CKD patients in Japan followed up for three years (n = 1,564). Moreover, the directional derivative of the model was developed as an index of CKD progression velocity. For ESKD prediction in three years, areas under the receiver operating characteristic curves (AUCs) were adjusted for baseline characteristics. Cox proportional hazards models with spline terms showed the exponential association between r and ESKD risk (ppp<0.0001). Then, a chart of the transformed CKD stage was developed for implementation in clinical settings. This study indicated that the transformed CKD stage as a vector field enables the easy and accurate estimation of ESKD risk and CKD progression and suggested that vector analysis is a useful tool for clinical studies of CKD and its related diseases.

A new marker reflecting the pathophysiology of chronic kidney disease (CKD) has been desired for its therapy. In this study, we developed a virtual space where data in medical words and those of actual CKD patients were unified by natural language processing and category theory. A virtual space of medical words was constructed from the CKD-related literature (n = 165,271) using Word2Vec, in which 106,612 words composed a network. The network satisfied vector calculations, and retained the meanings of medical words. The data of CKD patients of a cohort study for 3 years (n = 26,433) were transformed into the network as medical-word vectors. We let the relationship between vectors of patient data and the outcome (dialysis or death) be a marker (inner product). Then, the inner product accurately predicted the outcomes: C-statistics of 0.911 (95% CI 0.897, 0.924). Cox proportional hazards models showed that the risk of the outcomes in the high-inner-product group was 21.92 (95% CI 14.77, 32.51) times higher than that in the low-inner-product group. This study showed that CKD patients can be treated as a network of medical words that reflect the pathophysiological condition of CKD and the risks of CKD progression and mortality.

Introduction: Vascular access usage varies widely across countries. Previous studies have evaluated the association of clinical outcomes with the three types of vascular access, namely, arteriovenous fistula (AVF), arteriovenous graft (AVG), and tunneled and cuffed central venous catheter (TC-CVC). However, little is known regarding the association between arterial superficialization (AS) and the mortality of patients. Methods: A nationwide cohort study was conducted using data from the Japanese Society for Dialysis Therapy Renal Data Registry (2006–2007). We included patients aged ≥20 years undergoing hemodialysis with a dialysis vintage ≥6 months. The exposures of interest were the four types of vascular access: AVF, AVG, AS, and TC-CVC. Cox proportional hazard models were used to evaluate the associations of vascular access types with 1-year all-cause and cause-specific mortality. Results: A total of 183,490 maintenance hemodialysis patients were included: 90.7% with AVF, 6.9% with AVG, 2.0% with AS, and 0.4% with TC-CVC. During the 1-year follow-up period, 13,798 patients died. Compared to patients with AVF, those with AVG, AS, and TC-CVC had a significantly higher risk of all-cause mortality after adjustment for confounding factors: adjusted hazard ratios (95% confidence intervals) – 1.30 (1.20–1.41), 1.56 (1.39–1.76), and 2.15 (1.77–2.61), respectively. Similar results were obtained for infection-related and cardiovascular mortality. Conclusion: This nationwide cohort study conducted in Japan suggested that AVF usage may have the lowest risk of all-cause mortality. The study also suggested that the usage of AS may be associated with better survival rates compared to those of TC-CVC in patients who are not suitable for AVF or AVG.

Background: Hyperkalemia (HK) is a barrier to optimization of renin-angiotensin-aldosterone system inhibitor (RAASi) therapy in heart failure (HF) and chronic kidney disease (CKD). We investigated cardiorenal risk associated with changes in RAASi regimen after an episode of HK in patients with HF and/or CKD. Methods: This observational study utilized data from hospital records, claims, and health registers from the US (Optum’s de-identified Market Clarity Data) and Japan (Medical Data Vision). Included patients had an index episode of HK between July 2019 and September 2021 (US), or May 2020 and September 2021 (Japan), with prior diagnosis of HF or CKD (stage 3 or 4), and RAASi use. Risk of a cardiorenal composite outcome (HF emergency visit, HF hospitalization, or progression to end-stage kidney disease) was determined in patients who discontinued RAASi, down-titrated their dose by > 25%, or maintained or up-titrated their dose following the HK episode. Results: A total of 15,488 and 6020 patients were included from the US and Japan, respectively. Prior to the episode of HK, 59% (US) and 27% (Japan) of patients had achieved > 50% target RAASi dose. Following the episode of HK, 33% (US) and 32% (Japan) of patients did not fill a new RAASi prescription. Risk of the cardiorenal outcome at 6 months was higher in patients who discontinued or down-titrated versus maintained or up-titrated RAASi treatment both in the US (17.5, 18.3, and 10.6%; p < 0.001) and in Japan (19.7, 20.0, and 15.1%; p < 0.001). Conclusion: HK-related RAASi discontinuation or down-titration was associated with higher risk of cardiorenal events versus maintained or up-titrated RAASi.

Chronic kidney disease (CKD) patients have high risks of end-stage kidney disease (ESKD) and pre-ESKD death. Therefore, accurately predicting these outcomes is useful among CKD patients, especially in those who are at high risk. Thus, we evaluated whether a machine-learning system can predict accurately these risks in CKD patients and attempted its application by developing a Web-based risk-prediction system. We developed 16 risk-prediction machine-learning models using Random Forest (RF), Gradient Boosting Decision Tree, and eXtreme Gradient Boosting with 22 variables or selected variables for the prediction of the primary outcome (ESKD or death) on the basis of repeatedly measured data of CKD patients (n = 3,714; repeatedly measured data, n = 66,981) in their electronic-medical records. The performances of the models were evaluated using data from a cohort study of CKD patients carried out over 3 years (n = 26,906). One RF model with 22 variables and another RF model with 8 variables of time-series data showed high accuracies of the prediction of the outcomes and were selected for use in a risk-prediction system. In the validation, the 22- and 8-variable RF models showed high C-statistics for the prediction of the outcomes: 0.932 (95% CI 0.916, 0.948) and 0.93 (0.915, 0.945), respectively. Cox proportional hazards models using splines showed a highly significant relationship between the high probability and high risk of an outcome (p<0.0001). Moreover, the risks of patients with high probabilities were higher than those with low probabilities: 22-variable model, hazard ratio of 104.9 (95% CI 70.81, 155.3); 8-variable model, 90.9 (95% CI 62.29, 132.7). Then, a Web-based risk-prediction system was actually developed for the implementation of the models in clinical practice. This study showed that a machine-learning-based Web system is a useful tool for the risk prediction and treatment of CKD patients.

The munber of chronic kidney disease (CKD) patients is increasing globally because kidney function is affected by aging and lifestyle habits. Malnutrition, muscle weakness, and a decline in activities of daily living (ADL) are often observed in elderly CKD patients and dialysis patients, and are related to their CKD prognosis and life prognoses. Chronic inflammation and atherosclerotic disease are associated with malnutrition. Because malnutrition and its related factors affect patients’ prognoses, it is necessary to identify and treat patients with malnutrition at an early stage. The state in which the stored protein and energy sources are reduced in CKD is called protein energy wasting (PEW). PEW is diagnosed on the basis of biochemical test findings such as hypoalbuminemia, unhealthy physique, and decreased muscle mass and dietary intake. For evaluating PEW, a complex nutritional index taking into account the pathophysiology specific to CKD patients is useful. Not only nutritional therapy but also exercise therapy is necessary to stop the vicious cycle associated with PEW and the decline in ADL.

Hyperkalemia is associated with increased risks of mortality and adverse clinical outcomes. The treatment of hyperkalemia often leads to the discontinuation or restriction of beneficial but potassium-increasing therapy such as renin-angiotensin-aldosterone inhibitors (RAASi) and high-potassium diet including fruits and vegetables. To date, limited evidence is available for personalized risk evaluation in this heterogeneous and multifactorial pathophysiological condition. We developed risk prediction models using extreme gradient boosting (XGB), multiple logistic regression (LR), and deep neural network. Models were derived from a retrospective cohort of hyperkalemic patients with either heart failure or chronic kidney disease stage ≥3a from a Japanese nationwide database (1 April 2008–30 September 2018). Studied outcomes included all-cause death, renal replacement therapy introduction (RRT), hospitalization for heart failure (HHF), and cardiovascular events within three years after hyperkalemic episodes. The best performing model was further validated using an external cohort. A total of 24,949 adult hyperkalemic patients were selected for model derivation and internal validation. A total of 1452 deaths (16.6%), 887 RRT (10.1%), 1,345 HHF (15.4%), and 621 cardiovascular events (7.1%) were observed. XGB outperformed other models. The area under receiver operator characteristic curves (AUROCs) of XGB vs. LR (95% CIs) for death, RRT, HHF, and cardiovascular events were 0.823 (0.805–0.841) vs. 0.809 (0.791–0.828), 0.957 (0.947–0.967) vs. 0.947 (0.936–0.959), 0.863 (0.846–0.880) vs. 0.838 (0.820–0.856), and 0.809 (0.784–0.834) vs. 0.798 (0.772–0.823), respectively. In the external dataset including 86,279 patients, AUROCs (95% CIs) for XGB were: death, 0.747 (0.742–0.753); RRT, 0.888 (0.882–0.894); HHF, 0.673 (0.666–0.679); and cardiovascular events, 0.585 (0.578–0.591). Kaplan–Meier curves of the high-risk predicted group showed a statistically significant difference from that of the low-risk predicted groups for all outcomes (p < 0.005; log-rank test). These findings suggest possible use of machine learning models for real-world risk assessment as a guide for observation and/or treatment decision making that may potentially lead to improved outcomes in hyperkalemic patients while retaining the benefit of life-saving therapies.