Abstract
Background: An LC-MS/MS urine confirmation assay was developed using a “dilute and shoot” sample preparation method that was subject to interference arising from gabapentin column overload in approximately 4% of patient samples, leading to interference in amphetamine analysis.
Methods: The initial analysis method used dilute and shoot sample preparation followed by LC-MS/MS analysis. The improved assay used solid-phase extraction sample preparation followed by LC-MS/MS analysis.
Results: The improved assay using solid-phase extraction and alternative chromatographic conditions resolved the gabapentin interference in amphetamine analysis.
Conclusions: This experience illustrates the importance of thorough knowledge of likely comedications in a patient population and these drugs' elimination mechanisms. Laboratorians should be aware of the phenomenon of mass effect LC-MS/MS interference, in addition to the more common LC-MS/MS interferences caused by matrix effects and isobaric compounds.
Impact Statement
Patients taking gabapentin, a drug commonly used in pain management populations, and requiring urine amphetamine confirmation will benefit from the description of gabapentin interference in an LC-MS/MS amphetamine confirmation assay. In this article, the mechanism of the gabapentin interference and how to resolve it are described.
Urine drug testing is used for a wide variety of purposes, and testing volumes continue to rise for many laboratories. Pain management programs, drug and alcohol rehabilitation programs, labor/delivery departments, and employment drug testing are all examples of areas in which accurate results are critical to make appropriate decisions regarding healthcare and safety.
Urine drug testing typically starts with an immunoassay screen followed by confirmatory testing by GC-MS or LC-MS/MS techniques as needed. Immunoassay screens for amphetamines are known to be susceptible to false-positive results, often caused by common cold medications, antidepressants, and weight-loss drugs. Confirmatory tests need to be free of interferences that could cause false-positive or -negative results or that could delay reporting of results.
Gabapentin (Neurontin®, Pfizer) is a structural analog of the neurotransmitter γ-aminobutyric acid approved by the Food and Drug Administration for use as an adjunct treatment for patients with epileptic seizures and for those with postherpetic neuralgia. In addition to the Food and Drug Administration–approved indications, approximately 80% of gabapentin use is for off-label indications (1), including neuropathic pain and diabetic neuropathy, complex regional pain syndrome, anxiety, insomnia, and migraine (2). About 12% of patients receiving pain management treatment are prescribed gabapentin (3), and in our urine amphetamine confirmation population, gabapentin was present in approximately 8.5% of samples.
Gabapentin is also an effective treatment for mild alcohol withdrawal symptoms and sobriety maintenance (4, 5).
Gabapentin is dosed in healthy adults at 100 mg/day to 3600 mg/day and is excreted unchanged in urine (6). The large doses, coupled with urinary excretion of unchanged drug as the primary elimination route, account for the high levels of gabapentin that have been observed in patient urine, reported to be up to approximately 35 g/L (3).
In this report, we detail interference from gabapentin in an amphetamine urine confirmation assay causing suppression of analyte signal, poor peak shape, and retention time shift that was not mitigated by use of a deuterated internal standard. The interference was severe enough to require an alternative confirmation method for these samples and, ultimately, an improved sample preparation procedure and LC-MS/MS method to resolve the interference.
Materials and Methods
Chemicals
Amphetamine, amphetamine-d6, methamphetamine, methamphetamine-d5, MDMA, MDMA-d5, MDA, MDA-d5, benzoylecgonine, benzoylecgonine-d3, PCP, and PCP-d5 methanolic stocks were obtained from Cerilliant. Gabapentin powder standard, Chromasolv® HPLC water, 10 mol/L ammonium formate, methanol, potassium bicarbonate (KHCO3), and potassium carbonate (K2CO3) were purchased from Sigma Aldrich. Concentrated hydrochloric acid, ammonium hydroxide, and formic acid were purchased from Fisher Scientific. Drug-free human urine was purchased from UTAK.
Internal standards and calibrators
For the original assay, 5 levels of calibrators were prepared by diluting the 1-g/L methanolic stocks from Cerilliant into drug-free human urine. The concentration ranges for amphetamine, methamphetamine, MDMA, and benzoylecgonine were 50 ng/mL to 1000 ng/mL. The concentration range for MDA was 100 ng/mL to 1000 ng/mL and for PCP was 10 ng/mL to 100 ng/mL. The calibrators were aliquoted and stored at −70 °C with a tested stability of at least 3 months. Internal standard semistock solution containing 50 μg/mL amphetamine-d6, methamphetamine-d5, MDMA-d5, MDA-d5, benzoylecgonine-d3, and PCP-d5 was prepared in methanol and was stable at −20 °C for 1 month. The internal standard working solution was prepared fresh daily by diluting the semistock 20-fold with 50:50 methanol/water.
ELI urine controls obtained from ElSohly were used for amphetamine, methamphetamine, MDMA, and MDA. Custom urine quality control material obtained from UTAK was used for benzoylecgonine and PCP.
In the improved assay, 5 levels of calibrators were prepared from custom stock solutions made by UTAK in drug-free human urine. These were diluted into drug-free human urine to achieve the final calibrator concentrations. The concentration ranges for amphetamine, methamphetamine, MDMA, MDA, and benzoylecgonine were 50 ng/mL to 1000 ng/mL. The concentration range for PCP was 10 ng/mL to 100 ng/mL. The calibrators were aliquoted and stored at −20 °C, and measured stability was at least 74 days. Internal standard solutions were prepared in the same manner as for the original assay.
Quality control materials for the improved assay were the same ELI urine controls used for the amphetamine species, and the Bio-Rad Liquicheck Urine Toxicology Control Level 3 (Bio-Rad) was used for benzoylecgonine and PCP.
Apparatus
For both the original and the improved assays, chromatography was performed using a Shimadzu Prominence HPLC system (Shimadzu Scientific Instruments) coupled to a 3200 QTRAP® mass spectrometer (Sciex). The HPLC consisted of 2 Shimadzu LC-20AD pumps, 2 SIL-20AC autosamplers, and a Valco diverter valve (VICI). The system was duplexed to allow injection of samples onto 2 analytical columns with flow from the 2 systems alternately feeding into the mass spectrometer to increase sample throughput. The mass spectrometer was operated in positive electrospray ionization mode. Quantification and qualification ions were obtained by multiple reaction monitoring of the protonated precursor ion [M+H]+. The transitions monitored and the associated mass spectrometer source conditions are listed in Table 1 of the Data Supplement that accompanies the online version of this article at http://www.jalm.org/content/vol2/issue4. The same transitions were used for the original and the improved assays.
Sample preparation
In the original assay, sample preparation consisted of dilution and filtration before injection on the LC-MS/MS system. After appropriate dilution (determined from immunoassay screening values) with negative control urine (UTAK), 50 μL of the internal standard working solution was added to 250 μL of urine. Samples were mixed, then diluted with Chromasolv® HPLC-grade water (Sigma Aldrich) using a dilution factor of 10. Samples were mixed, and the total volume was transferred to a 0.2-μm protein precipitation filter plate (Sigma Aldrich) and centrifuged at 3500 rpm for 10 min. Following the filtration step, samples were injected on the LC-MS/MS system. A Hamilton STARlet liquid handling station was used for all liquid handling steps.
The improved assay used solid-phase extraction to help minimize interference from gabapentin. After appropriate dilution with negative control urine, 50 μL of the internal standard working solution was added to 250 μL of urine. Samples were mixed, and 500 μL of pH 9 bicarbonate buffer was added. After mixing, 750 μL was transferred to a 96-cartridge rack of CEREX polymeric strong cation exchange (PSCX) extraction columns (SPEware) and extracted using a CEREX ALDIII 96 automated liquid dispenser (SPEware). Samples were loaded onto the columns and then washed with 1 mL of pH 9 bicarbonate buffer, 1 mL of 0.1 mol/L HCl, and 1 mL of methanol. Samples were eluted into a clean 96-well deep-well collection plate with 1 mL of 2% ammonium hydroxide in methanol prepared fresh daily. After extraction, samples (200 μL) were diluted with 800 μL of 0.5% formic acid in water and were then injected on the LC-MS/MS system. A Hamilton STARlet liquid handling station was used for all liquid handling steps not performed by the CEREX ALDIII.
Chromatography
In the original assay, samples were injected on a Kinetex C18 3 × 100 mm, particle size 2.6 μm, analytical column with a Security Guard ULTRA C18 UHPLC guard column (Phenomenex). A gradient was used consisting of 2 mmol/L ammonium formate with 0.1% formic acid in water (mobile phase A) and methanol (mobile phase B). The gradient conditions are listed in Table 1. The mobile phase flow rate was 0.4 mL/min, and the injection volume was 10 μL. The analytical columns were heated to and maintained at 40 °C.
HPLC gradients for the original and improved LC-MS/MS assays.
The improved assay used a Kinetex F5 2.1 × 100 mm, particle size 2.6 μm, analytical column with a Security Guard ULTRA UHPLC F5 guard column (Phenomenex). A gradient was used consisting of 2 mmol/L ammonium formate with 0.1% formic acid in water (mobile phase A) and methanol (mobile phase B). The gradient conditions are listed in Table 1. The mobile phase flow rate was 0.4 mL/min, and the injection volume was 10 μL. The analytical columns were heated to and maintained at 40 °C.
Results
Interference testing during method validation of the original LC-MS/MS dilute and shoot assay included analyzing residual patient urine samples from pain management testing to look for any disturbances of internal standard signals indicative of interference (i.e., decreased peak areas and early or late retention times compared with calibrators, peak asymmetry). Substance Abuse and Mental Health Services Administration–recommended compounds, including ephedrine, pseudoephedrine, phenylephrine, phenylpropanolamine, phentermine, and pheniramine (spiked at concentrations of 100 μg/mL), in addition to N-desmethyl selegilene (spiked at a concentration of 100 μg/mL) and an over-the-counter mix from Cerilliant containing acetaminophen, caffeine, chlorpheniramine maleate, ibuprofen, and naproxen (spiked at concentrations of 10 μg/mL), were also tested. No interferences were observed.
On implementing the original LC-MS/MS assay for patient sample analysis, several samples displaying drastically reduced amphetamine and amphetamine-d6 areas (ranging from 50% to 10% of expected areas), along with retention time shifts to earlier elution times (approximately 0.2–0.4 min early) and asymmetric peak shapes, were observed. Only amphetamine and its internal standard (amphetamine-d6) were affected. These samples were analyzed on an LC-MS/MS comprehensive urine drug screen to try to identify a potential cause of this interference. All samples displaying this characteristic chromatographic appearance were positive for gabapentin. Examining patient charts revealed that all samples with the characteristic interference were from patients with gabapentin prescriptions. Patient charts for samples that did not have the characteristic interference (approximately 40) were checked for gabapentin prescriptions, and no gabapentin prescriptions were listed. For some samples, the interference could be resolved with a 2- or 5-fold dilution. However, this did not resolve the interference for all samples and was not feasible for samples that were true negatives for amphetamine or that fell below the lower limit of quantification when diluted. We resorted to analyzing samples that could not have the interference diluted out by a validated GC-MS method, which was not susceptible to the interference.
We characterized the gabapentin interference by adding a single transition for gabapentin to the LC-MS/MS assay, 172.0 > 154.0, and created a nonvalidated semiquantitative assay to estimate levels of gabapentin in our patient samples. Samples displaying the characteristic chromatographic interference were analyzed for the presence of gabapentin if patient charts indicated an active gabapentin prescription. Estimated gabapentin concentrations ranged from approximately 0.7 g/L to >40 g/L. Literature reports indicate urine gabapentin can reach levels as high as 35 g/L (3), which agreed with what was observed in our patients.
Our initial strategy to resolve the interference focused on finding extraction conditions to selectively retain the analytes of interest while washing away gabapentin. Using the PSCX columns, we achieved sufficient recovery for our analytes and reduced the interference from gabapentin, but not enough to resolve the problem. In the original dilute and shoot assay, we saw signal suppression and retention time shift at a gabapentin concentration of 0.2 μg/mL (data not shown). With the PSCX sample clean up, signal suppression and retention time shift still occurred at gabapentin concentrations >1 μg/mL (Fig. 1). Therefore, chromatographic conditions also required redevelopment to further resolve the interference. Mobile phase composition remained the same, and only minor adjustments to the gradient were made while evaluating several different columns (Phenomenex Kinetex F5, Phenomenex Evo C18, Phenomenex Biphenyl, Acentis C18, and Waters XTerra C18) for maximal separation of gabapentin from amphetamine. The only column able to baseline separate gabapentin from amphetamine was the Phenomenex Kinetex F5 (Fig. 2). The assay was validated using the PSCX sample preparation and the Kinetex F5 analytical column, and it was confirmed that gabapentin no longer interfered with the quantification of amphetamine. In the presence of 50-g/L gabapentin, the intraday accuracy and precision (n = 20) of the lower limit of quantification for all analytes met acceptance criteria of <20% imprecision (CV) and matching the nominal concentration within ±20%. At the cutoff values for each analyte (250 ng/mL for amphetamine, methamphetamine, MDMA, and MDA; 100 ng/mL for benzoylecgonine; and 25 ng/mL for PCP), intraday accuracy and precision (n = 20) were better than 15% CV and 10% CV, respectively (see Table 2 in the online Data Supplement). A series of 10 samples prepared from pooled patient samples known to contain gabapentin was analyzed by the improved LC-MS/MS method and the GC-MS/MS method, known to be free of the gabapentin interference. The results correlated well with a slope of 1.054, a bias of 5.1%, and a correlation coefficient of 0.9960 (Fig. 3).
The samples were analyzed on the Phenomenex Kinetex C18 analytical column.
We also investigated whether gabapentin affected our screening assay, the CEDIA™ Amphetamine/Ecstasy Drugs of Abuse immunoassay (Thermo Scientific). Negative control urine spiked to contain 25 g/L gabapentin was analyzed in triplicate according to the manufacturer's instructions. No false-positive signals were observed. Five urine samples spiked to contain 625 ng/mL amphetamine and 25 g/L gabapentin, along with 5 urine samples containing only 625 ng/mL amphetamine, were analyzed. The average amphetamine immunoassay concentrations in the presence vs absence of gabapentin were not significantly different (550 ng/mL with gabapentin vs 526 ng/mL without gabapentin).
Discussion
Discussions of interferences affecting LC-MS/MS assays generally center on matrix effects, isobaric compounds requiring HPLC separation, or structurally similar compounds with common fragmentation patterns. At low concentrations (<1 μg/mL with solid-phase extraction), gabapentin appears to cause matrix suppression in amphetamine signal, as evidenced by decreased peak areas with no retention time shift. This suppression was not completely compensated for by the internal standard, amphetamine-d6, because of imperfect coelution of amphetamine and amphetamine-d6. At high concentrations often seen in patient samples, gabapentin appears to cause a column overload chromatographic interference with amphetamine. Column overload occurs when there are more solute molecules than available binding sites in the column stationary phase, either because too much volume was injected on the column or because the solute concentration in the sample is too high (7). Gabapentin can elute at a similar retention time as amphetamine under certain chromatographic conditions, and the large amount of gabapentin in some samples may interfere with the ability of amphetamine to interact with the stationary phase of the chromatography column. This would explain the observed retention time shift and potentially account for a portion of the decreased amphetamine signal, as the amphetamine would not travel efficiently through the HPLC column as a discrete band and elute as a sharp peak, but rather would broaden and elute over a longer time frame. To illustrate, the Substance Abuse and Mental Health Services Administration amphetamine cutoff of 250 ng/mL yields 2.5 ng on a column when using a 10-μL injection volume. If the sample also contained 40 g/L gabapentin, there would be 400000 ng (0.4 mg) of gabapentin on the column. This is twice the maximum theoretical load of 0.2 mg for a 100 × 2.1-mm HPLC column, as was used in our assay (7). In addition to the chromatographic interference, a decrease in amphetamine signal could be partially caused by ion suppression in the mass spectrometer source if gabapentin and amphetamine coelute.
Gabapentin does not interfere with amphetamine in the GC-MS assay. Because gabapentin contains the primary amine necessary for the methylation derivatization reaction used in the GC-MS analysis, it is most likely that gabapentin is baseline resolved in the GC-MS assay.
Confirmation assays must be able to separate gabapentin from amphetamine, an analyte known to have a relatively high rate of false-positive findings in screening tests, to get accurate results. To our knowledge, this is the first published report detailing the potential of gabapentin to interfere with the LC-MS/MS confirmation of amphetamine. Because of the high concentrations of gabapentin present in urine, it could potentially interfere with LC-MS/MS assays for other urine drugs of abuse. Laboratorians need to be aware of this when they are developing methods so that gabapentin interference can be ruled out before implementation, avoiding the need for dilutions, alternative analyses, or having to report a test not done because of interference, which all delay turnaround time and are detrimental to patient care and satisfaction. Additionally, this case illustrates the importance of understanding how concomitant medications are dosed and how they are eliminated from the body. Drugs dosed at high levels and excreted primarily unchanged in urine will be potential causes of mass overload interferences in LC-MS/MS analyses.
Acknowledgments
The author would like to acknowledge Camilla Otten and Bridgette Henson, as well as David Hall from SPEware, for experimental help. The author would also like to acknowledge Dr. Julia Drees and Dr. Bridgit Crews for manuscript discussions and editing.
Footnotes
Authors' Disclosures or Potential Conflicts of Interest: The author declared no potential conflicts of interest.
Role of Sponsor: No sponsor was declared.
- Received May 22, 2017.
- Accepted July 18, 2017.
- © 2017 American Association for Clinical Chemistry
In this issue
