MSACL 2024 Abstract

INTRODUCTION: The applications of direct mass spectrometry strategies such as paper spray mass spectrometry (PS-MS) offer new alternatives to increase throughput in clinical workflows. In PS-MS, direct, measurements in complex samples such as biofluids can be made for small sample aliquots (i.e., ≤10µL). Samples are deposited on pointed paper strips with co-deposited internal standards for quantitative measurements. When the strips are moistened with a suitable solvent, and upon the application of high voltage, ions are generated in a manner akin to electrospray, allowing direct analyte quantitation via tandem mass spectrometry. The strips are inexpensive and disposed for each measurement, eliminating carryover. PS-MS offers the potential to make rapid, simultaneous measurements of both small and large molecules in a single run. Albuminuria is a clinical condition associated with poor kidney function, diagnosed by determining the ratio of albumin to creatinine concentrations in patient urine samples. We present a high-throughput paper spray mass spectrometry (PS-MS) method for simultaneous quantitation of urinary albumin and creatinine for potential diagnosis of albuminuria, and the use of high resolution accurate mass spectrometry (HRAM) PS-MS coupled with machine learning for the non-targeted detection of albuminuria progression.

OBJECTIVES: To demonstrate the use and effectiveness of PS-MS for the rapid, sensitive, and quantitative and simultaneous measurement of urinary albumin and creatinine as a candidate replacement for existing clinical albuminuria diagnostics. This includes the high throughput, quantitative ACr ratio determinations as well as the non-targeted detection of the progression of albuminuria utilizing PS-MS with high resolution accurate mass spectrometry.

METHODS: All quantitative ACr measurements were performed by paper spray tandem mass spectrometry with a high-throughput paper spray ion source (Thermo Scientific™ TSQ Altis™ triple quadrupole mass spectrometer with a VeriSpray™ source). Non-targeted HRAM-PS-MS was achieved with a bespoke PS-MS interface and an Exploris 120™ orbitrap mass spectrometer. For all quantitative ACr measurements, 10 uL urine samples were co-deposited with internal standards on PS-MS samples strips using barcode traceable VeriSpray™ PS-MS sample plates (24 strips/plate), allowing unattended measurement of up to 240 samples. Non-targeted measurements were made with the same sample volumes without using internal standards, deposited on individual PS-MS paper strips of the same geometry, but analyzed individually.

RESULTS: The analytical performance of the quantitative PS-MS ACr method was evaluated, achieving linear calibration curves (R2>0.99) with little inter-day variability in slopes (N = 5 days), and exhibiting coefficient of variation (CV) of 8% for albumin and 3% for creatinine. LOD and LOQ for albumin were 2.1 and 7.0 mg L-1, and for creatinine were 0.01 and 0.03 mmol L-1. Intra- and inter-day (N = 5) precisions (%CV) and accuracies (%bias) were <10% and ±11%, respectively, for both analytes. The method was applied to determine albumin-to-creatinine ratios in anonymous human patient urine samples (N = 56), and a correlation of R2 = 0.9744 was achieved between the PS-MS results and validated clinical method results utilizing the Jaffe method for creatinine and immunoturbidimetry for albumin. This work demonstrates the utility of PS-MS to simultaneously quantify a large (albumin) and a small (creatinine) molecule directly in patient urine samples with one method. Non-targeted HRAM-PS-MS measurements (full scan data) were analyzed using 80 samples (40 healthy, ACr <2 mg mmol-1; 40 diseased, ACr >3 mg mmol-1) in combination with the Random Forest machine learning method. Clinical diagnostic test parameters of accuracy (96.3%), sensitivity (92.5%), specificity (100.0%), positive predictive value (100.0%), and negative predictive value (93.0%) were obtained. These results demonstrate that HRAM-PS-MS combined with machine learning is a potential tool for rapid diagnosis of clinical albuminuria and its progression.

CONCLUSION: The 'one instrument' quantitative PS-MS measurement of urinary albumin/creatinine ratios and the use non-targeted HRAM-PS-MS combined with machine learning illustrate their potential as candidate new methods for clinical albuminuria diagnostics.