Our paper 20 years later: from acute renal failure to acute kidney injury—the metamorphosis of a syndrome

Machine learning-based risk prediction of acute kidney disease and hospital mortality in older patients

Introduction

Acute kidney injury (AKI) is a prevalent complication in older people, elevating the risks of acute kidney disease (AKD) and mortality. AKD reflects the adverse events developing after AKI. We aimed to develop and validate machine learning models for predicting the occurrence of AKD, AKI and mortality in older patients.

Methods

We retrospectively reviewed the medical records of older patients (aged 65 years and above). To explore the trajectory of kidney dysfunction, patients were categorized into four groups: no kidney disease, AKI recovery, AKD without AKI, or AKD with AKI. We developed eight machine learning models to predict AKD, AKI, and mortality. The best-performing model was identified based on the area under the receiver operating characteristic curve (AUC) and interpreted using the Shapley additive explanations (SHAP) method.

Results

A total of 22,005 patients were finally included in our study. Among them, 4,434 patients (20.15%) developed AKD, 4,000 (18.18%) occurred AKI, and 866 (3.94%) patients deceased. Light gradient boosting machine (LGBM) outperformed in predicting AKD, AKI, and mortality, and the final lite models with 15 features had AUC values of 0.760, 0.767, and 0.927, respectively. The SHAP method revealed that AKI stage, albumin, lactate dehydrogenase, aspirin and coronary heart disease were the top 5 predictors of AKD. An online prediction website for AKD and mortality was developed based on the final models.

Discussion

The LGBM models provide a valuable tool for early prediction of AKD, AKI, and mortality in older patients, facilitating timely interventions. This study highlights the potential of machine learning in improving older adult care, with the developed online tool offering practical utility for healthcare professionals. Further research should aim at external validation and integration of these models into clinical practice.

Internal and external validation of machine learning-assisted prediction models for mechanical ventilation-associated severe acute kidney injury

BACKGROUND

Currently, very few preventive or therapeutic strategies are used for mechanical ventilation (MV)-associated severe acute kidney injury (AKI).

OBJECTIVES

We developed clinical prediction models to detect the onset of severe AKI in the first week of intensive care unit (ICU) stay during the initiation of MV.

METHODS

A large ICU database Medical Information Mart for Intensive Care IV (MIMIC-IV) was analysed retrospectively. Data were collected from the clinical information recorded at the time of ICU admission and during the initial 12 h of MV. Using univariate and multivariate analyses, the predictors were selected successively. For model development, two machine learning algorithms were compared. The primary goal was to predict the development of AKI stage 2 or 3 (AKI-23) and AKI stage 3 (AKI-3) in the first week of patients' ICU stay after initial 12 h of MV. The developed models were externally validated using another multicentre ICU database (eICU Collaborative Research Database, eICU) and evaluated in various patient subpopulations.

RESULTS

Models were developed using data from the development cohort (MIMIC-IV: 2008-2016; n = 3986); the random forest algorithm outperformed the logistic regression algorithm. In the internal (MIMIC-IV: 2017-2019; n = 1210) and external (eICU; n = 1494) validation cohorts, the incidences of AKI-23 were 154 (12.7%) and 119 (8.0%), respectively, with areas under the receiver operator characteristic curve of 0.78 (95% confidence interval [CI]: 0.74-0.82) and 0.80 (95% CI: 0.76-0.84); the incidences of AKI-3 were 81 (6.7%) and 67 (4.5%), with areas under the receiver operator characteristic curve of 0.81 (95% CI: 0.76-0.87) and 0.80 (95% CI: 0.73-0.86), respectively.

CONCLUSIONS

Models driven by machine learning and based on routine clinical data may facilitate the early prediction of MV-associated severe AKI. The validated models can be found at: https://apoet.shinyapps.io/mv_aki_2021_v2/.

SENP1 protects cisplatin-induced AKI by attenuating apoptosis through regulation of HIF-1α

BACKGROUND

Acute kidney injury is a clinical syndrome with both high morbidity and mortality. However, the underlying molecular mechanism of AKI is still largely unknown. The role of SENP1 in AKI is unclear, while one of its substrates, HIF-1α possesses nephroprotective effect in AKI. Herein, this study aimed to reveal the role of SENP1/HIF-1α axis in AKI by using both cell and animal models.

METHODS

We investigated the effects of AKI on SENP1 expression using clinical samples, and cisplatin-induced AKI model based on mice or HK-2 cells. The influence of SENP1 knockdown or over-expression on cisplatin-induced AKI was studied in vitro and in vivo. Following the exploration of the change in HIF-1α expression brought by AKI, the synergistic effects of SENP1 knockdown and HIF-1α over-expression on AKI were examined.

RESULTS

The results showed the up-regulation of SENP1 in clinical specimens, as well as cell and animal models. The knockdown or over-expression of SENP1 in HK-2 cells could promote or inhibit AKI through regulating cell apoptosis, respectively. Moreover, SENP1^+/- mice suffered from much more serious AKI compared with mice in wild type group. Furthermore, we found that HIF-1α over-expression could attenuate the promoted cell apoptosis as well as AKI induced by SENP1 knockdown.

CONCLUSIONS

we showed that SENP1 provided protection for kidney in AKI via regulating cell apoptosis and through the regulation of HIF-1α. This study could benefit for the understanding of the pathogenesis of AKI and provide potential therapeutic target for AKI treatment.

Acute kidney injury and aging

Our aging population is growing and developing treatments for age-related diseases such as Alzheimer's and Parkinson's disease has taken on an increasing urgency and is accompanied by high public awareness. The already high and rising incidence of acute kidney injury (AKI) in the elderly, however, has received relatively little attention despite the potentially fatal outcomes associated with an episode of AKI in this age group. When discussing AKI and aging, one should consider two aspects: first, elderly patients have an increased susceptibility to an AKI episode, and second, they have decreased kidney repair after AKI given the high incidence of progression to chronic kidney disease (CKD). It is unclear if the same factors that drive the increased susceptibility to AKI could be playing a role in the decreased repair capacity or if they are totally different and unrelated. This review will examine current knowledge on the risk factors for the increased susceptibility to AKI in the elderly and will also explore potential aspects that might contribute to a decreased kidney repair response in this age group.

Learning Latent Space Representations to Predict Patient Outcomes: Model Development and Validation

Background

Scalable and accurate health outcome prediction using electronic health record (EHR) data has gained much attention in research recently. Previous machine learning models mostly ignore relations between different types of clinical data (ie, laboratory components, International Classification of Diseases codes, and medications).

Objective

This study aimed to model such relations and build predictive models using the EHR data from intensive care units. We developed innovative neural network models and compared them with the widely used logistic regression model and other state-of-the-art neural network models to predict the patient’s mortality using their longitudinal EHR data.

Methods

We built a set of neural network models that we collectively called as long short-term memory (LSTM) outcome prediction using comprehensive feature relations or in short, CLOUT. Our CLOUT models use a correlational neural network model to identify a latent space representation between different types of discrete clinical features during a patient’s encounter and integrate the latent representation into an LSTM-based predictive model framework. In addition, we designed an ablation experiment to identify risk factors from our CLOUT models. Using physicians’ input as the gold standard, we compared the risk factors identified by both CLOUT and logistic regression models.

Results

Experiments on the Medical Information Mart for Intensive Care-III dataset (selected patient population: 7537) show that CLOUT (area under the receiver operating characteristic curve=0.89) has surpassed logistic regression (0.82) and other baseline NN models (<0.86). In addition, physicians’ agreement with the CLOUT-derived risk factor rankings was statistically significantly higher than the agreement with the logistic regression model.

Conclusions

Our results support the applicability of CLOUT for real-world clinical use in identifying patients at high risk of mortality.

The Sick Adult Horse: Renal Clinical Pathologic Testing and Urinalysis

Clinicopathologic evaluation of renal function and renal disease in sick adult horses remains grounded in detection of azotemia, assessment of serum and urine electrolyte concentrations, and evaluation of urinalysis findings, including specific gravity, reagent strip analysis, and sediment examination. Because increases in serum or plasma urea nitrogen and creatinine concentrations are insensitive indicators of a decreased glomerular filtration rate, there is considerable interest in identifying novel biomarkers of renal function or injury in blood and urine, with serum symmetric dimethylarginine concentration being the most recent addition to the commercial market.

Does Beta-Trace Protein (BTP) Outperform Cystatin C as a Diagnostic Marker of Acute Kidney Injury Complicating the Early Phase of Acute Pancreatitis?

Acute pancreatitis (AP) belongs to the commonest acute gastrointestinal conditions requiring hospitalization. Acute kidney injury (AKI) often complicates moderately severe and severe AP, leading to increased mortality. Among the laboratory markers proposed for early diagnosis of AKI, few have been studied in AP, including cystatin C and neutrophil gelatinase-associated lipocalin (NGAL). Beta-trace protein (BTP), a low-molecular-weight glycoprotein proposed as an early marker of decreased glomerular filtration, has never been studied in AP. We investigated the diagnostic usefulness of serum BTP for early diagnosis of AKI complicating AP in comparison to previously studied markers. BTP was measured in serum samples collected over the first three days of hospital stay from 73 adult patients admitted within 24 h of mild to severe AP. Thirteen patients (18%) developed AKI in the early phase of AP. Serum BTP was higher in patients who developed AKI, starting from the first day of hospitalization. Strong correlations were observed between BTP and serum cystatin C but not serum or urine NGAL. On admission, BTP positively correlated with endothelial dysfunction. The diagnostic usefulness of BTP for AKI was similar to cystatin C and lower than NGAL. Increased BTP is an early predictor of AKI complicating AP. However, it does not outperform cystatin C or NGAL.

Acute respiratory compromise on hospital wards: Association between recent ICU discharge and outcome

INTRODUCTION

Acute respiratory compromise (ARC), respiratory distress requiring emergent assisted ventilation, has a mortality of 20-40%. The relationship between recent discharge from an intensive care unit (ICU) and outcomes of patients suffering ARC on hospital wards is not well known. We hypothesized that a significant percentage of ARC events would occur in patients recently discharged from an ICU, that these patients would have worse outcomes than those without prior ICU stays, and that weekend ICU discharge would be associated with higher than expected post-ICU ARC frequency.

METHODS

Using the Get-With-The-Guidelines-Resuscitation ARC registry, we included adult, index ARC events occurring on hospital wards. Our primary analysis used multivariable logistic regression accounting for clustering by hospital to examine the association between prior ICU discharge and survival after an ARC event.

RESULTS

Of 11,800 ARCs, 937 (8%) occurred within two calendar days and 1010 (9%) >two calendar days after an ICU discharge. Patients with ICU discharge within two days had higher survival compared to those with no prior ICU stay (odds ratio 1.28 (95% CI: 1.11-1.48, p = 0.001)). Survival was not different in those with an ICU discharge more than two days prior and no prior ICU stay. Patients with ARC within two days of ICU discharge were not more likely to have left the ICU on a weekend.

CONCLUSIONS

Contrary to our hypothesis, discharge from an ICU within two calendar days was associated with better odds for survival compared to no prior ICU discharge or ICU discharge more than two days prior.

Association between mineral and bone disorder in patients with acute kidney injury following cardiac surgery and adverse outcomes

BACKGROUND

Numerous studies have evaluated the prevalence and importance of mineral and bone disorders among patients with chronic kidney disease (CKD) and end-stage renal disease (ESRD). However, little is known about dysregulated mineral and bone metabolism in acute kidney injury (AKI).

METHODS

We evaluated the association between mineral and bone metabolites and clinical outcomes in 158 patients who underwent cardiac surgery and developed AKI between June 2014 and January 2016. The baseline characteristics of the patients were recorded, and the levels of mineral and bone metabolites, including calcium, phosphate, intact parathyroid hormone (iPTH), 25-hydroxyvitamin D (25D), bone-specific alkaline phosphatase (BAP), tartrate-resistant acid phosphatase 5b (TRACP-5b) and C-terminal fibroblast growth factor 23 (cFGF23) were measured within 12 h after establishing the clinical diagnosis.

RESULTS

The serum phosphate, iPTH and cFGF23 levels were significantly associated with the 28-day mortality (phosphate: Hazard Ratio [HR] =2.620, 95% CI: 1.083 to 6.338, p = 0.035; iPTH: HR = 1.044, 95% CI: 1.001 to 1.090, p = 0.046; cFGF23: HR = 1.367, 95% CI: 1.168 to 1.599, p < 0.001). Moreover, higher serum cFGF23 and BAP levels were independently associated with an increased risk of adverse outcomes. Additionally, we found that the serum cFGF23 levels rose most significantly and were associated with the severity of AKI (P < 0.001).

CONCLUSIONS

Mineral and bone metabolites are dysregulated and are associated with adverse clinical outcomes among patients with AKI.

TRIAL REGISTRATION

www.clinicaltrials.gov NCT00953992. Registered 6 August 2009.

Nephrology in the Academic Intensive Care Unit: A Qualitative Study of Interdisciplinary Collaboration

RATIONALE & OBJECTIVE

Collaboration between nephrology consultants and intensive care unit (ICU) teams is important in light of the high incidence of acute kidney injury in today's ICUs. Although there is considerable debate about how nephrology consultants and ICU teams should collaborate, communicative dynamics between the 2 parties remain poorly understood. This article describes interactions between nephrology consultants and ICU teams in the academic medical setting.

STUDY DESIGN

Focused ethnography using semi-structured interviews and participant observation.

SETTING & PARTICIPANTS

Purposive sampling was used to enroll nephrologists, nephrology fellows, and ICU practitioners across several roles collaborating in 3 ICUs (a medical ICU, a surgical ICU, and a cardiothoracic surgical ICU) of a large urban US academic medical center. Participant observation (150 hours) and semi-structured interviews (35) continued until theoretical saturation.

ANALYTICAL APPROACH

Interview and fieldnote transcripts were coded in an iterative team-based process. Explanation was developed using an abductive approach.

RESULTS

Nephrology consultants and surgical ICU teams exhibited discordant preferences about the aggressiveness of renal replacement therapy based on different understandings of physiology, goals of care, and acuity. Collaborative difficulties resulting from this discordance led to nephrology consultants often serving as dialysis proceduralists rather than diagnosticians in surgical ICUs and to consultants sometimes choosing not to express disagreements about clinical care because of the belief that doing so would not lead to changes in the course of care.

LIMITATIONS

Aspects of this single-site study of an academic medical center may not be generalizable to other clinical settings and samples. Surgical team perspectives would provide further detail about nephrology consultation in surgical ICUs. The effects of findings on patient care were not examined.

CONCLUSIONS

Differences in approach between internal medicine-trained nephrologists and anesthesia- and surgery-trained intensivists and surgeons led to collaborative difficulties in surgical ICUs. These findings stress the need for medical teamwork research and intervention to address issues stemming from disciplinary siloing rooted in long-term socialization to different disciplinary practices.

Assessment of Canine Pancreas-Specific Lipase and Outcomes in Dogs with Hemodialysis-Dependent Acute Kidney Injury

Background

Renal replacement therapies can be life‐saving for dogs with severe acute kidney injury (AKI), however, comorbidities including pancreatitis might affect outcome.

Hypothesis/Objectives

To investigate the prevalence of pancreas‐specific lipase (Spec cPL) measurements consistent with pancreatitis (≥400 μg/L) in dogs undergoing intermittent hemodialysis (IHD) for treatment of AKI and to determine whether there were associations between 30‐days outcomes and Spec cPL measurements.

Animals

Fifty‐three client‐owned dogs presented to teaching hospitals between November 2008 and September 2016 that underwent IHD.

Methods

Retrospective medical record review from dogs that received IHD for management of AKI and also had a Spec cPL measurement. Association between survival, dialysis‐dependency, and Spec cPL measurements was assessed.

Results

Forty of 53 (76%) dogs were alive at 30‐days and 33/53(62%) had a Spec cPL result ≥400 μg/L. Spec cPL was not significantly different either between surviving (635.5 μg/L, range 29–1,001) and nonsurviving dogs (860 μg/L, range 56–1,001; P = 0.75) or between dialysis‐dependent (1,001 μg/L, range 177–1,001) and nondialysis‐dependent dogs (520 μg/L, range 29–1,001; P = 0.08). Spec cPL ≥400 μg/L was not significantly associated either with survival (P = 0.74) or dialysis‐dependency (P = 0.33).

Conclusions and Clinical Importance

Results revealed a high prevalence of Spec cPL ≥400 μg/L in dogs with AKI treated with IHD. No significant associations between Spec cPL and survival or dialysis‐dependency in dogs with AKI at 30 days were identified in the current study, however, the latter could be due to lack of power in this study.

The urine biomarker panel [IGFBP7]x[TIMP-2] (NephroCheck® parameter) does not correlate with IGFBP7 and TIMP-2 gene expression in urinary sediment

Background

Acute kidney injury (AKI) is frequently observed in serious infections, following nephrotoxic medication, surgery and trauma. Here we tested whether the detection of two recently identified biomarkers for AKI, Tissue Inhibitor of Metalloproteinase-2 (TIMP-2) and Insulin-Like Growth Factor Binding Protein 7 (IGFBP7), depends on the expression of these proteins in cells of the urinary sediment.

Method

We collected urine samples of 33 kidney transplant recipients and 14 non-transplanted patients who all had AKI (stages 1–3 according to KDIGO), and measured [IGFBP7]x[TIMP-2] using the NephroCheck^® Astute1 40 ^™ meter. Concomitantly, we analyzed IGFBP7 and TIMP-2 mRNA expression by quantitative polymerase chain reaction (qPCR) from urinary sediment of the same patients, and correlated the results with [IGFBP7]x[TIMP-2] (protein), by linear regression analysis. We also determined the association between [IGFBP7]x[TIMP-2] and estimated glomerular filtration rate (eGFR), and between IGFBP7 and TIMP-2 mRNA expression and markers of inflammation. Light microscopy and confocal immunofluorescence served to illustrate changes in the urinary sediment over the time course of renal function improvement.

Results

Of the 47 analyzed AKI patients, 14 presented with ascending urinary tract infection. Serum creatinine (sCr), blood urea nitrogen (BUN) and eGFR in all patients were 3.9±2.28 mg/dL, 47.59±23.1 mg/dL and 22.88±16.0 mL/min/1.73m², respectively, on average ±standard deviation. [IGFBP7]x[TIMP-2] was 2.33±9.95 (ng/ml)²/1000, and did not associate with IGFBP7 and TIMP-2 gene expression (r = -0.0220, p = 0.4216; respectively r = 0.0972, p = 0.1909). [IGFBP7]x[TIMP-2] did not associate with eGFR; IGFBP7 and TIMP-2 mRNA expression. Improvement of renal function went along with disappearance of casts, decrease in aquaporin1 positive renal epithelial cells and leukocytes from the urinary sediment.

Conclusion

The gene expression pattern of IGFBP7 and TIMP-2 from urinary sediment, which contains desquamated renal tubular epithelial cells, did not correlate with [IGFBP7]x[TIMP-2] protein, indicating that IGFBP7 and TIMP-2 measured in the NephroCheck^® test originated predominantly from intact but stressed cells of the kidney itself.

Diagnósticos, resultados e intervenções de enfermagem em pacientes com lesão renal aguda

Resumo Objetivo: Identificar prevalência de diagnósticos (DE), resultados (RE) e intervenções de enfermagem (IE) em pacientes com lesão renal aguda (LRA) internados em unidade de terapia intensiva (UTI). Correlacionar DE, RE e IE identificados. Métodos: Estudo transversal, conduzido em Unidades de Terapia Intensiva de um hospital público de grande porte da cidade de São Paulo. Foram incluídos, numa amostra intencional, 98 pacientes com LRA em tratamento hemodialítico internados em UTI, maiores de 18 anos. Foi utilizado software SPSS v21.0 para estimação da prevalência, fixando a estimativa no intervalo de confiança (IC) de 95% e erro amostral de 0,05. A coleta de dados foi realizada por meio da consulta de enfermagem, composta de entrevista estruturada, anamnese e exame físico dos pacientes, utilizando instrumento elaborado pelos pesquisadores. A coleta de dados foi realizada no período de março a julho de 2016, e o instrumento de coleta de dados foi preenchido pela pesquisadora principal. A consulta de enfermagem teve duração de aproximadamente 30 minutos. Do total da amostra, 10% foi selecionada aleatoriamente e checada, com o propósito de avaliar a qualidade dos dados e valores atípicos. Foi também realizado teste piloto em dois pacientes, previamente, para verificar se as informações contidas no instrumento atingiriam os objetivos da pesquisa. Resultados: Participaram 98 pacientes, predominantemente com idade ≥60 anos (33%), sexo masculino (60%) e classificados com lesão pré-renal (54%). DE prevalentes (100%): risco de infecção, risco de perfusão gastrointestinal ineficaz, risco de perfusão renal ineficaz, risco de desequilíbrio eletrolítico, volume de líquidos excessivos e risco de volume de líquidos desequilibrados. RE prevalentes (100%): gravidade da infecção, acesso para hemodiálise, perfusão tissular: órgãos abdominais, equilíbrio hídrico, mobilidade, remoção de toxinas e função renal. IE prevalentes (100%): promoção contra infecção, controle de infecção, manutenção de acesso para diálise, controle hidroeletrolítico, controle de eliminação urinária, controle ácido-básico, controle de eletrólitos, controle de hipervolemia, controle hídrico, monitorização hídrica, fisioterapia respiratória, monitorização respiratória e posicionamento. Correlações foram significativas (p< 0,001) entre DE e IE e entre IE e RE. Conclusão: Os principais DE, RE e IE foram relacionados à perda da função renal, origem das alterações na perfusão renal, volemia, distúrbios hidroletroliticos e risco para infecção. O número de DE atribuídos mostrou-se relacionado ao número de IE, assim como, das IE aos RE.

Transpulmonary thermodilution (TPTD) before, during and after Sustained Low Efficiency Dialysis (SLED). A Prospective Study on Feasibility of TPTD and Prediction of Successful Fluid Removal

BACKGROUND

Acute kidney injury (AKI) is common in critically ill patients. AKI requires renal replacement therapy (RRT) in up to 10% of patients. Particularly during connection and fluid removal, RRT frequently impairs haemodyamics which impedes recovery from AKI. Therefore, "acute" connection with prefilled tubing and prolonged periods of RRT including sustained low efficiency dialysis (SLED) has been suggested. Furthermore, advanced haemodynamic monitoring using trans-pulmonary thermodilution (TPTD) and pulse contour analysis (PCA) might help to define appropriate fluid removal goals.

OBJECTIVES, METHODS

Since data on TPTD to guide RRT are scarce, we investigated the capabilities of TPTD- and PCA-derived parameters to predict feasibility of fluid removal in 51 SLED-sessions (Genius; Fresenius, Germany; blood-flow 150 mL/min) in 32 patients with PiCCO-monitoring (Pulsion Medical Systems, Germany). Furthermore, we sought to validate the reliability of TPTD during RRT and investigated the impact of "acute" connection and of disconnection with re-transfusion on haemodynamics. TPTDs were performed immediately before and after connection as well as disconnection.

RESULTS

Comparison of cardiac index derived from TPTD (CItd) and PCA (CIpc) before, during and after RRT did not give hints for confounding of TPTD by ongoing RRT. Connection to RRT did not result in relevant changes in haemodynamic parameters including CItd. However, disconnection with re-transfusion of the tubing volume resulted in significant increases in CItd, CIpc, CVP, global end-diastolic volume index GEDVI and cardiac power index CPI. Feasibility of the pre-defined ultrafiltration goal without increasing catecholamines by >10% (primary endpoint) was significantly predicted by baseline CPI (ROC-AUC 0.712; p = 0.010) and CItd (ROC-AUC 0.662; p = 0.049).

CONCLUSIONS

TPTD is feasible during SLED. "Acute" connection does not substantially impair haemodynamics. Disconnection with re-transfusion increases preload, CI and CPI. The extent of these changes might be used as a "post-RRT volume change" to guide fluid removal during subsequent RRTs. CPI is the most useful marker to guide fluid removal by SLED.