Phenotyping prerenal and intrarenal acute kidney injury by urine mass spectrometry

ABSTRACT

Purpose of Study Postrenal causes of acute kidney injury (AKI) can generally be determined by imaging but differentiation of prerenal (PR) from intrinsic renal (IR) can be more difficult. A point of care test to quickly differentiate these two entities would be useful. Mass spectrometry (MS) can visualize small molecules in urine. We utilized a portable, single quadrupole mass spectrometer with a simple atmospheric pressure ionization interface to measure small molecules in urine. The interface allows for direct analysis of samples without the need for chromatography or other time-consuming sample preparation. The goal was to distinguish PR and IR AKI by urine analyte profiles. Methods Used Inpatients that developed AKI were eligible for enrolment. Patients with COVID-19 were excluded. Informed consent was obtained under a protocol approved by the UAMS IRB. Patients were categorized as either PR (n=9) or IR (n=7) AKI etiology using the diagnosis by the on-service nephrology attending. Etiologies for IR AKI included tumor lysis syndrome, contrast-induced nephropathy, myeloma light chain disease, pyelonephritis, and cisplatin-induced kidney injury. Two microliters of urine was dispensed onto a stainless steel probe without prior processing and analyzed by MS with an Advion Expression CMS Mass Spectrometer. Peaks within the 20 to 500 m/z range were recorded. MS spectra were processed and binned in 1 m/z increments for peak clustering using MATLAB. The frequency of binned peaks in the PR AKI group were compared to the IR group and marked as peaks of interest if the difference in number of peaks across the two categories was four or more. Missing peak values were replaced with the minimum value for that peak across all samples. Summary of Results Eleven peaks met our initial criteria for difference in frequency between intrinsic and prerenal cases. From the eleven, the mean intensity was different between the groups for only the sample at m/z 242 (p=0.03). In the PR group, mean and SD of the peak intensities were 1.7E+07 ±1.2E+07 compared to 7.0E+06 ±7.7E+06 Using a cut-off value of 1.55E+07, overall accuracy was 75%. The predictor correctly classified 6/9 in the PR group (67%) and 6/7 in the IR group (86%). Conclusions Point of care MS has the potential to rapidly differentiate PR from IR and potentially to further phenotype causes of AKI. Based on previous studies, the peaks we identified are likely small molecule metabolites. Larger sample size will be needed to better phenotype patients and will enable the use of machine learning algorithms to classify the kidney disease that is present. Further correlation of the different analytes in prerenal and intrarenal AKIs is needed, however this study shows the potential utility of mass spectrometry to rapidly phenotype AKI in the clinical setting.