Arginine, as a Key Indicator for Real-Time Stability Monitoring of Quality Control in the Newborn Screening Test Using Dried Blood Spot

: Dried blood spots (DBS) have advantages such as minimizing blood collection volume and the distress to neonate. DBS have been used for tandem mass spectrometry (MS/MS)-based newborn screening tests (NST) of amino acid (AA) and acylcarnitine. The Newborn Screening Quality Assurance Program (NSQAP) have been provided quality control (QC) materials for MS/MS, as DBS cards. The NSQAP is generally provided within 14 months of the shelf life and the recommended storage condition is at − 10 ◦ C to − 30 ◦ C. Previously, several accelerated degradation studies had been performed to determine the transportation stability and short-term stability of AAs and acylcarnitines in DBS. However, the experimental condition is markedly different to the storage condition. We performed long-term monitoring for the real-time stability of seven AAs and 14 acylcarnitines from three levels of 2012 NSQAP QC materials across a time period of 788 days. Arginine suddenly yielded a catastrophic degeneration pattern, which started around D300. When comparing this with previous accelerated degradation studies, methionine, tyrosine, citrulline, and acetylcarnitine did not show a remarkable measurand drift for the real-time stability, except for arginine. Our study showed that arginine would require intensive QC monitoring in routine practice, and should be used for the assessment of the stability in long-term storage of DBS samples for biobanking.


Introduction
Inborn errors of metabolism (IEM) are treatable disorders for which timely treatment is critical to prevent mortality and to improve the outcome [1][2][3]. The newborn screening test (NST) is helpful for early detection of IEM [4], and it has been performed annually for approximately 10 million neonates worldwide [5]. In the 1990's, tandem mass spectrometry (MS/MS)-based NST for amino acid (AA) and acylcarnitine (AC) was developed and was implemented as a routine test for the simultaneous screening of various IEM to be carried out in an efficient and cost-effective manner [6,7]. Dried blood spots (DBS) have advantages such as minimizing blood collection volume and the distress to neonate [8], therefore DBS has been used for MS/MS-based NST [9,10]. Biobanking for residual DBS specimens may be useful for additional diagnostic use in unexpected causes of IEM for the child and family, research projects, and in the development of new NBS assays [11].
MS/MS-based NST is easily modified according to the laboratory's demands [12]. These in-house modifications, which are categorized as laboratory-developed tests (LDTs) cause the difficulty of quality assurance (QA) and inter-laboratory comparison [13], because there are no reference methods of MS/MS-based NST [14]. The Centers for Disease Control and Prevention (CDC) have operated the Newborn Screening Quality Assurance Program (NSQAP) for quality management of NST [15]. The NSQAP have provided QA services including quality control (QC) materials to participating laboratories. The NSQAP QC materials for MS/MS had been provided as DBS cards containing a basal pool and a basal

Sample Preparation for Extraction and Derivatization
The DBS cards were punched as a 3.2 mm spot disc using DBS puncher (PerkinElmer, Waltham, MA, USA). The punched spot was extracted with 100 µL IS working solution for 30 min at ambient temperature in 96-well microplates shaken using a plate shaker (Wallac, Turku, Finland). IS working solution was diluted with 100-fold IS stock solution with methanol, which was consisting of one vial of amino acid reference standards with 50% methanol 1 mL and one vial of carnitine/acylcarnitine reference standards kits with 100% methanol 1 mL. As the QC solution, MassCheck Amino Acids, Acylcarnitines, Succinylacetone Dried Blood Spot Controls were added into two empty wells of plate. After extraction, mixtures were transferred to new microplate and were dried by nitrogen gas for 15 min at 40 • C. Derivatization of analytes was performed by adding 50 µL of 3 N HCL in N-Butanol into each well and second drying for 20 min at 40 • C. After the derivatization procedure, 75 µL of redissolved solution was added into each well and centrifuged at 1500× g for 5 min at ambient temperature.

LC-MS/MS Analysis
The analytes in blood spots were analyzed in the same day at the sample preparation using the API 2000 triple quadrupole LC/MS/MS (AB SCIEX, Framingham, MA, USA) equipped with an electrospray ionization (ESI) source in positive scan mode. The samples in each well were injected 10 µL at a flow rate of 0.08 mL/min using autosampler with flow gradient, and mobile phase solution consisting of 80% acetonitrile and 0.01% formic acid. A 5.5 kV ion spray voltage at 300 • C was applied to the ESI system. Pressures for curtain gas, ion source gas 1, and ion source gas 2 were 20, 60, and 60 psi, respectively. Neutral loss scan mode of m/z 102 and of m/z 85.1 were used for the analysis of most AAs and ACs. The multiple reaction monitoring (MRM) scan mode was used for Arg (Q1/Q3: 231.2/70.1) and Cit (Q1/Q3: 232.2/113.1), respectively. ChemoView 1.4.2 software (AB SCIEX, Framingham, MA, USA) was used for AAs and ACs quantification.

Real-Time Stability Monitoring
Day zero (D0) was set to 3 September 2012, which was the 55th day after the samples were received and stored. According for NSQAP's recommendation, the shelf life of these material was considered until 31 August 2013 (D364). Single measurement of NSQAP QC materials per each batch was performed between 3 September 2012 and 31 October 2014 (D778). The initial data was obtained by collecting data during first two month runs, from D0 to D60. Using these data, the initial mean of each analyte was calculated and was compared with peer group data with derivatized-MS/MS non-kit. According to CLSI EP25, acceptable differences criteria was set to extra 20% of the initial mean [23]. We alternatively assessed the allowable drift of the real-time stability by the comparison of predicted value and measured value to acceptable differences criteria, when compared to the initial mean. Between D0 and D778, commercial QCs were changed twice, on D81 and on D413. Schematic time course of NSQAP QC materials and study design of real-time stability test were illustrated in Figure 1.

Statistical Analysis
ROUT method was used for removing outliers. Student's t-test was used for comparing measured means between initial mean and within the shelf life or after expired shelf life. Linear regression was performed for predicting the time point of in-use stability metrics. Pearson correlation was used for assuming the relationship between time and measured value of each analyte. The Statistical Package for the Social Sciences version 25.0 (IBM Corporation, Armonk, NY, USA) and the GraphPad Prism (version 9.1.2.; GraphPad Software, La Jolla, CA, USA) were used for statistical analyses and graphs. p value of less than 0.05 was considered significant.

Characteristics of Routine Practice
A total of 331 procedures were performed between September 2012 and October 2014. Of 331 data pieces, one outlier was removed for statistical analyses,using the ROUT method (Q = 0.5). 151 procedures were performed within the shelf life from D0 to D364 (week1 to week52), while 179 were performed after the shelf life from D365 to D788 (week53 to week113). Each interval of procedures was as same, 2.4 day regardless of the shelf life.

Statistical Analysis
ROUT method was used for removing outliers. Student's t-test was used for comparing measured means between initial mean and within the shelf life or after expired shelf life. Linear regression was performed for predicting the time point of in-use stability metrics. Pearson correlation was used for assuming the relationship between time and measured value of each analyte. The Statistical Package for the Social Sciences version 25.0 (IBM Corporation, Armonk, NY, USA) and the GraphPad Prism (version 9.1.2.; GraphPad Software, La Jolla, CA, USA) were used for statistical analyses and graphs. p value of less than 0.05 was considered significant.

Characteristics of Routine Practice
A total of 331 procedures were performed between September 2012 and October 2014. Of 331 data pieces, one outlier was removed for statistical analyses, using the ROUT method (Q = 0.5). 151 procedures were performed within the shelf life from D0 to D364 (week1 to week52), while 179 were performed after the shelf life from D365 to D788 (week53 to week113). Each interval of procedures was as same, 2.4 day regardless of the shelf life.

Method Validation
The repeatability and total imprecision were evaluated with replicated measures at three levels over 20 days. The repeatability of each analytes varied from 2.3% to 19.9% of coefficient of variation (CV), and total imprecision varied from 3.8% to 19.6% of CV, respectively. The matrix effect of each analyte ranged from 76.7% to 121.6%, and extraction recovery ranged from 88.6% to 137.0%, respectively.

Comparison between the Initial Data and Peer Group Data with Derivatized-MS/MS Non-Kit
A comparison between the initial data and peer group data with derivatized-MS/MS non-kit [24] was performed. The initial mean and SD were calculated, using 25 obtained data from D0 to D60. The coefficient of variation (CV) of each analyte varied from 4.0% to 15.7%. Of 21 analytes, means of 20 were compared to peer group data, except for Arg. Among 20 analytes, 18 revealed standard deviation index (SDI) within ±2.0 in all three levels, except for Phe and C0 (Phe medium, 256.9 µmol/L vs. 190.8 µmol/L, SDI 2.40; C0 low, 39.7 µmol/L vs. 28.9 µmol/L, SDI 2.06; Phe medium, 56.8 µmol/L vs. 42.0 µmol/L, SDI 2.04) [24].

Real-Time Stability Analysis
Using data obtained from D0 to D788, we predicted the allowable time point as D542 for real-time stability of NSQAP QC materials. We calculated the predicted measurand drift of D542 (eD542) and the predicted percent difference of D542 compared with the initial mean (e%D542). The real-time stability of NSQAP QC materials was reviewed by the comparison between e%D542 and the measured percent difference of D542 compared with the initial mean (%D542). The results of real-time stability analysis are summarized in Table 2. Of 21 analytes, most analytes except for Arg were predicted that eD542 met within the allowable acceptance criteria, ±20% difference compared with the initial mean. Among 21 analytes, 18 analytes were predicted that e%D542 was within ±10% difference compared with the initial mean, while three analytes, Leu, Arg, and C5DC were predicted that e%D542 was over ±10% (Leu low, 10.3%; Arg low, 15.9%; Arg high, −20.1%; C5DC high, −10.8%, respectively) ( Table 3 and Figure 2). Each mean of %D542 and 4 neighborhoods between D538 and D546 (%D542 adj ) was compared with paired e%D542. Overall, e%D542 yielded 0.2% lower estimations rather than %D542 adj . Among 21 analytes, 20 analytes could meet the allowable acceptance criteria, while Arg could not meet the criteria (Arg high, −24.5%; Arg low, 30.2%).

Discussion
Quantitation of AAs and ACs for MS/MS-based NST using the DBS requires a degree of uniformed absorbance volume of analytes on the DBS card [25]. Hematocrit and filter paper are most critical factors for determining absorbance volume and the chromatographic distribution of the analytes on the DBS [26]. Hematocrit might influence flux and diffusion properties of the blood because it has a profound effect on viscosity [27]. When hematocrit was increased from 40% to 65%, serum volume per 3.2 mm disk decreased by 27%, while a diameter of spot decreased by less than 8% [28]. Filter paper might show different levels of imprecision according to individual products and their lots [29]. To overcome the hematocrit effect, a calibration strategy for quantitation in the DBS of patients with limited blood volume less than 50 µL and unknown hematocrit level had been suggested [27]. Moreover, the DBS spot made from verified filter paper with intact blood of fixed volume and fixed hematocrit level could reveal the acceptable quality and homogeneity, by controlling the hematocrit effect. The NSQAP QC materials were made using 100 µL of washed intact red blood cells at a 55% hematocrit with FDA approved filter papers [30]. The mean serum absorbance volumes for used lots of filter paper were ranged from 1.44 µL to 1.49 µL per 3.2 mm disk for intact red blood cells, in the 2012 NSQAP QC materials [24]. Thesevalues were within the acceptable range from CLSI NBS01 ED7, 1.454 ± 0.11 µL [25]. In current study, the concentration of analytes from the qualified DBS cards was measured, using the ratio between the peak intensity of analyte and that of the stable isotope labeled IS.
Verification of the real-time stability in the clinical laboratory is generally disturbed by bias or variability due to various factors such as instrument hardware changes and laboratory environment fluctuations over the study duration [23]. We compared commercial QCs and Arg high, and reviewed the historical chart and the nomogram (Figure 3). The results revealed that there was no definitive measurement error during the study, such as calibration failure or rejected QC, except for one procedure. One occurred by a specimen injection problem, and was removed by an outlier identification process. Our data showed periodic fluctuation patterns, which may highlight or mask the effect of measurand drift, without an instrument replacement. Lot 2912 and lot 2613 of commercial QC for AAs showed the similar fluctuation pattern of Arg high, while no remarkable measurand drift or correlation with time was observed in the commercial QCs, contrast to Arg high ( Figure 3). In the current study, the discrepancies between real-time measured values and predicted values from a linear regression analysis might be results from variability due to laboratory environment fluctuations.  Temperature and humidity are factors that affect the denaturation of analytes in DBS during storage or transport to the laboratory [17][18][19][20]31]. NSQAP QC materials were repeatedly exposed to ambient temperature and altered humidity during each routine procedure, though exposure time and manner were minimized. While several excursions to room temperature were not affected for stability of DBS [32], the high humidity might have a critical role for stability of analytes from DBS. The accelerated degradation study had been conducted at 37 °C for predetermined intervals in low-humidity and high-humidity environments [17]. According for the research by Adam et al., Arg had been re- Temperature and humidity are factors that affect the denaturation of analytes in DBS during storage or transport to the laboratory [17][18][19][20]31]. NSQAP QC materials were repeatedly exposed to ambient temperature and altered humidity during each routine procedure, though exposure time and manner were minimized. While several excursions to room temperature were not affected for stability of DBS [32], the high humidity might have a critical role for stability of analytes from DBS. The accelerated degradation study had been conducted at 37 • C for predetermined intervals in low-humidity and high-humidity environments [17]. According for the research by Adam et al., Arg had been reported as the most sensitive AA to the effects of high humidity, and Arg lost 95% of its concentration at 37 • C within 35 days in the over 50% humidity condition [17]. The research also showed that galactose-1-phosphate uridyltransferase, biotinidase, succinylacetone, C2, malonylcarnitine, decenoylcarnitine, and tetradecenoylcarnitine lost 95% of their concentration within 25 to 35 days in the same condition [17]. However, C2 showed a negligible negative correlation with time, and other analytes were not measured in our study ( Table 4). The gaps between accelerated degradation study and real-time stability monitoring were significantly observed in this study. Therefore, we agree with the suggestion of a recent study that it is necessary to highlight the limitations of the accelerated stability model [21]. The storage stability of DBS specimens is an important issue for biobanking or second usage of residual samples. However, the long-term stability studies which were conducted at −20 • C for AAs and ACs in the DBS were limited [33,34]. Our data showed that most AAs and ACs have storage stability at −20 • C at least 2 years, and these results were in line with the previous studies [33,34]. However, the studies had not included Arg, while high susceptibility of Arg in DBS to heat-related and humidity-related degradation was reported [17].
In the current study, Arg showed different degeneration patterns between a low and high level. To evaluate the initial mean, we compared the initial mean and Y-intercept from each linear regression equation. All of the CVs were within 15% for the three level of 21 analytes over 778 days, including CV of Arg low. Most analytes showed no significant difference between the initial mean and Y-intercept, except for Arg low (initial mean vs. Y-intercept, 9.0 µmol/L vs. 9.8 µmol/L, p = 0.0001). Furthermore, Arg low showed a negligible positive correlation with time (ρ = 0.26, p < 0.001). These figures indicated that the predicted measurand drift of Arg low was overestimated due to the undervalued initial mean of Arg low ( Figure 2U).
The current study has a limitation in assessing sample storage stability, because the study was performed using readymade QC material, not freshly prepared DBS. To assess the storage stability, sources of variation related to sample collection, transport, and storage would be clearly evaluated [35]. However, uncertainties around transport and pre-storage duration might disturb the assessment of short-term stability for DBS. A recent study showed that the decline of AAs and ACs in the freshly prepared DBS primarily occurred from one to three months of storage [36]. However, the storage instability of DBS samples in a short time period could not be observed in this study, because D0 was the 55th day after the samples were received and stored at −20 • C. According for the nomograms, a linear regression line or a quadratic regression curve is not suitable for reflecting a degeneration pattern of AAs and ACs. High order kinetics including humidity factor with a polynomial curve [37] may be appropriate for the interpolation of continuous measured values of 21 analytes. Notably, Arg high suddenly yielded a catastrophic degeneration pattern, which started around D300 ( Figure 2U). Further trend analysis using other lots of NSQAP QC materials for AAs high should be performed for elucidating whether this pattern and trend are observed at similar times. Taken together, these results indicate that there is a need for QC monitoring of MS/MS-based NST, including Arg high, and, furthermore, more intensive monitoring is required for Arg.

Conclusions
In conclusion, we showed that the real-time stability of NSQAP QC materials was allowable until D542, longer than 14 months from shipping date, as the shelf life from NSQAP's recommendation. There was a discrepancy of relatively sensitive analytes between our real-time stability monitoring and previous accelerated degradation study. In the current study, Arg showed the remarkable measurand drift from D300 to D778, therefore Arg would require intensive QC monitoring, in routine practice of MS/MS-based NST. Additionally, our study suggests that Arg should be used for the assessment of the stability in long-term storage DBS samples for biobanking.