Machine Learning Improves Upon Clinicians' Prediction of End Stage Kidney Disease.

Background and Objectives: Chronic kidney disease progression to ESKD is associated with a marked increase in mortality and morbidity. Its progression is highly variable and difficult to predict. Methods: This is an observational, retrospective, single-centre study. The cohort was patients attending hospital and nephrology clinic at The Canberra Hospital from September 1996 to March 2018. Demographic data, vital signs, kidney function test, proteinuria, and serum glucose were extracted. The model was trained on the featurised time series data with XGBoost. Its performance was compared against six nephrologists and the Kidney Failure Risk Equation (KFRE). Results: A total of 12,371 patients were included, with 2,388 were found to have an adequate density (three eGFR data points in the first 2 years) for subsequent analysis. Patients were divided into 80%/20% ratio for training and testing datasets.ML model had superior performance than nephrologist in predicting ESKD within 2 years with 93.9% accuracy, 60% sensitivity, 97.7% specificity, 75% positive predictive value. The ML model was superior in all performance metrics to the KFRE 4- and 8-variable models.eGFR and glucose were found to be highly contributing to the ESKD prediction performance. Conclusions: The computational predictions had higher accuracy, specificity and positive predictive value, which indicates the potential integration into clinical workflows for decision support.

Cystatin C kidney functional reserve: a simple method to predict outcome in chronic kidney disease.

BACKGROUND: Kidney functional reserve (KFR), the only clinical kidney stress test, is not routinely measured because the complexity of measurement has limited clinical application. We investigated the utility of plasma cystatin C (CysC) after oral protein loading (PL) to determine KFR in Stages 3 and 4 chronic kidney disease (CKD). METHODS: Following a 24-h low-protein diet, KFR was measured after oral protein by hourly plasma CysC and compared with simultaneous creatinine clearance (CrCl) and radionuclide 99technetium diethylenetriaminepentaacetatic acid (Tc-99m-DTPA) measured glomerular filtration rate (mGFR) measurement in an observational, single-centre cohort study of adults with CKD Stages 3 and 4. Subjects were followed for 3 years for fast (F) or slow (S) CKD progression, dialysis requirement or death or a combination of major adverse kidney events (MAKEs). RESULT: CysC, CrCl and Tc-99m-DTPA mGFR measurements of KFR in 19 CKD Stage 3 and 21 CKD Stage 4 patients yielded good agreement. KFR was not correlated with baseline kidney function. Eight CKD Stage 3 (42%) and 11 CKD Stage 4 (52%) subjects reached their lowest serum CysC concentration 4 h after PL. CysC KFR and baseline serum creatinine (sCr) predicted death or dialysis or MAKE-F with a respective area under the curve (AUC) of 0.73 [95% confidence interval (CI) 0.48-0.89] and 0.71 (95% CI 0.51-0.84). Including CysC KFR, age, baseline sCr and nadir CysC predicted a decrease in sCr-estimated GFR >1.2 mL/min/year (MAKE-S) with an AUC of 0.89. CONCLUSIONS: Serial CysC avoided timed urine collection and radionuclide exposure and yielded equivalent estimates of KFR. Serial CysC may facilitate monitoring of KFR in clinical practice.

Increased burden of rare variants in genes of the endosomal Toll-like receptor pathway in patients with systemic lupus erythematosus.

OBJECTIVE: To compare the frequency of rare variants in genes of the pathophysiologically relevant endosomal Toll-like receptor (eTLR) pathway and any quantifiable differences in variant rarity, predicted deleteriousness, or molecular proximity in patients with systemic lupus erythematosus (SLE) and healthy controls. PATIENTS AND METHODS: 65 genes associated with the eTLR pathway were identified by literature search and pathway analysis. Using next generation sequencing techniques, these were compared in two randomised cohorts of patients with SLE (n = 114 and n = 113) with 197 healthy controls. Genetically determined ethnicity was used to normalise minor allele frequencies (MAF) for the identified genetic variants and these were then compared by their frequency: rare (MAF < 0.005), uncommon (MAF 0.005-0.02), and common (MAF >0.02). This was compared to the results for 65 randomly selected genes. RESULTS: Patients with SLE are more likely to carry a rare nonsynonymous variant affecting proteins within the eTLR pathway than healthy controls. Furthermore, individuals with SLE are more likely to have multiple rare variants in this pathway. There were no differences in rarity, Combined Annotation Dependent Depletion (CADD) score, or molecular proximity for rare eTLR pathway variants. CONCLUSIONS: Rare non-synonymous variants are enriched in patients with SLE in the eTLR pathway. This supports the hypothesis that SLE arises from several rare variants of relatively large effect rather than many common variants of small effect.

The kinetic estimated glomerular filtration rate ratio predicts acute kidney injury.

AIM: Kinetic estimated Glomerular Filtration Rate (KeGFR) approximates GFR under non-steady-state conditions. We investigated whether the ratio of KeGFR difference to baseline eGFR could predict acute kidney injury (AKI) earlier than a creatinine-based algorithm that triggered an AKI electronic Alert (eAlert). METHODS: This retrospective, single-centre, proof-of-concept cohort study assessed all patients diagnosed with AKI by an automated serum creatinine-based eAlert. The kinetic eGFR, the kinetic eGFR difference from baseline and the ratio of difference to baseline was calculated in subjects with at least two serum creatinine (sCr) measurements within 72 h of AKI. RESULTS: Patients in the AKI cohort (n = 140) had a significant decline in KeGFR ratio (AKI: 17% IQR 7% to 29%, Non-AKI: 0 IQR -12% to 9%; P-value <.0001). A decrease of the ratio greater than 10% predicted AKI with a sensitivity of 66%, a specificity of 77%, a positive predictive value of 63%, and negative predictive value of 80%. The median lead time between KeGFR ratio decrease and AKI was 24 h (IQR: 19-27 h). CONCLUSIONS: KeGFR ratio is a cheap, simple method that predicted AKI 24 h before laboratory detection. KeGFR may facilitate triaging patients to increased monitoring or intervention.

Successful Treatment of PD Peritonitis Due to Brevundimonas vesicularis.

Brevundimonas vesicularis is a rare cause of peritoneal dialysis-associated peritonitis (PD peritonitis). One documented case report described poor outcome despite treatment with appropriate antibiotics. Here, we report the successful treatment of PD peritonitis due to Brevundimonas vesicularis with 21 days of intraperitoneal (IP) antibiotics using the regimen described for Pseudomonas aeruginosa, notably IP gentamicin (10 days) followed by IP cefepime and oral ciprofloxacin (11 days). Despite the favorable outcome, measurement of antibiotic concentrations in the PD effluent suggests that the cefepime and possibly ciprofloxacin regimens do not achieve key antibiotic concentration targets that are reported to maximize bacterial kill. The role of routine therapeutic drug monitoring to maximize clinical outcomes from antibiotic therapy for PD peritonitis requires further consideration.

Calciphylaxis in a dialysis patient successfully treated with high-dose vitamin K supplementation.

Calciphylaxis has high mortality. Vitamin K deficiency is common in haemodialysis patients and may be a trigger for calciphylaxis due to its role in activating matrix Gla protein (a tissue inhibitor of calcification). We report the case of a 43-year-old female haemodialysis patient who developed calciphylaxis. Two months prior to the diagnosis she was found to have an undetectable plasma vitamin K concentration. The calciphylaxis completely resolved with vitamin K supplementation and an increase in haemodialysis frequency. She did not receive sodium thiosulphate or bisphosphonates. Supplementation of vitamin K in deficient patients may improve the outcome of this condition.

Regional citrate anticoagulation in membrane based plasma exchange: Safety, efficacy and effect on calcium balance.

AIM: To assess the efficacy, safety and calcium balance of a membrane based regional citrate anticoagulation plasma exchange protocol. METHODS: This was an observational, prospective, single centre study of membrane separation plasma exchange using regional citrate anticoagulation. It was performed using a fixed dose pre-filter citrate infusion that was based on the plasma flow rate. Patients received a post filter calcium infusion that was modified during treatment based on systemic ionized calcium monitoring. Post filter ionized calcium was not assessed. Safety and efficacy were assessed by extraction of clinical events and laboratory data contemporaneously recorded in electronic health records. RESULTS: Thirty-six sessions in five patients were performed. No patients developed symptomatic hypocalcaemia, and no patient had a recorded ionized calcium below 0.81 mmol/L. Filter clotting occurred in two sessions. The mean net calcium gained was 9.6 ± 1.8 mmol per session. CONCLUSION: Regional citrate anticoagulated membrane separation plasma exchange can be performed safely and effectively without the need for post filter ionized calcium monitoring. The algorithm employed resulted in a net calcium gain.