Candidate High-Resolution Mass Spectrometry-Based Reference Method for the Quantification of Procalcitonin in Human Serum Using a Characterized Recombinant Protein as a Primary Calibrator

Procalcitonin (PCT) is a widely used biomarker for rapid sepsis diagnosis and antibiotic stewardship. Variability of results in commercial assays has highlighted the need for standardization of PCT measurements. An antibody-free candidate reference measurement procedure (RMP) based on the isotope dilution mass spectrometry and protein calibration approach was developed and validated to quantify PCT in human serum. The method allows quantification of PCT from 0.25 to 13.74 μg/L (R > 0.998) with extension up to 132 μg/L after dilution of samples with PCT concentration above 13.74 μg/L. Intraday bias was between −3.3 and +5.7%, and interday bias was between −3.0 and −0.7%. Intraday precision was below 5.1%, and interday precision was below 4.0%. The candidate RMP was successfully applied to the absolute quantification of PCT in five frozen human serum pools. A recombinant PCT used as a primary calibrator was characterized by high-resolution mass spectrometry and amino acid analysis to establish traceability of the results to the SI units. This candidate RMP is fit to assign target values to secondary certified reference materials (CRMs) for further use in external quality assessment schemes to monitor the accuracy and comparability of the commercially available immunoassay results and to confirm the need for improving the harmonization of PCT assays. The candidate RMP will also be used to evaluate whether the correlation between the candidate RMP and immunoassays is sufficiently high. Overall, this candidate RMP will support reliable sepsis diagnosis and guide treatment decisions, patient monitoring, and outcomes.

Human serum (reference H4522) was purchased from Sigma Aldrich (St. Quentin Fallavier, France). The material was obtained from healthy male subjects and is noted blank serum afterward. Its PCT concentration was below the limit of detection (0.02 µg/L) of commercially available immunoassay (Roche Cobas).

Document 2: Preparation of solutions
All steps of preparation of solution were performed gravimetrically using calibrated precision balances. Therefore, its concentration is expressed by mass of compound of interest per mass of solution, except for final samples (calibrant, quality control) where its concentration is converted in µg/L in concordance with clinical use.

Stock solutions and working solutions
Individual primary stock solutions were gravimetrically prepared in human serum for Met-PCT [3-116] and SIL-protein Met-PCT at 12.18 µg/g and 14.23 µg/g, respectively. Solutions were aliquoted and stored at -80°C ± 10°C.
The two individual primary stock solutions of Met-PCT [3-116] and SIL-protein Met-PCT were diluted in human serum to obtain individual working solutions of unlabelled protein and labeled protein at 22.15 ng/g and 44.93 ng/g, respectively. Solutions were aliquoted and stored at -80°C ± 10°C.

Protein calibrators and internal quality control materials
Six points calibration curves were prepared by mixing an increasing volume of Met-PCT  working solution with a constant volume of SIL-protein Met-PCT working solution to obtain ratios ranging from 0.14 to 12.10 (corresponding to a PCT concentration from 0.25 µg/L to 13.74 µg/L).
Three quality control (QC) materials were prepared by mixing three different volumes of Met-PCT  working solution and a constant volume of SIL-protein Met-PCT working solution to reach an amount ratio of 0.51, 2.03, and 5.08 (corresponding to a PCT concentration of 0.88, 3.61 and 8.93 µg/L, respectively). The Met-PCT [3-116] working solutions used for calibration solutions and QC materials were prepared from two different aliquots of the individual primary stock solutions.

Protein samples for dilution test
PCT samples at 132 µg/L were prepared by diluting Met-PCT  working solution in human serum. This solution was then diluted with human serum by a factor of about 20 to prepare six PCT samples at 6.77 µg/L. The diluted PCT samples were then mixed with SIL-protein Met-PCT working solution to reach a concentration of 6.52 µg/L.

Document 3: Method validation
Linearity: The calibration curves were obtained with non-zero five to six calibrators depending on peptide by plotting the ratios of the peak areas of the peptides over their isotopically labelled counterparts against the corresponding ratios of the amounts of the peptides over their isotopically labelled counterparts. Linearity was evaluated by three sets of calibration samples prepared from three independent aliquots of Met-PCT [3-116] stock solutions that were then diluted in the serum samples obtained from healthy subjects. The linearity was evaluated according to the Pearson correlation coefficient, which was required to be >0.995. Besides, the accuracy of the quantification of each calibrant was evaluated by considering the calibrants as unknown samples. The bias between the measured concentration and the theoretical concentration determined gravimetrically should be within ± 20% for the LLOQ level and ± 15% for other levels according to the guideline of bioanalytical method validation from FDA and EMA 1,2 . For each batch, a maximum of one calibrator was excluded from the curve. The individual concentration per peptide was determined using its respective calibration curve; the mean concentration was calculated as the average of two individual concentrations.
Trueness and precision: The trueness and precision were evaluated in intermediate precision conditions of measurement using three processed replicates of QC materials at three concentration levels (about 1.0, 4.0 and 9.0 µg/L) over three independent experiments (intra-assay: N=3 per concentration level, inter-assay: N= 9 per concentration level). The concentration of each sample was determined using the calibration curve prepared on the same day. Trueness was evaluated by the bias between the measured concentration and the theoretical concentration determined gravimetrically. The precision was calculated as the coefficient of variation (CV) of the measured concentrations per peptide and the mean concentration of two peptides. Bias and precision (CV) should be within ±15 % according to the guideline of bioanalytical method validation from FDA and EMA 1,2 .
Lower limit of quantification: The LLOQ was defined as the lowest concentration of the calibration curve with bias and precision within 20% according to the guideline of bioanalytical method validation from FDA and EMA 1,2 . The LLOQ was assessed by analyzing six processed replicate samples at 0.25 µg/L for SAL peptide and 0.5 µg/L for FHT peptide.
Higher limit of quantification: To determine whether a serum sample with a PCT concentration exceeding the highest concentration of the calibration curve can be diluted before being processed as described above, an experiment was conducted by diluting 20-folds with human serum obtained from healthy subjects. A spiked QC material (HLOQ) was gravimetrically prepared at a concentration of about 132 µg/L, which is 9-folds higher than the highest concentration of the calibration curve. The experiment was performed with six processed replicate samples. The bias and precision of the diluted samples should be within ± 15% according to the guideline of bioanalytical method validation from FDA and EMA 1,2 .
Autosampler stability: The auto-sampler stability (7 C) of the final extract from QC materials was evaluated by comparing the concentration obtained at the beginning of the study (day 0) and those obtained 7 days later. The final extract was considered stable if the concentration was quantified within 20% of the initial concentration (day 0). Carryover: Carryover is the ratio (in percentage) of peak area of the calibrator in blank sample acquisition, just after a prior injection of the calibrator at the highest concentration over the peak area of the LLOQ level. Carry-over should not exceed 20%.  Each condition being evaluated in four processed replicates using one of the aliquots provided by the manufacturer. Table S5. AAA results obtained from 29 processed replicates over six independent gas-phase hydrolyses of the primary calibrator stock solution.
The primary calibrator stock solution was obtained by combining ten aliquots provided by the manufacturer in order to avoid inhomogeneity between aliquots. Cal 0 0 0 -- The uncertainty of the linear regression was obtained using Excel (Microsoft Office). The uncertainty associated with the measurement result obtained from the linear calibration was calculated by the random error associated with both the intercept and the slope.   (A) Linearity of the signal response for SAL peptide obtained with non-zero six points serum-based calibrators from three independent experiences (linearity equation was obtained by the average of results obtained from three independent days); (B) Percentage of deviation (%) of the back-calculated ratio compared to the theoretical quantity ratio of the SAL peptide; (C) Linearity of the signal response for FHT peptide obtained with non-zero five points serum-based calibrators from three independent experiences (linearity equation was obtained by averaging results obtained from three independent days); (D) Percentage of deviation (%) of the back-calculated ratio compared to the theoretical quantity ratio of the FHT peptide. Deviation % from all calibrator levels were between ±15%.