|
Abstract INTRODUCTION:
2,3-dinor 11β-Prostaglandin F2α (2,3-BPG) is released during mast cell activation making it a non-invasive urinary biomarker for mast cell disorders.1 While 2,3-BPG is the most abundant prostaglandin D2 metabolite, it poses an analytical challenge for LC-MS/MS analysis due to numerous structurally related interferences in urine.1 Prior research has demonstrated that differential mobility spectrometry (DMS) improves specificity for 2,3-BPG when combined with off-line anion exchange solid phase extraction.2 We sought to develop an on-line two-dimensional (2-D) LC-MS/MS method using mixed-mode anion exchange (MAX) coupled with reverse phase chromatography and SelexION DMS to reduce interferences while preserving the existing liquid-liquid sample extraction method.2 A preliminary study also assessed whether the on-line 2-D method could support dilute-and-shoot preparation to reduce hands-on time.
METHODS:
An on-line 2-D MAX-reverse phase LC method was developed on a Thermo Scientific Transcend TLX-4 UHPLC with TriPlus RSI autosampler, Vanquish binary and quaternary pumps, and valve interface module. The first dimension used a Waters Oasis MAX Cartridge (80Å, 30 µm, 2.1 x 20 mm) loaded with 0.05% ammonium hydroxide and eluted with an acidic organic plug into a Waters CORTECS T3 column (120Å, 1.6 µm, 2.1 x 50 mm, 60 °C) for analytical separation. 40 µL of extracted urine was injected and loaded for 1.5 min. Flow was reduced and valves switched to place the elution loop in line with the trap column while combining flow with the second dimension. After transfer over 1 min, valves were switched back to isolate each column. After a gradient ramp, 40 %B was held for 6.84 min. Flow was diverted to the MS from 8.5 to 11.5 min, followed by an organic wash for 1.67 min. Total run time was 14.08 min. During separation, the trap was washed at high flow (1-2 mL/min) followed by loop filling and re-equilibration. MS was performed using a Sciex 6500+ with SelexION in negative ion mode (IS -4500, TEM 550 °C, CUR 35, CAD 9, GS1 75, GS2 20; DP -40, EP -10, CXP -12). DMS: DT low, no modifier, SV 3875, COV 10, DMO 20, DR low. Three MRM transitions were monitored: 325.3 > 163.0 (CE -24), 325.3 > 145.1 (CE -20), and IS 334.3 > 145.1 (CE -24).
The reference method was performed as previously described.2 Briefly, 1 mL of urine was combined with 25 µL of IS, acidified with 50% acetic acid, and extracted with ethyl acetate. The organic phase was evaporated and reconstituted in 135 µL mobile phase. 10 µL of sample was injected onto a legacy Transcend TLX-4 system with Shimadzu LC-20 pumps and a Sciex 5000 MS.
Standards, QCs, and 41 patient samples were analyzed by both methods. Results were processed in Sciex OS using linear regression (1/x weighting) from 312.5-10,000 pg/mL. Statistical analysis was performed in Microsoft Excel and Python 3.11.
RESULTS:
2,3-BPG results were compared between the reference and 2D-SelexION methods using concentrations derived from calibrating each transition separately to assess specificity. For m/z 163, comparison of 2D-SelexION to reference yielded a Passing-Bablok slope of 0.772 (95% CI 0.576-0.859, y-int=5.01, r=0.938) and Bland-Altman percent bias of -37.0%. For m/z 145, the slope was 0.805 (95% CI 0.707-0.925, y-int=38.2, r=0.969) with percent bias of -20.3%.
Within-method m/z 163/145 fragment ion agreement differed substantially between the reference and 2D-SelexION methods. The reference method gave a slope of 0.896 (95% CI 0.850-0.965, y-int=-6.05, r=0.990) with percent bias of -17.1%. The 2D-SelexION method gave a slope of 0.980 (95% CI 0.936-1.012, y-int=11.49, r=0.999) with percent bias of 2.6%. The 2D-SelexION method demonstrated reduced ion ratio variability across the comparison batch compared to the reference method (CV 10.9% vs 19.2%). Applying the reference method acceptance criterion for quantifier-qualifier agreement (±20% difference) reduced the number of samples exceeding this threshold from 10 to 3.
In a dilute-and-shoot pilot study (n=10), comparison of 2D-SelexION m/z 163 results to reference LLE m/z 163 results gave a slope of 0.965 (95% CI 0.682-1.243, y-int=-51.90, r=0.962) with percent bias of -11.1%. Within the dilute-and-shoot 2D-SelexION method, comparison of m/z 163 and 145 results gave a slope of 0.985 (95% CI 0.885-1.106, y-int=50.43, r=0.994) with percent bias of 3.2%. 2 of 10 samples exceeded the ±20% acceptance criterion. Despite reduced signal-to-noise compared with LLE, these findings suggest dilute-and-shoot may be feasible after further optimization.
DISCUSSION:
Initial experiments during method development using a hydrophilic-lipophilic balanced cartridge provided no improvement over the existing method alone. In contrast, 2,3-BPG was strongly retained on MAX under basic conditions and efficiently released under acidic organic conditions. Method optimization showed that 2% formic acid in methanol was required for narrow-band elution, low initial organic conditions combined with at-column dilution were needed to refocus analyte during transfer, and a loading flow of 0.5 mL/min was necessary to avoid breakthrough. Although the on-line 2-D MAX method alone provided only modest improvement, combining it with the SelexION improved fragment-ion agreement and reduced interference-related discordance while preserving the existing extraction workflow. Furthermore, a pilot study demonstrates the potential efficacy for on-line extraction to reduce the manual sample preparation burden. The data suggest that the remaining interferences may be structurally similar acidic compounds also retained by the anion exchange dimension.
REFERENCES:
1. JH Butterfield, CR Weiler. Utility of urinary mast cell mediator metabolites in systemic mastocytosis and MCAS. J Allergy Clin Immunol Pract. 2020;8:2533–41.
2. K Moehnke, et al. Differential mobility spectrometry improves specificity of 2,3-dinor-11β-PGF2α measurement. Clin Biochem. 2024;126:110745. |
= Discovery stage. (57.21%, 2026)
= Translation stage. (23.38%, 2026)
= Clinically available. (19.40%, 2026)
