Estimation of measurement uncertainty for the quantification of protein by ID-LC–MS/MS

The emergence of mass spectrometry (MS)-based methods to quantify proteins for clinical applications has led to the need for accurate and consistent measurements. To meet the clinical needs of MS-based protein results, it is important that the results are traceable to higher-order standards and methods and have defined uncertainty values. Therefore, we outline a comprehensive approach for the estimation of measurement uncertainty of a MS-based procedure for the quantification of a protein biomarker. Using a bottom-up approach, which is the model outlined in the “Guide to the Expression of Uncertainty of Measurement” (GUM), we evaluated the uncertainty components of a MS-based measurement procedure for a protein biomarker in a complex matrix. The cause-and-effect diagram of the procedure is used to identify each uncertainty component, and statistical equations are derived to determine the overall combined uncertainty. Evaluation of the uncertainty components not only enables the calculation of the measurement uncertainty but can also be used to determine if the procedure needs improvement. To demonstrate the use of the bottom-up approach, the overall combined uncertainty is estimated for the National Institute of Standards and Technology (NIST) candidate reference measurement procedure for albumin in human urine. The results of the uncertainty approach are applied to the determination of uncertainty for the certified value for albumin in candidate NIST Standard Reference Material® (SRM) 3666. This study provides a framework for measurement uncertainty estimation of a MS-based protein procedure by identifying the uncertainty components of the procedure to derive the overall combined uncertainty. Graphical Abstract Supplementary Information The online version contains supplementary material available at 10.1007/s00216-023-04705-8.

Experimental Design. The samples were digested via the trypsin digestion protocol included in the NIST candidate RMP [5] and the digested samples were analyzed via LC-MS/MS using an Orbitrap Elite (Thermo Scientific) ion trap mass spectrometer. The Orbitrap Elite was operated in the positive ion mode using CID. The high resolution (R=120,000) full MS data was processed manually to assess the peak intensities of the isotopic distribution of the twelve (12) tryptic peptides.
Results/Discussion. The isotope enrichment of the 15 N-labeled IS was determined using the peak intensity data [three process replicates (n=36)]. The average 15 N incorporation for the IS was 99.7% (CV of 0.5 %) with an average r-value of 0.9999 (CV of 0.02 %) (Table S1). Figure S1 illustrates the theoretical and experimental (observed) isotopic distribution for unlabeled (NIST SRM 2925) and 15 N-labeled IS tryptic peptides: YLYEIAR, TYETTLEK and QTALVELVK. The isotopic distribution of the NIST SRM 2925 (unlabeled) is equivalent to that of the 0 % theoretical isotopic probability distribution and the 15 N-Labeled material is equivalent to that of the 99 % theoretical isotopic probability distribution. Table S1. Percent label incorporation (Incor) and r-value for the twelve peptides for the 15 N-labeled IS.

B. DESIGN OF EXPERIMENT (DOE) OPTIMIZATION STUDY
To reduce the PAR uncertainty ( ) of the National Institute of Standards and Technology (NIST) candidate reference measurement procedure (RMP) [5], the design of experiment (DOE) Optimization study was conducted to optimize trypsin digestion protocol of the measurement of albumin. The DOE Optimization study was performed to establish the optimal settings for the trypsin digestion protocol of the candidate RMP [5].
Experimental Design. The Response Surface Method (RSM) was applied using the Central Composite Design (CCD). The design consisted of the following: full factorial matrix (2 3 ), two center points (red) with six (6) replicates per sample, and six (6) star points (± α) ( Figure S2). The α value of 1.684 was determined using the following equation: where k is 3 for the number of factors. The three-factors included in the design were: enzyme (trypsin)to-protein ratio (X1), digestion reaction time (X2), and digestion reaction temperature (X3  Table S3. To analyze the data, the peak area results for SRM 2925 and the IS were converted to z-scores. Results/Discussion. All 11 MRM peptides (SRM 2925 and IS) were observed in the eight (8) conditions ( Figure S3). The MRM peak area ratio (PAR) values and the performance of the IS for the 16 conditions across the MRM transition are consistent, as illustrated in Figure S4 and S5, respectively. Using the peak area results for both SRM 2925 (unlabeled) and IS to generate the z-score graph for each optimization condition, the optimal condition (highest z-score across all MRM peptides) for trypsin digestion of albumin (unlabeled and IS) was Optimization Sample #12 (enzyme (trypsin)-to-protein mass ratio (X1) of 1:30; digestion reaction time (X2) of 23 h; digestion reaction temperature (X3) of 37.0 °C) (Table S3).

Conclusion.
The DOE optimization approach was used to statistically determine the optimal trypsin digestion conditions for albumin to achieve the highest response from the NIST LC-MS/MS method. From the data, we observe that the optimal condition (highest z-score across all MRM peptides) for trypsin digestion of albumin (unlabeled and IS) was Optimization Sample #12. By optimizing the trypsin conditions for albumin and identifying the quantitative MRM peptides/transitions, we can determine the content of albumin in human urine samples with the highest degree of confidence and precision. The urine albumin trypsin digestion protocol and LC-MS/MS method are fit-for-purpose to accomplish value-assignment of the candidate NIST SRM 3666 Albumin and Creatinine in Frozen Human Urine material for urine albumin.