Identification of diagnostic urinary biomarkers for acute kidney injury

Abstract

Acute kidney injury (AKI) is an important cause of death among hospitalized patients. The 2 most common causes of AKI are acute tubular necrosis (ATN) and prerenal azotemia (PRA). Appropriate diagnosis of the disease is important but often difficult. We analyzed urine proteins by 2-dimensional gel electrophoresis from 38 patients with AKI. Patients were randomly assigned to a training set, an internal test set, or an external validation set. Spot abundances were analyzed by artificial neural networks to identify biomarkers that differentiate between ATN and PRA. When the trained neural network algorithm was tested against the training data, it identified the diagnosis for 16 of 18 patients in the training set and all 10 patients in the internal test set. The accuracy was validated in the novel external set of patients where conditions of 9 of 10 patients were correctly diagnosed including 5 of 5 with ATN and 4 of 5 with PRA. Plasma retinol-binding protein was identified in 1 spot and a fragment of albumin and plasma retinol-binding protein in the other. These proteins are candidate markers for diagnostic assays of AKI.

Figure 1

Unsupervised simultaneous clustering of gels and spots by UPGA. Individual gels (patients) are shown on the y axis. Spots are shown on the x axis. Protein intensity is shown in the color map. UPGA did not reveal any consistent patterns of similarity between patients with the same disease, age, urine protein concentration, run batch for 2-DE or serum creatinine concentration.

Figure 2

Two dimensional gel electrophoresis of urine protein from a patient with ATN. The two spots necessary for predicting the cause of AKI are shown. Two proteins were identified in spot 40 is a fragment of albumin and a fragment of plasma retinol binding protein. Spot 133 is intact plasma retinol binding protein.

Figure 3

Plot of quantile ranks for two spots which can distinguish between ATN and PRA. X and Y planes show quantile ranks for spot 133 (PRBP) and 40 (fragments of PRBP and albumin). Spots in the shaded area on the graph predicts PRA and in the white area predicts ATN. The prediction algorithm was derived from the training set. Only patients in the external validation set are plotted on the graph. All patients were predicted correctly except the patient with PRA at the far left of the figure.

Figure 4

Match of peptides and sequence to identified proteins. Underlined sequence was matched to predicted peptides from MALDI-TOF spectrum. Red sequence was identified by tandem mass spectrometry. A. Spot 133 was identified as Plasma retinol binding protein. 53% of the intact protein was covered by the assigned peptides. B. Spot 40, Protein 1. An internal sequence of plasma retinol binding protein was identified by tandem MS sequence of one peptide and predicted peptide masses of three peptides. The corresponding peptides covered 100% of a 5.8 kDa protein fragment of plasma retinol binding protein. C. Spot 40, protein 2. An internal sequence of serum albumin was identified by tandem MS sequence of two peptides and predicted masses of 4 peptides. The corresponding peptides covered 50% of a 12.2 kDa protein fragment of serum albumin.

Figure 5

MALDI-TOF MS spectrum of tryptic digest of protein spot 40. Blue peaks match to a fragment of plasma retinol binding protein. Green peaks match to a fragment of albumin. Ten of the 12 most abundant peptides in this view can be attributed to one of the two proteins or a known contaminant. Assignment of the masses to the appropriate sequence is shown in figure 4B and C and in table 4.