A Human Oral Fluid Assay for D- and L- Isomer Detection of Amphetamine and Methamphetamine Using Liquid-Liquid Extraction

Medical providers are increasingly confronted with clinical decision-making that involves (meth)amphetamines. And clinical laboratories need a sensitive, efficient assay for routine assessment of D- and L-isomers to determine the probable source of these potentially illicit analytes. This paper presents a validated method of D- and L-isomer detection in human oral fluid from an extract used for determination of a large oral fluid assay (63 analytes) on an older AB SCIEX 4000 instrument. Taken from the positive extract, D- and L-analytes were added. The method for extraction included addition of internal standard and a 2-step liquid-liquid extraction and dry-down step to concentrate and clean the samples. The samples were suspended in 50% MeOH in water, diluted with mobile phase, with separation and detection accomplished using LC-MS/MS to determine analyte concentration. Once samples were confirmed positive for (meth)amphetamine from the large oral fluid assay, they were further examined for the enantiomeric forms with 50 μl aliquots of the standards and samples of interest combined with 450 μl of D- and L-assay mobile phase, then analyzed using chiral column separation, and LC-MS/MS detection with standard curve spanning the range from 2.5 to 1000 ng/mL. The result is a sensitive and accurate detection of D- and L-isomers of amphetamine and methamphetamine in human oral fluid performed on an older model mass spectrometer (AB SCIEX 4000). The novelty of this assay is twofold (a) the 2-step liquid-liquid extraction and dry-down step to concentrate and clean the samples, and (b) its adoption characteristics as a reflex test from a large ODT panel without the need to invest in newer or expensive LC-MS/MS instruments. Finally, this assay also has potential to add a valuable option to high-throughput laboratories seeking a D- and L-testing alternative to urine drug testing methods.


Introduction
Amphetamines and methamphetamine are popular illicit drugs of abuse for their stimulation of the central nervous system. Tese central nervous system (CNS) stimulants exist as two enantiomeric forms, dexter (D-) or laevus (L-), which produce radically diferent efects on the human system. Notwithstanding the potential use disorder from the heighted dopamine response that D-(meth)amphetamine asserts [1,2], the L-form is quite common and is an efective vasoconstrictor used in the over-the-counter formulation of Vicks ® VapoInhaler ™ . Moreover, the D-form can be used therapeutically as a treatment for overeating disorders, narcolepsy, and attention defcit disorder, but produces the sought-after habit-forming central nervous system (CNS) efects that can be abused and long term neurotoxicity [2][3][4]. Tus, medical providers are increasingly confronted with clinical decision-making that involves (meth)amphetamines-including illicit use.
Routine assessment for noncompliance or nonmedical use of (meth)amphetamine is frequently accomplished through urine drug testing (UDT) based on risk of drug misuse, abuse, and diversion. Although UDT is considered the common practice for detecting scheduled drug compliance or illicit use, often medical providers are unable to procure a urine sample for various reasons. In this case, oral fuid drug testing (ODT) can serve as an efective alternative to UDT [5]. ODT is increasingly emerging as an alternative biological matrix for detecting drugs and monitoring patient medication compliance [6][7][8]. Although the detection window for oral fuid is small compared to matrices like urine, hair, or sweet, in certain clinical situations-like immediate detection of recent marijuana use-oral fuid shows to be more benefcial over UDT [9,10]. Te matrix allows for easy collection, but attention to recovery, stability, and dilutions issues of some collection devices should be given consideration for pharmacokinetic studies [10,11].
Although ODT opioid assays that use dilute-and-shoot methods with little sample manipulation have been developed and validated on AB SCIEX 4500 instruments with excellent calibration ranges (2.5-1,000 ng/mL) [12] robust ODT assays that also quantify the D-and L-isomers of (meth)amphetamine are less common. Notably, there are several methods available for determining D-and L-(meth) amphetamine including immunoassays which have been designed to react with one or the other but often have problems with cross-reactivity. Gas chromatography-mass spectrometry (GC-MS) can also be used but often requires an extra step requiring the synthesis of derivatives resulting in potential purity issues and errors in concentration estimation. Other methodologies including some liquid chromatography with tandem mass spectrometry (LC-MS/MS) methods also require purifcation steps with solid phase extraction cartridges involving additional equipment and the cartridges themselves adding extra cost per sample. Accordingly, there is a need for developing an efcient method to detect enantiomeric forms of D-and L-(meth) amphetamine, especially in small resource-limited laboratories without the capacity to invest in newer LC-MS/MS instruments.
In this paper, we ofer a novel approach for a fast, accurate, and applicable method to quantify the D-and Lisomers of (meth)amphetamine in human oral fuid specimens using liquid-liquid extraction and LC-MS/MS with an older model AB SCIEX 4000 instrument as an extension of larger ODT assay. A preprint version of this assay paper has been published [13]. Each patient sample was initially analyzed for 63 targeted analytes using LC-MS/MS with the same extracts injected a second time using a delta-9 tetrahydrocannabinol (THC)-specifc (ES) negative assay to capture the THC. Ten, for samples that exhibited a positive confrmation result for (meth)amphetamine, an additional sample was taken from the previously extracted specimen. Tis sample was then analyzed with a newly developed assay designed specifcally to assess the D-and L-isomer status to defne nonillicit versus illicit etiology. Te assay development and validation ofered here is for the beneft of highthroughput laboratories seeking novel solutions for a quantitative D-and L-isomer test from oral fuid with fast and accurate chemical analysis using less expensive older model AB SCIEX instruments. Although other methodologies may be more economic for this purpose, this assay provides small resource-limited laboratories with older model AB SCIEX 4000 instruments to operationalize discovery of enantiomeric forms of D-and L-(meth)amphetamine without incurring additional equipment expensive.

Methods and Materials
2.1. Reagents and Standards. All analyte stock solutions at 1 mg/mL concentrations and deuterated internal standards at 100 μg/mL were purchased from Cerilliant Corporation (Round Rock, TX, USA). All organic solvents including methanol, acetonitrile, formic acid (88%), dichloromethane, 2 propanol, and ethyl acetate were obtained from Fisher Scientifc (Pittsburgh, PA, USA). Because experiments have demonstrated considerable variation in recovery of (meth) amphetamine from various oral fuid collections devices [14], the Quantisal ™ oral fuid collection device was used based upon its exhibited recovery of (meth)amphetamine shown to exceeded 93%, and the device has seen good recovery in liquid-liquid extraction techniques [15,16]. Te Quantisal ™ oral fuid collection device and extraction bufer were obtained from Immunalysis Corporation (Pomona, CA, USA).

Mobile Phase and Extraction Solutions.
A D-and Lmobile phase (MPDL) solution was created by adding ∼993.2 mL of methanol to a 1 L bottle. Ten using a 1 mL pipettor, 5 mL of type I water, 1.5 mL of acetic acid, and 0.3 mL of ammonium hydroxide were added. Tis solution can be kept at room temperature for up to 1 year. Extraction solution 1 (ES1) was created with 50% dichloromethane and 50% 2-propanol by using a graduated cylinder under a fume hood. Equal volumes of dichloromethane and 2-propanol were added to a clean reagent bottle which was capped and mixed well. Extraction solution 2 (ES2) was created with 50% dichloromethane and 50% ethyl acetate by using a graduated cylinder under a fume hood. Equal volumes of dichloromethane and ethyl acetate were added to a clean reagent bottle which was capped and mixed well.

Standard
Preparation. An 8000 ng/mL stock solution was made by combining analyte stock controls and diluting it with MPA. In contrast, D-and L-amphetamine and methamphetamine were added in an amount to make a 4000 ng/mL stock of each isomer so that when combined they would produce an 8000 ng/mL solution of total amphetamine and methamphetamine. Tis means that the range of the D-and L standard curve (SC) is from 2.5 to 1000 ng/mL (half the concentration). Te resulting stock standard was diluted with mobile phase A (MPA) to produce the SC. Concentrations were 8000 (undiluted), 4000, 2000, 1000 400, 200, 100, 40, 20, 10, 4, and 2 ng/mL. Tese solutions were stored at the concentrations above. Tey underwent a dilution during the assay (1 part standard to 3 parts mobile phase and THC standard) to achieve the concentration desired in sample analysis with oral fuid (saliva). Te standards and quality control (QC) were diluted (0.5 mL) with 1.5 mL of extraction bufer. Tis approximates the condition seen with saliva after collection with the Quantisal ™ oral fuid sample collection device. Te fnal concentration in the 0.5 mL sample SC included the following points: 2000, 1000, 500, 250, 100, 50, 25, 10, 5, 2.5, 1, and 0.5 ng/mL.
Te assay QCs were made similarly; frst making a 7200 ng/mL spiking solution in MPA then diluting to 3200, 2400, 300, 60, 12, and 2 ng/mL. Te D-and L-amphetamine and methamphetamine QCs were made at half concentrations. Final concentrations of each QC were 1800, 800, 600, 75, 15, 3, and 0.5 ng/mL, after the 1 : 4 dilution with MPA and THC QC same as the SC points noted above.
Te internal standard working solution (ISWS) for the large oral fuid assay and the D-and L-assay was made by flling a 100 mL graduated cylinder to the 50 mL mark with 10% methanol in water and adding 250 μL of each of the internal standards listed above. Te volume was brought to 100 mL with additional 10% methanol producing a concentration of 250 ng/mL.

Analyte Optimization.
Individual analytes and internal standards were optimized by using T-infusion with 50% B mobile phase and tuning for declustering potential (DP), entrance potential (EP), collision energy (CE), and exit potential (CXP) at a fow rate of 0.7 mL/min. Te two most abundant fragments were selected for monitoring using multiple reaction monitoring (MRM).

Sample Preparation and Procedures.
Samples were collected using the Quantisal ™ oral fuid collection device.
Te manufacturers collections instructions were followed (https://immunalysis.com/products/oral-fuid/quantisal/). Te samples, standards, and QC were extracted using two liquid-liquid extractions with 1 : 1 DCM : IPA and 1 : 1 DCM : EtAc. Tey were combined, dried, reconstituted with 50% MeOH water, and combined with mobile phase for separation of the analytes. Sample preparation for D-and L-analysis by LC-MS/MS involved transferring 50 μL of the already extracted standards, QC, and any samples of interest to a new plate. Ten 450 μL of MPDL was added to each well and mixed with a multichannel pipette, the plate was covered with a plate mat and analyzed for the D-and L-isomers of amphetamine and methamphetamine using the listed chiral column. Te LC-MS/MS conditions and separation parameters are presented in Table 1.

Method Validation Procedures
Te assay was developed as an extension of a larger 63 analyte assay whereby samples that were confrmed positive for (meth)amphetamine could be refex tested to identify nonillicit versus illicit etiology. Te assay was initially optimized for extraction of all the required analytes in the larger panel. Samples were spiked with a combination of all the drugs of study and multiple solvent systems and single and two step processes were evaluated including the combinations fnally selected; these were dichloromethane plus isopropyl alcohol for solution 1 and dichloromethane plus ethyl acetate for solution 2 as these appeared to be the most broad-spectrum solvents for extraction of the diferent analytes tested-including methamphetamine and its metabolite amphetamine, and analytes with diferent chemical solubility characteristics such as meprobamate. Moreover, the D-and L-isomer component of this assay was adapted from a dilute and shoot method for analysis of urine samples and was modifed to the back end of the oral fuid assay after extraction. Importantly, it was found during development that the elimination of phentermine was achieved by both chromatographic separation and elimination by careful selection of the secondary transition.

Matrix Lot-to-Lot Comparison.
Individual lots of human matrix (saliva) difer according to a person's overall health and hydration status [17]. A single lot of oral fuid is not enough to demonstrate the ruggedness of the assay system when such variability in the matrix exists [18]. Due to this, and in accordance with the current College of American Pathologists (CAP) standards, a minimum of 10 lots of human matrix were collected from drug-free donors. Tese oral fuid samples were spiked at a low-level concentration with each analyte. Tese samples were prepared, extracted, and run as described above. Te responses were calculated and the analyte to internal standard (IS) ratio and %CV is shown in Table 2.

Analytical Measurement Range.
Te analytical measurement range (AMR) of the assay refers to the concentration range that the assay is validated within and is determined by running a series of calibration curve standards covering a concentration range that encompass the concentration of analyte expected to fnd in patient samples [19]. Te limits of the AMR were bounded by the lower limit of quantitation (LLOQ) and the upper limit of quantitation (ULOQ). Te dynamic range may be described by a linear or quadratic ft [19,20]. Calibration curves were created using a Journal of Analytical Methods in Chemistry 3 minimum of six nonzero calibration points. To be accepted as the AMR, all points describing the calibration curve must pass within ±20% of the nominal concentration [19]. Furthermore, the correlation coefcient (R 2 ) for the calibration curve must be ≥ 0.99, or R should be ≥ 0.98 to be acceptable [21,22].

Sensitivity.
Te sensitivity of the assay system refers to the ability to reliably produce a signal throughout the entire calibration range, but specifcally at the low-end of the calibration curve (the lower limit of quantitation, LLOQ) [23]. In hyphenated mass spectrometry assays, a signal that produces a signal to noise ratio (S/N) of ≥10 is considered valid for the LLOQ of an assay system [24]. Further, an S/N ratio of ≥5 is considered clear enough for the limit of detection. Te sensitivity of the assay system was tested by injecting 6 replicates of the LLOQ over 3 days and evaluating the resulting analytical determinations. Standard acceptance criteria of ±20% of nominal concentration apply.

Intraday Precision and Accuracy.
Intraday precision and accuracy were determined using six replicates of each of three QC sample determinations and LLOQ from across at least three validation runs. Concentrations of the QC samples ranged across the curve, with the low QC set at approximately 3 times the LLOQ or less, the mid QC near the mid-range of the linear range of the curve, and the high QC set at 80-90% of the ULOQ. Percent accuracy and precision were determined for each individual measurement. To be accepted, the precision and accuracy for the replicate determinations must be ≤ 20% at each level.

Interday Precision and Accuracy.
Interday precision and accuracy were determined using all replicates of each of three quality control (QC low, QC mid, and QC high) and LLOQ sample determinations from the analytical runs performed on 3 separate days. Concentrations of the QC samples ranged across the curve, with the low QC set around 3 times the LLOQ, the mid QC near the middle of the linear range, and the high QC set at 80-90% of the ULOQ. To be accepted, the precision and accuracy for the replicate determinations must be ≤ 20% at each level.   Journal of Analytical Methods in Chemistry 3.6. Exogenous Interfering Substances. Drugs that are known or suspected of interfering with similar bioanalytical systems should be evaluated to ensure that they do not suppress ionization or cause false-positive results for a given analyte [25,26]. Te following medications were evaluated: overthe-counter mix, acetaminophen, ibuprofen, pseudoephedrine, cafeine, and naproxen. Te following individual analytes were also tested: salicylic acid, phenylephrine, phentermine, diphenhydramine, and dextromethorphan. A high concentration of the possible interfering drug (typically 2,000 ng/mL or greater) was spiked into a low QC sample (15-75 ng/mL low QC). Acceptance criteria for a substance to be deemed as noninterfering is that the quantitated value for the low QC should be within ±20% of the nominal value [27]. Furthermore, the spiked substance should not cause a false-positive or a false-negative result.

Partial Volumes and Dilutions.
A spiked solution was created at a concentration above the ULOQ, in this case, 4000 ng/mL. Te sample was run at discrete dilutions of 1 : 5, 1 : 10, 1 : 20, and 1 : 50. Concentration determinations for all dilutions should be within ±20% of the nominal value following correction for the dilution factor [27,28]. More recent literature suggests that the signal to noise ratio of both the quantifcation trace and the qualifying ion trace be 3-10 [29]. On occasion, an analyte will not have a quantifying ion that passes this criterion while still permitting the quantifcation trace to remain in a meaningful range. Tese instances should be documented in the laboratory standard operating procedure or validation report.

Carryover.
Carryover is the presence of an analyte in a blank injection following a positive injection, resulting in a false-positive sample [30]. Te injection needle should be washed in-between samples with a needle wash solution that is intended to remove contamination from the surface of the needle. Te efciency of this process is monitored during validation by assessing carryover in the following manner. Samples are injected in the following sequence: high QC, wash, high QC, wash, high QC, wash. Peak areas are integrated for both the analyte and internal standard. Peak area in the wash solutions should be 0.1% or less of that found in the high QC standard. In addition, the mean of the peak area in the three wash solutions following the high QC replicates should be less than 20% of the LLOQ being used for the assay [30].

Interday Average back Calculated Calibration Standards.
Each validation run contained calibration standards with theoretical concentrations of 1, 2.5, 5, 10, 25, 50, 100, 500, 1000, and 2000 ng/mL of each of the analytes with an additional negative run at 0.5 ng/mL. Table 3 shows the range of standard curves of the combined amphetamine and the individual Dand L-analytes and the correlation information. D-and L-curve concentrations were half the above concentration ranging from 0.25 (neg) to 1000 ng/mL. Mean R values were all at least 0.99 indicating good ft to the data.  Table 7 indicate that all analytes can be diluted at all levels.

Room Temperature, Refrigerator, and Freezer Stability.
Samples with concentrations of 75, 800, or 1800 ng/mL were prepared in triplicate. One set was kept at room temperature overnight (RT), a second set was kept in the refrigerator overnight (RF), and a third set was kept in the freezer overnight (FZ). Tese validation samples were then run and compared to a triplicate preparation of QC samples that had been analyzed as normal. All results show less than 20% deviation from expected (Table 8).

Freeze-Taw (FT) Stability.
Validation samples with concentrations of 75, 800, or 1800 ng/mL were frozen at −20°C and thawed in sequence with samples taken after each freeze-thaw cycle for a maximum of three cycles. Tese validation samples were analyzed in triplicate and compared to a triplicate preparation of validation samples that had not been subjected to this freeze-thaw cycle. Te experimental results showed all meeting acceptance criteria.

Extracted Sample Stability.
A stability experiment was performed where samples were stored in the instrument (3 day) or refrigerator (7 day) and reinjected after 3 and 7 days. All samples were within 20% of the initial results.

Stability in Matrix.
A series of triplicate samples were analyzed over 7 days for stability at room temperature, 4°C and −20°C. Te results indicated that all analytes were stable for at least 7 days refrigerated and frozen. Te analytes were stable at room temperature for 24 hours. Table 9 indicates the efect of 10 diferent matrix lots tested by using a series of 7.5 ng/mL samples prepared in water, MPA, and 10 diferent matrices. Te results were acceptable with less than 20% CV across oral fuid, water, and MPA meeting acceptance criteria. Tis is likely due to dilution in 1.5 mL Quantisal ™ extraction bufer before extraction.     4.9. Selectivity. Multiple drugs that might have a potential for interfering with the assay analytes were run in the assay. Samples of 500 μL of 7.5 ng/mL QC were placed in a series of tubes to be run in triplicate. To the frst set, 50 μL of MeOH was added to act as the control. To the remaining tubes, 50 μL of sample containing dextromethorphan, diphenhydramine, phenylephrine, salicylic acid, or combo (includes acetaminophen, cafeine, chlorpheniramine, ibuprofen, naproxen, and pseudoephedrine). Tese solutions were obtained from Cerilliant and were at a concentration of 1 mg/mL each except for the over-the-counter mix which was 100 μg/mL. Each solution was diluted to 20 μg/mL in methanol and this solution was used to spike samples as indicated above. All samples met the accepted criteria and are diplayed in Table 10 and Table 11.

Discussion
Te determination of prescription medications and illicit substances is needed for medical compliance [31]. And human oral fuid is one of the most noninvasive and easily observed sample collection methods. It provides a relatively simple and reliable means of sample collection coupled with a reduced chance of sample adulteration. Oral fuid also provides a viable alternative for measurement in patients that cannot provide an adequate urine sample volume such as catheterized patients. Te drawbacks of the oral fuid assay are that it has a shorter detection window and requires a more sensitive assay. Accordingly, this paper demonstrates a developed and validated cost-efective means of analysis using older, less sensitive instruments (API SCIEX 4000) by using a 2 step liquid-liquid extraction method and concentration of the samples with a nitrogen dry-down and a resuspension step. Te method of development of this Dand L-assay was validated in accordance with the United States Food and Drug Administration and the College of American Pathologists guidelines [30] with an LLOQ of 2.5 ng/mL and ULOQ of 1000 ng/mL. Te most important aspect of this assay was its specifcity. It has the ability to reliably and defnitively diferentiate between the isomeric forms of methamphetamine and its metabolite amphetamine, as well as common decongestants and weight loss medication. Phentermine in particular is a positional isomer of methamphetamine that laboratories need to ensure does   not interfere in (meth)amphetamine confrmation. As a positional isomer, it shares a molecular weight and fragment pattern nearly indiscernible from methamphetamine by many LC-MS/MS methods. Tis assay showed good chromatographic separation of methamphetamine, amphetamine, and phentermine (Figure 1), permitting clear diferentiation between these positional isomers.

Conclusion
Te development and validation of an ODT assay designed specifcally to assess the D-and L-isomer status of (meth) amphetamine to defne nonillicit versus illicit etiology is presented. Te novelty of this assay is twofold (a) the 2step liquid-liquid extraction and dry-down step to concentrate and clean the samples, and (b) its adoption characteristics as a refex test from a large ODT panel without the need to invest in newer or expensive LC-MS/ MS instruments. Te assay is quite sensitive with a cutof of 2.5 ng/mL and has good precision and accuracy. Although other methodologies may be more economic for enantiomeric discovery, this assay provides small resource-limited laboratories with existing older model AB SCIEX 4000 instruments an efective way to operationalize the identifcation of D-and L-forms of (meth)amphetamine without incurring additional equipment expense. Finally, this assay also has potential to add a valuable option to high-throughput laboratories seeking a robust testing alternative to UDT methods.

Data Availability
https://doi.org/10.7910/DVN/CQTWE1, Harvard Dataverse. Note. Retention times are updated using the internal standard each time that the data are analyzed.