A Practical Approach to Determination of Laboratory GC-MS Limits of Detection*

Transcription

1 A Practical Approach to Determination of Laboratory GC-MS Limits of Detection* Paula J. Underwood t, Gerald E. Kananen, and Edwin K. Armitage Forensic Toxicology Drug Testing Laboratory, 2490 Wilson Street, Fort George G. Meade, Maryland [Abstract [ Determination of limit of detection (LOD) values in a forensic laboratory serves a fundamental forensic requirement for assay performance. In addition to demonstrating assay capability, IOD values can also be used to fulfill certification requirements of a high-volume forensic drug laboratory. The LOD was defined as the lowest concentration of drug that the laboratory can detect in a specimen with forensic certainty at a minimum of 85% of the time. Overall batch acceptance criteria included acceptable quantitation of control materials (within 20% of target), acceptable chromatography (symmetry, peak integration, peak shape, peak, and baseline resolution), retention time within +1% of the extracted standard, and mass ion ratios within of the extracted standard mass ion ratios. Individual specimen acceptance criteria were the same as the batch acceptance criteria excluding the quantitation requirement. Data were collected from all instruments on different runs. A minimum of ten data points was required for each certified instrument, and a minimum of 85% of data points was acceptable. Quantitation within _+20% of the LOD concentration was not required, but acceptable mass ratios were required. with poor chromatography (internal standard failed mass ratios; interference of the baseline, for example, shoulders; asymmetry; and baseline resolution) was omitted from the acceptable rate calculation. with good chromatography with failed mass ion ratios were included in the acceptable rate calculation. With these criteria, we established the following LODs: 11-nor-Ag-tetrahydrocannabinol-9-carboxylic acid, 2 ng/ml; benzoylecgonine, 5 ng/ml; phencyclidine, 2.5 ng/ml; amphetamine, 150 ng/ml; methamphetamine, 100 ng/ml; codeine, 500 ng/ml; and morphine, 1000 ng/ml. Introduction Determination of the limit of detection (LOD) in a forensic laboratory is a fundamental forensic requirement to document assay performance. The College of American Pathologists Forensic Urine Drug Testing Checklist requires that a certified * The opinions expressed by the authors are not necessarity those of the Department of Defense or the Department of the Army, but are solely the opinions of the authors. t Author to whom correspondence should be addressed. laboratory determine a limit of detection for each gas chromatography-mass spectrometry procedure (GC-MS) (1). However, the Forensic Urine Drug Testing Checklist does not define how a laboratory should approach the determination of the LOD. The International Federation of Clinical Chemistry defines the detection limit as "the smallest single result which, with a stated probability (commonly 95%), can be distinguished from a suitable blank" (2). The International Union of Pure and Applied Chemistry defines the limit of detection similarly as the lowest concentration or amount that an analyst can detect to be statistically different from an analytical blank (3). The literature suggests two different approaches, statistical and empirical, for determining an LOD. In the statistical approach, a series of blanks without the analyte of interest is analyzed to determine mean background response and variation of the response. The LOD is calculated from the blank mean by adding two to four standard deviation units (4). This approach establishes an LOD that should be detectable from zero with a defined statistical level of confidence, but it only defines "the ability to measure nothing" (5). The empirical approach to determining the LOD involves measuring samples of decreasing concentrations of the analyte of interest to establish the lowest concentration that a method can detect with a consistent response to the actual solution concentration (4). This defines the point in a method where "analysis becomes just feasible" (2). LOD determination is a fundamental forensic requirement for a urine drug-testing laboratory. Accrediting organizations require laboratories to determine the LOD of a method as part of the method- or instrument-certification process. How to accomplish this is a practical question that any urine drugtesting laboratory must answer. A practical approach to the determination of LOD values should be realistically achievable and challenge assay and instrument capabilities under routine instrument operating conditions to produce supportable qualitative and quantitative data, and it should provide a baseline of analytical performance for daily operation. In a production laboratory, the number of data points needed to establish an LOD and the amount of time that can be invested without using valuable resources from routine testing are 1 2 Reproduction (photocopying) of editorial content of this journal is prohibited without publisher's permission.

2 Journal of Analytical Toxicology, Vol, 21, January/February 1997 important considerations. The Forensic Toxicology Drug Testing Laboratory (FTDTL) is a Department of Defense- (DoD) certified laboratory operated by the Department of the Army for deterring the use of drugs of abuse in uniformed services personnel. The FTDTL uses DoD mandated cutoff values to determine positive samples, which in some cases differ from the United States Department of Health and Human Services cutoff values. The FTDTL started the determination of a laboratory LOD by creating a definition of "limit of detection" to fit laboratory needs. Information, outlined by Armbruster et al. (6), was used as a nucleus to commence currently documented certification procedures for a battery of GC-MS instrumentation. This approach was modified to accommodate a larger number of GC-MS instruments and a significantly higher work load. FTDTL currently has a workload of approximately 500, ,000 samples per year. Stringent quality control certification for the vast battery of quality control material to be certified, coupled with increasing requirements by DoD for reduced turn around time, led to internal revisions for documenting reliable instrument performance characteristics. Table I. Summarized Extraction Details Drug Sample size (mt) Internal standard THC* 5.0 THC-d3 BZE 4.0 BZE-d 3 PCP 5.0 PCP-ds COD 1.0 COD-d3 MOR 1.0 MOR-d3 AMP 3.0 AMP-d s MAMP 3.0 MAMP-d11 * Abbreviations: THC, 11-nor-Ag-tetrahydrocannabinol-9-carboxylic acid; BZ[, benzoyecgonine; PCP, phencyclidine; COD, codeine; MOR, morphine; AMP, amphetamine; MAMP, methamphetamine. Table II. Summarized GC-MS Details Drug Ions monitored Derivatizing reagent THC + 313, 357, 360, 372*, 375 TMAH: DMSO lodomethane BZE 224%227,272,345,348 PCP 186,200',205,243,248 COD 234, 237, 343, 371', 374 MOR 236, 239, 401,429*, 432 AMP MAMP 91,118,123,240', ,210,213,254',260 TMAH: TEPAH: DMSO 1-1odobutane none BSTFA with I% TMCS BSTFA with I% TMCS HFBA HFBA Materials and Methods * Quantitating ion f Abbreviations: THC, 11-nor-Ag-tetrahydrocannabinol-9-carboxylic acid; BZE, benzoyecgonine; PCP, phencyclidine; COD, codeine; MOR, morphine; AMP, amphetamine; MAMP, methamphetamine. The FTDTL used the methods and instruments certified for use with military service member specimens. LODs were determined for the following assays: 11-nor-Ag-tetrahydrocannabinol- 9-carboxylic acid (THC), benzoylecgonine (BZE), phencyclidine (PCP), amphetamine (AMP), methamphetamine (MAMP), codeine (COD), and morphine (MOR). Internal standards were deuterated analogues of the compounds being assayed~ THC was extracted on AX solid phase columns (Creative Technology, Newark, DE). BZE was extracted on RP-W solid phase columns (Creative Technology). THC and BZE assays used an automated sample processor (Creative Technology X-TRXTM). Opiates and PCP were extracted on solid-phase columns (Worldwide Monitoring, Horsham, PA). The amphetamines procedure used a liquid-liquid extraction. Additional information about the extraction procedures is summarized in Table I. All analyses were performed on Hewlett-Packard (Atlanta, GA) 5890, 5970, 5971, or 5972 GC-MS systems using selected ion monitoring (SIM). All GC columns were HP Ultra 2, 0.2 rnm x 12 m (5% phenyl methyl silicone). Additional information about the GC-MS assays is summarized in Table [I. The analytical batch consisted of a calibrator at the DoD cutoff (single-point calibration), a low control at 40% of the DoD cutoff, a negative control (0 ng/ml), a high control at % of the DoD cutoff, a nonextracted specimen at the DoD cutoff, plus up to eight LOD specimens. The batch was considered acceptable when all concentrations of the controls were within of target, retention times were within of the extracted standard, mass ion ratios were within of the extracted mass ion ratios, and acceptable chromatography criteria were met. Acceptable chromatography criteria included symmetry, peak integration, peak shape, peak, and baseline resolution. The LOD specimens were evaluated as previously described for the analytical batch, but the requirement for concentration within of target was not required. The first batch analyzed for LOD contained various concentrations designed to narrow the focus for LOD concentration. The concentrations for the LOD specimens were obtained by serial dilution of certified control materials used at DoD cutoff the FTDTL. The requirements set to declare (ng/mt) a specific concentration at the LOD wereas follows: 1) LOD definition: the lowest con- 15 centration of drug that this laboratory will detect in a sample with forensic certainty at 100 a minimum of 85% of the time. The lowest detectable concentrations meeting the fol- 25 lowing criteria were GC-MS review criteria, 2ooo quality control review criteria, and mass ion ratios within of calibration. 2) Analyt- 4ooo ical criteria: a minimum of two analytical 5oo (batch) runs per certified instrument; min- 5OO imum of ten data points per certified instrument; acceptance of a minimum of 85% of data points; failed mass ratio data points for drug ions, which are included as a failed data point in the acceptable rate calculation; and 13

3 data points with poor chromatography (internal standard failed mass ratio; interference of the baseline, for example, shoulders; asymmetry; and baseline resolution), which are omitted for the acceptable rate calculation. The acceptable rate calculation was calculated for each instrument as follows: [(A)/(T-C)] x 100 =,4/? where A is the number of acceptable points, T is the total number of points, C is the number of points rejected for unacceptable chromatography, and,4/? is the acceptable rate, which is a percentage expressing achievability of the LOD concentration. Results The laboratory established a LOD for THC of 2 ng/ml (Table III) and for benzoylecgonine of 5 ng/ml (Table IV) with acceptable rates greater than 92% on all instruments. The laboratory established LODs for amphetamine of 150 ng/ml (Table V) and methamphetamine of 100 ng/ml (Table V) with acceptable rates at 100%. The laboratory established LODs for codeine of 500 ng/ml (Table VI) and morphine of 1000 ng/ml (Table VI) with acceptable rates greater than 96% and 88%, respectively. The laboratory established a LOD for phencyclidine of 2.5 ng/ml (Table VII) with acceptable rates greater than 86%. Table III. THC I II III IV V VI QC run # Acceptable* Chromatography Total Mean, ng/ml(1 SD*) 2.1 (0.12) 2.1 (0.11) 1.6 (0.07) 1.8 (0.10) %CV(n) 5.6 (13) 5.5 (14) 4.6 (13) 5.7 (23) Acceptable rate Mean, ng/ml SD 0.22 %CV 11.5 n Number of data points that were omitted from calculations because of unacceptable chromatography. r SD, standard deviation. Table IV. Benzoylecgonine I II QC run # Acceptable* 7 5 Chromatography t 1 2 Total 8 8 Mean, ng/ml (1 SD*) 7.0 (0.44) %CV (n) 6.2 (12) Acceptable rate 92.3 Mean, ng/ml SD 0.70 %CV 10.5 n (0.22) 3.2 (17) (0.07) 1.8 (0.15) 2.2 (12) 8.6 (16) III IV V (0.37) 6.1 (0.55) 5.9 (0.44) 5.1 (16) 8.9 (12) 7.4 (15) Number of data points that were omitted from calculations because of unacceptable chromatography. :~ SD, standard deviation. 14

4 Discussion Any urine drug-testing laboratory must deal with increasing documentation and certification requirements. This includes documentation of a variety of analytical performance criteria to include the following for each assay: precision around the cutoff, LOD, linearity for quantitative methods, and detection of carryover. High-volume production laboratories with multiple GC-MS instruments in operation require a practical approach to the determination of LOD. Determining an LOD for each individual method and instrument combination could be cost prohibitive or confusing for multiple combinations of method, instrument, and laboratory certifying official. Therefore, a single LOD was established for each drug method. Although this does not take each individual GC-MS to the absolute limit, it does satisfy the routine operating requirements of a urine drug-testing laboratory. The most common use of an LOD at the FTDTL is Table V. Amphetamines QC run # Acceptable* 8 4 Chromatography 0 0 Total 8 4 Mean, ng/ml (I SD)* (34.75) %CV(n) 25.7 (t4) Acceptable rate 100 Mean, ng/ml SD %CV 18.3 n 26 Amphetamine Methamphetamine I II I II * Number ofdata points acceptable by all review criteria. t Number of data points that were omitted from calculations because of unacceptable chromatography. SD, standard deviation. Table Vl. Opiates (2.77) (2.85) 97.1 (3.53) 1.9 (I 2) 2.8 (I 4) 3.6 (I 2) I00 I Drug Codeine Morphine I II I II QC run # Acceptable* 4 Chromatography* 0 Total 4 Mean, ng/ml (1 SD)* %CV(n) 4.2 Acceptable rate Mean, ng/ml SD %CV 11.2 n (23.45) (21.50) (30.86) (46.35) (20) 4.7 (25) 2.8 (18) 5.01 (23) i- Number of data points that were omitted from calculations because of unacceptable chromatography. :~ SD, standard deviation. 15

5 Table VII. Phencyclidine I II QC run # Date points Acceptable* 6 Chromatography t 0 Total 7 Mean, ng/ml (1 SD*) 2.6 %CV (n) 7.3 Acceptable rate 86.7 Mean, ng/ml SD 0.13 %CV 5.2 n (0.19) 2.6 (0.04) (13) 1.7 (14) 100.i- Number of data points that were omitted from calculations because of unacceptable chromatography. * SD, standard deviation. a specimen retest requested by the proper authority; the reported result is reconfirmed if the specimen retest result is greater than the LOD of the drug being analyzed. The development of a definition of LOD, which is the lowest concentration of drug that this laboratory can detect in a specimen with forensic certainty at a minimum of 85% of the time within all normal assay requirements except the requirement to meet quantitation criteria (within 20% of target value), was an important first step. The second step was to develop analytical criteria for the evaluation of the LOD data. Two analytical runs per certified instrument was the minimum number required for the assessment of run-to-run variability. A minimum of ten data points per instrument was required to prove compliance with the LOD definition. The acceptance of a minimum of 85% of data points was chosen to reflect the loss of one LOD data point per analytical run. with unacceptable chromatography were removed from the LOD acceptable rate calculation. When the first analytical batches were reviewed by the laboratory certifying officials, it was evident that different laboratory certifying officials interpreted data acceptability differently at the lower limits of the assays, especially chromatography review criteria. The chromatography review criteria were specific, but interpretation of the acceptance of a shoulder near 10% was dependent on the individual who reviewed the analytical batch. This opinion was often based on whether a shoulder was truly a shoulder or baseline noise near 10% of the LOD drug signal. Therefore, the laboratory chose not to limit interpretation and excluded any data points that failed because of chromatography. with failed mass ratios were always included because they were examples that the lower limit of the assay was being reached. The first runs of each drug were performed to narrow the LOB focus by analyzing serial dilutions of the low control. For example, when the sample size for THC was 10.0 ml, there was no problem meeting a 1.0-ng/mL LOD. However, when DoD reduced the specimen collection requirement from 60 to 30 ml, it was necessary to reevaluate the THC sample size. Currently, with a 5.0-mL sample size for THC, an LOD of I ng/ml with 85% acceptable rate was not attainable. The laboratory's LOD for THC was determined to be 2 ng/ml. The laboratory did not explore the possibility of an LOD between ] and 2 ng/ml because the additional experimental effort would not improve the laboratory's ability to successfully perform retests. The laboratory was able to certify six GC-MS using multiple personnel, including a new instrument eight months after the initial group of instruments was certified. The laboratory used a similar approach for the determination of benzoylecgonine and PCP LODs. The LOD focus for opiates (codeine and morphine) presented a different situation. The DoD cutoff for confirmation of opiates is 2000 ng/ml for codeine and 4000 ng/ml for morphine. Because of the high cutoff, the laboratory could only achieve LOD concentrations of 500 ng/ml for codeine and 1000 ng/ml for morphine. The laboratory's attempts to demonstrate acceptable performance below these concentrations failed (codeine, 400 ng/ml). Before the adoption of the current opiate cutoff, the laboratory used a cutoff of 300 ng/ml for codeine and 300 ng/ml for morphine with a sample size of 5.0 ml, which enabled a LOD of less than 75 ng/ml for each. The laboratory determined a LOD of 150 ng/ml for amphetamine and 100 ng/ml for methamphetamine and found that ion contribution to the qualifying ions from the internal standard caused failed mass ratios below these concentrations (both drugs at 50, 75, and 100 ng/ml). Conclusion The Forensic Toxicology Drug Testing Laboratory continues to use this practical approach for the determination of LODs. These LODs have been supportable in the normal course of business. Certification of new instruments has met the current LOD. Each instrument is recertified on a yearly basis and has continued to meet these LODs. References 1. Forensic urine drug testing section 50, Commission on Laboratory Accreditation Inspection Checklist, College of American Pathologists, (1995). 2. J. B0ttner, R. Borth, J.H. BoutweJl, and ffm.g. Broughton. ]nternational Federation of Clinical Chemistry provisional recommendation on quality control in clinical chemistry. Clin. Chem. 22" (1976). 3. Nomenclature, symbols, units and their usage in spectrochemical analysis - II. Spectrochim. Acta, Part B 33:242 ( 978). 4. G.L. Long and J.D. Winefordner. Limit of detection. Anal. Chem. 55" 712A-19A (1983). 5. S.B. Needleman and R.W. Romberg. Limits of linearity and detection for some drugs of abuse. J. Anal. Toxicol. 14:34-38 ( 990). 6. D.A. Armbruster, M.D. Tillman, and L.M. Hubbs. Limit of detection (LOD)/limit of quantitation (LOQ): Comparison of the empirical and the statistical methods exemplified with GC-MS assays of abused drugs. Ctin. Chem. 40(7): (1994). Manuscript received July 30,