Deep learning algorithms for predicting renal replacement therapy initiation in CKD patients: a retrospective cohort study

BACKGROUND

Chronic kidney disease (CKD) requires accurate prediction of renal replacement therapy (RRT) initiation risk. This study developed deep learning algorithms (DLAs) to predict RRT risk in CKD patients by incorporating medical history and prescriptions in addition to biochemical investigations.

METHODS

A multi-centre retrospective cohort study was conducted in three major hospitals in Hong Kong. CKD patients with an eGFR < 30ml/min/1.73m2 were included. DLAs of various structures were created and trained using patient data. Using a test set, the DLAs' predictive performance was compared to Kidney Failure Risk Equation (KFRE).

RESULTS

DLAs outperformed KFRE in predicting RRT initiation risk (CNN + LSTM + ANN layers ROC-AUC = 0.90; CNN ROC-AUC = 0.91; 4-variable KFRE: ROC-AUC = 0.84; 8-variable KFRE: ROC-AUC = 0.84). DLAs accurately predicted uncoded renal transplants and patients requiring dialysis after 5 years, demonstrating their ability to capture non-linear relationships.

CONCLUSIONS

DLAs provide accurate predictions of RRT risk in CKD patients, surpassing traditional methods like KFRE. Incorporating medical history and prescriptions improves prediction performance. While our findings suggest that DLAs hold promise for improving patient care and resource allocation in CKD management, further prospective observational studies and randomized controlled trials are necessary to fully understand their impact, particularly regarding DLA interpretability, bias minimization, and overfitting reduction. Overall, our research underscores the emerging role of DLAs as potentially valuable tools in advancing the management of CKD and predicting RRT initiation risk.

An atypical airway cast

A 60-year-old man intubated for airway protection after smoke inhalation was found to have decompensated hypercapnic respiratory failure. Fiberoptic bronchoscopy revealed obstructive airway slough and pseudomembrane, a manifestation of severe inhalation injury. Veno-venous extracorporeal membrane oxygenation was established for stabilization. The airway casts were removed successfully with periprocedural veno-venous extracorporeal membrane oxygenation support.

Pyroglutamic Acidosis - An Underrecognised Entity Associated with Acetaminophen Use

Pyroglutamic acidosis (PGA) is an underrecognized entity characterised by raised anion gap metabolic acidosis (RAGMA) and urinary hyper-excretion of pyroglutamic acid. It is frequently associated with chronic acetaminophen (APAP) ingestion. We report the case of a 73-year-old man with invasive pulmonary aspergillosis treated with voriconazole and APAP for analgesia with a cumulative dose of 160 g over 40 days. PGA was suspected as he developed severe RAGMA and common causes were excluded. Diagnosis was confirmed via urinary organic acid analysis which showed significant hyper-excretion of pyroglutamic acid. APAP was discontinued, and N-acetylcysteine (NAC) was administered. His RAGMA rapidly resolved following treatment.

Emerging Infections Due to Shewanella spp.: A Case Series of 128 Cases Over 10 Years

Background

Shewanella species are emerging pathogens that can cause severe hepatobiliary, skin and soft tissue, gastrointestinal, respiratory infections, and bacteremia. Here we reported the largest case series of infections caused by Shewanella species.

Aim

To identify the clinical features and risk factors predisposing to Shewanella infections. To evaluate resistance pattern of Shewanella species and appropriateness of antibiotic use in the study cohort.

Methods

Patients admitted to a regional hospital in Hong Kong with Shewanella species infection from April 1, 2010 to December 31, 2020 were included. Demographics, antibiotics, microbiology, and outcomes were retrospectively analyzed.

Findings

Over the 10 years, we identified 128 patients with Shewanella species infection. 61.7% were male with a median age of 78 (IQR 65–87). Important underlying diseases included hepatobiliary diseases (63.3%), malignancy (26.6%), chronic kidney disease or end-stage renal failure (25.8%), and diabetes mellitus (22.7%). Hepatobiliary infections (60.4%) were the most common clinical manifestation. Majority (92.2%) were infected with Shewanella algae, while 7.8% were infected with Shewanella putrefaciens. The identified organisms were usually susceptible to ceftazidime (98.7%), gentamicin (97.4%), cefoperazone-sulbactam (93.5%) and ciprofloxacin (90.3%). Imipenem-susceptible strains were only present in 76.6% of isolates.

Conclusion

This largest case series suggested that Shewanella infections are commonly associated with underlying comorbidities, especially with hepatobiliary diseases and malignancy. Although Shewanella species remained largely susceptible to third and fourth generation cephalosporins and aminoglycosides, carbapenem resistance has been on a significant rise.

A newly discovered verotoxin variant, VT2g, produced by bovine verocytotoxigenic Escherichia coli

A new verotoxin (VT) variant, designated vt2g, was identified from a bovine strain of verocytotoxigenic Escherichia coli (VTEC) serotype O2:H25. When vt2g was aligned with published sequences of vt2 and vt variants, it exhibited the highest DNA sequence homology with vt2 and vt2c. However, vt2g was not detected by vt2-specific primers and probes, although it was partially neutralized by an antiserum to the VT2A subunit. VT2g was cytotoxic for Vero and HeLa cells and was not activated by mouse intestinal mucus. The vt2g gene was detected in 3 of 409 (0.7%) bovine VTEC strains, including serotypes O2:H25, O2:H45 and Ont:H-.

Polyclonal antibodies to glutathione S-transferase--verotoxin subunit a fusion proteins neutralize verotoxins

The A1 subunits of verotoxin-1 (VT1) and VT2 genes were cloned into pGEX-4T-2 for the expression of glutathione S-transferase (GST) fusion proteins. The N-terminal and the transmembrane regions of the A1 subunits were excluded from the constructs in order to increase the product yields. Polyclonal anti-VT1A1 and anti-VT2A1 antibodies were produced by immunizing rabbits with GST-VT1A1 and GST-VT2A1 fusion proteins, respectively. The antibodies were tested for their ability to neutralize active toxins from 45 VT-producing Escherichia coli (VTEC) strains. The antibodies had significantly high neutralizing activities against their homologous toxins. The average percentages of neutralization of VT1 by anti-GST-VT1A1 and anti-GST-VT2A1 were 76.7% +/- 7.9% and 3.6% +/- 2.3%, respectively, and those of VT2 were 1.7% +/- 2.3% and 82.5% +/- 13.9%, respectively. VT2 variant toxin was neutralized by anti-GST-VT2A1, with cross neutralization being a possible consequence of sequence homology between VT2 and a VT2 variant. To our knowledge, this is the first report on the production of polyclonal antibodies from GST-VT fusion proteins. The antibodies were shown to exhibit specific toxin neutralizing activities and may be useful for immunological diagnosis of VTEC infections.