Doping Control Analysis of Intact Rapid-Acting Insulin Analogues in

Anal. Chem. 2006, 78, 1897-1903

Doping Control Analysis of Intact Rapid-Acting Insulin Analogues in Human Urine by Liquid Chromatography-Tandem Mass Spectrometry Mario Thevis,*,† Andreas Thomas,† Philippe Delahaut,‡ Alain Bosseloir,§ and Wilhelm Scha 1 nzer†

Institute of Biochemistry and Center for Preventive Doping Research, German Sport University Cologne, Carl-Diem Weg 6, 50933 Cologne, Germany, Laboratory of Hormonology, Centre d’Economie Rurale, Rue du Point du Jour 8, 6900 Marloie, Belgium, and ZenTech, Parc Scientifique du Sart-Tilman, Avenue du Pre´ -Aily 10, 4031 Angleur, Belgium

Insulin and related synthetic therapeutics have been prohibited by the World Anti-Doping Agency for athletes demonstrably not suffering from diabetes mellitus. The primary specimen for doping controls has been urine, but the renal excretion of intact human insulin as well as synthetic analogues such as the rapid-acting products Humalog LisPro, Novolog Aspart, and Apidra Glulisine has been reported negligible owing to metabolic degradation. Nevertheless, employing solid-phase extraction in combination with immunoaffinity purification followed by a top-down sequencing-based mass spectrometric approach, an assay was established allowing the identification of three intact rapid-acting synthetic insulins in doping control urine samples. A volume of 25 mL of urine was concentrated, insulin analogues were isolated from the concentrate by immunoaffinity chromatography, and the eluate was analyzed using microbore liquid chromatography/tandem mass spectrometry. Characteristic product ion spectra obtained from 5-fold protonated intact analytes as well as isolated insulin B-chains allowed the unambiguous identification of target analytes with detection limits of 0.05 ng/mL (9 fmol/mL). Moreover, assay validation demonstrated recoveries between 72 and 80% for Humalog LisPro, Novolog Aspart, and Apidra Glulisine, and assay precisions ranged from 9 to 16%. A reliable tool is provided that allows the qualitative determination of rapid-acting insulins in urine specimens collected for sports drug testing. Since 1999, the use of insulins has been prohibited by the International Olympic Committee (IOC)1 and the World AntiDoping Agency (WADA)2 for professional and amateur athletes that are demonstrably not suffering from diabetes mellitus. The misuse of insulin in sport and its hazardous consequences have * Corresponding author. Tel.: 0221-4982-7070. Fax: 0221-497-3236. E-mail: [email protected]. † German Sport University Cologne. ‡ Centre d’Economie Rurale. § Parc Scientifique du Sart-Tilman. (1) List of prohibited classes of substances and prohibited methods of doping; International Olympic Committee: Lausanne, 2003. (2) World Anti-Doping Agency. The 2005 Prohibited List, http://www. wada-ama.org/rtecontent/document/list_2005.pdf; 05-09-2005. 10.1021/ac052095z CCC: $33.50 Published on Web 02/16/2006

© 2006 American Chemical Society

been reported frequently during the last years,3-6 although the question whether insulins have performance-enhancing properties has been discussed controversially.7-10 Insulin is a peptide hormone consisting of two peptide chains that are cross-linked by two disulfide bonds (Figure 1a) with a molecular mass of 5807 Da. Due to the protein’s strong affinity to self-association to noncovalent hexamers11 and a resulting lag phase between subcutaneous injection and bioavailability of insulin monomers,12 rapid-acting insulins such as Humalog LisPro, Novolog Aspart, and Apidra Glulisine (Figure 1b-d) have been introduced, the bioavailability of which is accomplished within 10-15 min after administration.13-18 These insulin derivatives possess significantly reduced tendencies toward the formation of hexamers because of modified primary structures (Figure 1). The correspondingly increased controllability of the insulin action has raised concerns about their preference for being misused in sports,19 and a test method for synthetic insulins in human plasma based on immunoaffinity chromatography (IAC) followed by liquid chromatography/tandem mass spectrometry (LC-MS/MS) has recently been published20 for doping control purposes. However, the most commonly collected doping control specimens are urine samples, which have, so far, been considered less suitable for the (3) Reverter, J. L.; Tural, C.; Rosell, A.; Dominguez, M.; Sanmarti, A. Arch. Intern. Med. 1994, 154, 225-226. (4) Rich, J. D.; Dickinson, B. P.; Merriman, N. A.; Thule, P. M. JAMA, J. Am. Med. Assoc. 1998, 279, 1613. (5) Evans, P. J.; Lynch, R. M. Br. J. Sports Med. 2003, 37, 356-357. (6) Dawson, R. T.; Harrison, M. W. Br. J. Sports Med. 1997, 31, 259. (7) Sonksen, P. H. J. Endocrinology 2001, 170, 13-25. (8) Wolfe, R. R. Curr. Opin. Clin. Nutr. Metab. Care 2005, 8, 61-65. (9) Wolfe, R. R. Curr. Opin. Clin. Nutr. Metab. Care 2000, 3, 67-71. (10) Tipton, K. D.; Wolfe, R. R. Int. J. Sport Nutr. Exercise Metab. 2001, 11, 109-132. (11) Fabris, D.; Fenselau, C. Anal. Chem. 1999, 71, 384-387. (12) Barnett, A. H.; Owens, D. R. Lancet 1997, 349, 47-51. (13) Rosak, C. Internist 2001, 42, 1523-1535. (14) Lindstro¨m, T.; Hedman, C. A.; Arnqvist, H. J. Diabetes Care 2002, 25, 10491054. (15) Plum, A.; Agers, H.; Andersen, L. Drug Metab. Dispos. 2000, 28, 155-160. (16) Becker, R. H.; Frick, A. D.; Burger, F.; Potgieter, J. H.; Scholtz, H. Exp. Clin. Endocrinol. Diabetes 2005, 113, 435-443. (17) Danne, T.; Becker, R. H.; Heise, T.; Bittner, C.; Frick, A. D.; Rave, K. Diabetes Care 2005, 28, 2100-2105. (18) Becker, R. H.; Frick, A. D.; Burger, F.; Scholtz, H.; Potgieter, J. H. Exp. Clin. Endocrinol. Diabetes 2005, 113, 292-297. (19) Steiner, A.; Wagner, R. A. ISP-Verlag, Arnsberg 2002, 3, 4-73. (20) Thevis, M.; Thomas, A.; Delahaut, P.; Bosseloir, A.; Schanzer, W. Anal. Chem. 2005, 77, 3579-3585.

Analytical Chemistry, Vol. 78, No. 6, March 15, 2006 1897

Figure 1. Primary structures of investigated insulins: (a) human insulin (mol wt ) 5807), (b) Humalog (mol wt ) 5807), (c) Novolog (mol wt 5826), and Apidra (mol wt ) 5823). Modifications of amino acid sequences compared to human insulin are indicated in gray.

determination of insulin owing to its metabolic fate. But, as early as 1948, insulin was quantitatively measured in urine specimens21 followed by numerous additional studies employing immunological assays that have proven the presence of low amounts of intact insulin in human urine22-28 of ∼50 fmol/mL.24 Still, until today no attempt has been made to determine intact synthetic insulins in (21) Mirsky, I. A. J. Clin. Invest. 1948, 27, 515-519. (22) Jorgensen, K. R. Acta Endocrinol. (Copenhagen) 1966, 51, 400-410. (23) Crossley, J. R. J. Lab. Clin. Med. 1974, 84, 752-758. (24) Aun, F.; Soeldner, J. S.; Meguid, M. M.; Stolf, N. A. G. Postgrad. Med. J. 1975, 51, 622-626. (25) Chamberlain, M. J.; Stimmler, L. J. Clin. Invest. 1967, 46, 911-919. (26) Polonsky, K. S.; O’Meara, N. M. In Endocrinology, 4th ed.; DeGroot, L. J., Jameson, J. L., Eds.; Elsevier: Philadelphia, 2001; pp 697-727. (27) De Palo, E. F.; Gatti, R.; Lancerin, F.; De Palo, C. B.; Cappellin, E.; Solda, G.; Spinella, P. Clin. Chem. Lab. Med. 2003, 41, 1308-1313. (28) Duckworth, W. C.; Bennett, R. G.; Hamel, F. G. Endocr. Rev. 1998, 19, 608-624.

1898

Analytical Chemistry, Vol. 78, No. 6, March 15, 2006

human urine by mass spectrometry, and in the present study, we demonstrate the possibility of reproducibly and reliably identifying three intact rapid-acting insulins in human urine using top-down sequencing-based LC-MS/MS strategies for doping control purposes. EXPERIMENTAL SECTION Materials and Chemicals. Oasis HLB solid-phase extraction (SPE) cartridges (60 mg) were obtained from Waters (Milford, MA). Acetonitrile (HPLC grade), trisodium phosphate dodecahydrate (p.a.), sodium chloride (p.a.), and acetic acid (glacial) were purchased from Merck (Darmstadt, Germany), and trifluoroacetic acid (99+%), tris(carboxyethyl)phosphine hydrochloride (TCEPHCl), and bovine insulin were from Sigma (St. Louis, MO). Insulin analogues Humalog, Novolog, and Apidra were supplied by Eli Lilly (Indianapolis, IN), Novo Nordisk (Princeton, NJ), and Aventis

Pharma (Frankfurt a. M., Germany), respectively, and the antiinsulin immunoaffinity gel (0.5 mL/column, 10 mg of IgG/mL) was obtained from CER (Marloie, Belgium). All buffers and solutions were prepared using deionized water (MilliQ grade). Urine Samples. All experiments and validation steps were performed using fortified blank urine samples obtained from healthy volunteers (6 male and 4 female donors). To probe for specificity and applicability of the assay, four urine samples of patients and elite athletes suffering from diabetes mellitus and declaring the treatment with rapid-acting insulin analogues were prepared and analyzed. All specimens were stored at +4 °C until analysis. Stock and Working Solutions. All solutions were prepared in polyethylene tubes to avoid loss of target analytes on glass surfaces. Stock solutions were stored at 2-8 °C and used for no longer than one week. A solution containing 10 pmol/µL bovine insulin in 2% acetic acid was used as internal standard (ISTD) stock solution. Humalog, Novolog, and Apidra stock solutions contained 100 pmol/µL in 2% acetic acid and were freshly diluted before use to a final concentration of 0.01 pmol/µL in 2% acetic acid. These working standard solutions also contained a 10-fold excess of bovine insulin (0.1 pmol/µL), which was added to the Eppendorf tube prior to the target analytes in order to saturate active surfaces. Reduction of Disulfide Bonds. To obtain detailed complementary information about the fragmentation behavior of target analytes, disulfide bonds of insulins were reduced using 900 µL of a 20 pmol/µL insulin solution in 2% acetic acid and 100 µL of a 100 mM TCEP-HCl solution by incubation at 60 °C for 10 min. After dilution with acetonitrile (ACN; 1:1, v/v), B-chains were analyzed by ESI-MS/MS. Mass Spectrometry. All reference mass spectra of intact or disulfide-reduced compounds were measured on an Applied Biosystems Qtrap mass spectrometer (Forster City, CA). The analytes were introduced by means of a syringe pump at a flow rate of 5 µL/min using solutions containing 1 pmol/µL. The ion spray voltage was set to 5500 V, and collision-induced dissociation (CID) was performed at a collision gas pressure of 5.33 × 10-3 Pa and with collision offset voltages of 50 or 70 V for B-chains or intact insulins, respectively. Sample Preparation. The sample preparation was modified from a previously published assay.20 Briefly, 25 mL of urine was fortified with 50 µL of internal standard stock solution, acidified with 200 µL of glacial acid, and vortexed properly. The sample was transferred in 2-mL aliquots onto an Oasis SPE cartridge, preconditioned with 2 mL of ACN and 2 mL of acetic acid (2%). The cartridge was washed with 2 mL of acetic acid (2%) and eluted with 1.8 mL of a mixture of acetic acid (2%) and ACN (1:1, v/v) directly onto the IAC column, which contained 3 mL of phosphatebuffered saline (PBS, consisting of 120 mM Na3PO4 and 0.5 M NaCl in deionized water; pH was adjusted to 8 by 3 M aqueous hydrochloric acid). After loading the SPE eluate, another 3 mL of PBS was added, and the mixture was vortexed for 5 s and incubated for 30 min at room temperature with two vortex repetitions after 10 and 20 min. The sample solution was eluted from the IAC column while the target analytes were retained by the antigen-antibody interaction. The column was rinsed 3 times with 3 mL of PBS, and insulins were eluted onto another

preconditioned Oasis SPE cartridge using 3 × 1 mL of acetic acid (2%). The cartridge was washed with 2 mL of acetic acid (2%), and elution of target analytes into a 1.5-mL Eppendorf tube was accomplished using 1.2 mL of a mixture of acetic acid (2%) and acetonitrile (1:1, v/v). The sample was evaporated to dryness utilizing a vacuum centrifuge at 40 °C for ∼90 min. The residue was reconstituted in 40 µL of a mixture of acetic acid (0.5%) containing 0.01% TFA and acetonitrile (72:28, v/v). An aliquot of 20 µL of the reconstituted sample was used for subsequent LCMS/MS analysis. To measure corresponding insulin B-chains only, a volume of 2 µL of 100 mM aqueous TCEP-HCl solution was added to the residual sample of ∼20 µL, which was incubated at 50 °C for 10 min prior to injection into the LC-MS/MS system. Liquid Chromatography-Tandem Mass Spectrometry. LC-MS/MS was performed on an Agilent 1100 Series highperformance liquid chromatograph (Palo Alto, CA) coupled to an Applied Biosystem Qtrap mass spectrometer (Forster City, CA). The LC was equipped with a Zorbax 300SB-C8 guard column (1 mm × 17 mm, 5-µm particle size) and a Zorbax 300SB-C18 analytical column (1 × 50 mm, 5-µm particle size, 300-Å pore size) using a column oven temperature of 40 °C. The mobile phases consisted of 0.5% acetic acid with 0.01% TFA (A) and a mixture of 0.5% acetic acid with 0.01% TFA and acetonitrile (1:4, v/v) (B). A linear gradient was used starting at 72% A and ending at 35% A after 15 min with a flow rate of 70 µL/min. Subsequently, a 25min equilibration period was added. The mass spectrometer was operated in positive ion spray mode with a needle voltage of 5500 V. Parameters such as declustering potential, ion trap fill time, and entry barrier were optimized for isolation and detection of the 5-fold protonated molecules of human insulin, Humalog, Novolog, and Apidra. Product ion spectra were measured at collision offset voltages of 70 V, utilizing nitrogen as collision gas (5.33 × 10-3 Pa). Owing to similar precursor ions for Novolog (m/z 1166.2) and Apidra (m/z 1165.5), detection was accomplished using one product ion scan experiment at m/z 1165.5 with a Q1 resolution of (0.8 u. Assay Validation. Specificity was demonstrated by preparing and analyzing 10 different urine samples as described above. No interfering signals in product ion scan chromatograms for Novolog and Apidra at expected retention times should be generated. The differentiation between Humalog and human insulin (both having identical molecular weights; see Results and Discussion) was based on the characteristic product ion at m/z 217 of Humalog (Figure 2b), which is not detectable in the case of human insulin (Figure 2a). The lower limit of detection (LLOD) was defined as the “lowest content that can be measured with reasonable statistical certainty”29 at a signal-to-noise ratio of g3. It was determined by measuring 6 urine samples, which were fortified with 9 fmol/mL of each insulin analogue, and 10 blank urine specimens analyzed for specificity. The assay precision at the LLOD was determined for each analyte at a concentration level of 9 fmol/mL by calculating standard deviations from six replicates.30 (29) Kromidas, A. Validierung in der Analytik; Wiley-VCH: Weinheim, 1999. (30) International Conference on Harmonisation. Validation of Analytical Procedures: Methodology; http://www.ich.org/MediaServer.jser?@ _ID)418&@_MODE)GLB.; 22-12-04.

Analytical Chemistry, Vol. 78, No. 6, March 15, 2006

1899

Figure 2. Product ion mass spectra of (a) human insulin obtained by CID of (M + 5H)5+ at m/z 1162.3, (b) Humalog obtained by CID of (M + 5H)5+ at m/z 1162.3, (c) Novolog obtained by CID of (M + 5H)5+ at m/z 1166.3, and (d) Apidra obtained by CID of (M + 5H)5+ at m/z 1165.5. All spectra were recorded on an Applied Biosystems Qtrap instrument using a collision offset voltage of 70 V.

The recovery of synthetic insulins was determined at a concentration of 18 fmol/mL. Six blank specimens were fortified with target analytes before sample preparation, and another six blank samples were prepared according to described protocol followed by addition of insulins prior to the evaporation step using the vacuum centrifuge. Recovery was calculated by comparison of mean peak area ratios of analytes and ISTD of samples fortified prior to and after the consecutive SPE/IAC/SPE procedure. To probe for linearity of relative signal intensities of rapidacting insulins in a selected concentration range, six urine samples were fortified with 9.0, 13.5, 18.0, 22.5, 27.0, and 35.0 fmol/mL of each reference compound and measured once. RESULTS AND DISCUSSION The scope of the study was the development of an assay allowing the mass spectrometric determination of rapid-acting synthetic insulin analogues in human urine doping control samples. Normal urinary insulin levels for fasting or nonfasting nonobese subjects have been reported between 20 and 550 fmol/ mL24,31 with an average of 50 fmol/mL considering an excretion of 2 L of urine/24 h. In addition, pre-exercise urinary insulin levels of well-trained cyclists were reported at ∼100 fmol/mL.27 However, the amount of synthetic insulins potentially administered surreptitiously by cheating athletes in a sport is difficult to estimate. Assuming their intention to accomplish local hyperinsulinemia that is supposed to have anticatabolic effects,8,32,33 urinary levels are (31) Najjar, S. S.; Stephan, L. Metabolism 1970, 19, 301-308.

1900

Analytical Chemistry, Vol. 78, No. 6, March 15, 2006

expected at least as high as found under normal nonfasting conditions. Hence, the presented assay was designed to allow for the determination of urinary insulin levels at ∼20 fmol/mL. Liquid Chromatography-Tandem Mass Spectrometry. The identification of synthetic insulins was based on their affinity to immobilized anti-insulin antibodies, chromatographic retention times, and mass spectrometric characteristics in LC-MS/MS analyses with or without cleavage of analytes by disulfide reduction. The product ion mass spectra of human (endogenous) insulin, Humalog, Novolog, and Apidra obtained by CID of 5-fold protonated precursor ions are shown in Figure 2. Diagnostic product ions for human insulin and Humalog, the amino acid sequences of which differ only by switched residues at positions B28 and B29, are found at m/z 226 (y3 - y1) and 217 (y2), respectively20 (Figure 2a,b). The proline-directed dissociation of the charged analytes34 allowed a mass spectrometric distinction despite identical precursor ions and closely related chromatographic properties. The differentiation of Novolog and Apidra from human insulin was accomplished by considerably different masses and product ion spectra due to the exchange of the proline residue (B28) of human insulin by aspartic acid (Novolog) or the substitution of the asparagine (B3) and lysine (B29) residues by lysine and glutamic acid (Apidra), respectively (Figure 2c,d). In the case (32) Tessari, P.; Inchiostro, S.; Biolo, G.; Vincenti, E.; Sabadin, L. J. Clin. Invest. 1991, 88, 27-33. (33) Biolo, G.; Declan Fleming, R. Y.; Wolfe, R. R. J. Clin. Invest. 1995, 95, 811819. (34) Loo, J. A.; Edmonds, C. G.; Smith, R. D. Anal. Chem. 1993, 65, 425-438.

Figure 3. Product ion mass spectra of B-chains of (a) Novolog obtained by CID of (M + 4H)4+ at m/z 863.1 and (b) Apidra obtained by CID of (M + 4H)4+ at m/z 862.2. Both spectra were recorded at collision offset voltages of 50 V.

of Novolog, the proline-directed dissociation of the multiply charged precursor ion to m/z 226 is missing due to the removal of the proline residue (B28). In contrast, the characteristic dissociation pathway of human insulin to abundant y3, y3 - H2O, and y3 - y1 was found also with Apidra yielding product ions at m/z 346, 328, and 227, respectively (Figure 2d). Additionally, an intense and diagnostic fragment ion at m/z 199 was observed that was identified in MS3 experiments as a product ion of m/z 227 eliminating carbon monoxide (-28 u). To obtain additional mass spectrometric evidence for the composition of target analytes, synthetic insulins were treated with TCEP-HCl providing separated A- and B-chains. All modifications of rapid-acting therapeutics compared to human insulin are found in the B-chain, and thus, their product ion mass spectra were of particular interest to allow an identification of prohibited compounds. Resulting spectra obtained from Novolog and Apidra reference compounds are presented in Figure 3, while corresponding spectra of human insulin and Humalog were described elsewhere.20 The quadruply charged precursor ion of the B-chain of Novolog was observed at m/z 863.1, and CID provided characteristic fragment ions that were primarily formed by a series of doubly charged b-ions. The obtained sequence information tag allowed an unambiguous assignment of the measured peptide as the B-chain of Novolog (Figure 3a). The CID spectrum of Apidra (Figure 3b) was also obtained from the quadruply charged precursor ion at m/z 862.2, yielding informative sequence information tags based on doubly charged b-ions from b102+ - b162+.

Figure 4. Total ion chromatograms of product ion scan experiments obtained from urine samples fortified with 18 fmol/mL of (a) Humalog, (b) Novolog, and (c) Apidra. Distinct signals representing the respective target analytes are observed at retention times between 20.43 and 20.51 min.

Additionally, three y-ions were obtained with intense abundance at m/z 346 (y3), 227 (y3 - y1), and 199 corresponding to the product ion mass spectrum obtained from the intact structure of Apidra (Figure 2d). In the course of method development, urine samples were fortified with the rapid-acting synthetic insulins at various concentration levels. Typical analysis results of urine specimens enriched with 18 fmol/mL (0.1 ng/mL) of either Humalog, Novolog, or Apidra are depicted in Figure 4. Total ion chromatograms obtained from product ion scans of respective precursor ions allow the identification of intact synthetic insulins and Analytical Chemistry, Vol. 78, No. 6, March 15, 2006

1901

Figure 5. Total ion chromatograms of product ion scan experiments for Apidra using (a) a blank urine specimen that does not generate any considerable signal at a retention time between 17 and 24 min measuring the intact analyte and (b) (top) a doping control urine specimen of a male elite athlete using Apidra to treat diabetes mellitus yielding an abundant signal at 20.39 min. The corresponding product ion spectrum (bottom) contains characteristic product ions of Apidra at m/z 346 (y3), 227 (y3 - y1), and 199. Chromatogram c (top) was obtained from the same doping control sample after treatment of the insulin-containing extract with TCEP-HCl. The resulting B-chain was analyzed using CID of its quadruply charged precursor ion at m/z 862.2, providing a product ion mass spectrum with comprehensive information on the amino acid sequence of the target analyte (bottom).

demonstrate the feasibility to isolate the target compounds from urine and determine intact insulins by top-down mass spectrometry. Blank urine samples that were not spiked with synthetic insulins did not show interfering signals at expected retention times (vide infra). In addition to fortified urine specimens, doping control samples with an indicated use of rapid-acting insulins due to diabetes mellitus disease and urine specimens obtained from patients also using rapid-acting insulin analogues were analyzed providing the proof of principle. In Figure 5a, the results of a blank urine sample is presented in contrast to a specimen obtained from a male elite athlete using the rapid-acting insulin derivative Apidra (Figure 5b). The product ion spectrum obtained from the intact analyte is shown in Figure 5b below the total ion chromatogram of the 1902 Analytical Chemistry, Vol. 78, No. 6, March 15, 2006

product ion scan experiment containing the characteristic fragment ions of Apidra at m/z 346, 227, and 199. The treatment of the same urine extract with TCEP-HCl yielded chemically separated A- and B-chains, and the analysis of the latter by LC-MS/ MS provided complementary and substantiating data unambiguously proving the identity of the synthetic insulin derivative with comprehensive amino acid sequence tags (Figure 5c). Assay Validation. To test the applicability of the established procedure to doping controls, selected parameters such as specificity, LLOD, recovery, precision, and linearity were validated for Humalog, Novolog, and Apidra using spiked urine specimens. In addition, urine samples of athletes suffering from diabetes mellitus using rapid-acting synthetic insulins were analyzed to prove the reliability of the assay with authentic urine samples

Table 1. Summarized Validation Results

compd

mol wt

recovery (%)

LLOD (fmol/mL)

precision at LLOD (n ) 6) (%)

r

Humalog Novolog Apidra

5807 5826 5823

78 72 80

9 9 9

9 11 16

0,992 0,980 0,992

containing excreted synthetic insulins. The tested parameters are summarized in Table 1. The determined recoveries for Humalog, Novolog, and Apidra ranged from 72 to 80% at precisions between 9 and 16%. The efficient extraction and purification of trace amounts of insulins allowed an LLOD of 9 fmol/mL (0.05 ng/mL) for all three synthetic analogues. The noise in blank urine samples was measured, and a 3-fold standard deviation was added. The signals obtained for Humalog, Novolog and Apidra provided a signal-tonoise ratio of g3 at 9 fmol/mL. According to normal nonfasting insulin urine levels, calibration curves were prepared for synthetic insulins from 9 to 35 fmol/mL, all of which allowed approximation of linearity according to the Mandel test.35 The obtained equations were y ) 1.5417x - 0.002 for Humalog at r ) 0.992, y ) 1.7224x + 0.004 for Novolog at r ) 0.980, and y ) 1.5231x - 0.035 for Apidra at r ) 0.992. Owing to product ion scan experiments utilized for all analyses in the present study, peak areas included the total ion current at respective retention times of target analytes. The assay was tested for specificity with 10 different urine samples prepared and analyzed as described above. No sample generated interfering signals at expected retention times of target analytes, and signals corresponding to human insulin were distinguished from Humalog by the absence of the diagnostic fragment ion at m/z 217. (35) Mandel, J. The Statistical Analyses of Experimental Data; John Wiley & Sons: New York, 1964.

calibration curve (9-35 fmol/mL) equation approximation y ) 1.542x - 0.002 y ) 1.722x + 0.004 y ) 1.523x - 0.035

linear linear linear

CONCLUSION The identification of intact synthetic insulins in human urine was accomplished by allowing the determination of the rapidacting therapeutics Humalog, Novolog, and Apidra in doping control urine samples at low femtomole per milliliter levels. Informative product ion mass spectra, obtained by a top-down sequencing-based assay, enabled the unambiguous identification of the prohibited drugs and will allow sports drug testing laboratories to uncover their misuse. Earlier studies regarding a determination of insulin misuse in sports were focused on blood samples, which have been rarely available in doping controls. Owing to the noninvasive nature of urine analysis and the much higher frequency of drug tests based on urine sampling, the efficiency and comprehensiveness of the antidoping fight will be improved by an incorporation of the presented strategy. ACKNOWLEDGMENT This project has been carried out with support from the World Anti-Doping Agency (WADA). The authors are grateful to Dr. J. Thomas of the Hospital Siegburg for providing excretion study urine samples. Received for review November 29, 2005. Accepted January 24, 2006. AC052095Z

Analytical Chemistry, Vol. 78, No. 6, March 15, 2006

1903