High‐Performance Liquid Chromatographic Determination of 1,1′‐Ethylidenebis(L‐tryptophan) in L‐Tryptophan Preparations

High-Performance Liquid Chromatographic Determination of 1,1'-Ethylidenebb@-tryptophan) in L-Tryptophan Preparations MARY W. TRUCKSESS~, FREDERICK S. THOMAS,AND SAMUEL W. PAGE Received May 21, 1993, from the Food and Drug Administration, Center for Food Safety and Applied Nutrition. 200 C Street SW, Washington, DC 20204. Accepted for publication November 24. 1993@. Abstract 0 In studies to determine the cause or causes of the

eosinophilia myalgic syndrome (EMS) and to monitor the purity of L-tryptophan preparations, an HPLC method has been developed for determining 1,1'-ethylidenebis(L4ryptophan)(EBT) in L-tryptophan (W) preparations. The W preparations are extracted with 0.1 % trifluoroacetic acid (TFA) and filtered, and the EBT is purified by passage through a Sep-Pak Cle cartridge. The cartridge is washed with water and 6% acetonitrile in water, and EBT is eluted with methanol. The waterdiluted eluate is then chromatographed on a silica-based, reversed-phase HPLC column with a gradient of water and 8 0 % acetonitrile, both solvents containing 0.1 % TFA. EBT absorbance is measured at 280 nm. The average recovery of EBT from L-tryptophan powder, spiked over the range 1.2-4.8 pglg, was 9 1 %. The limit of determination was -0.6 pglg. Sixteen test samples of W products manufactured by the company to which most of the cases of EMS have been traced contained >70 pg of EBTIg. Three nonpatientrelated test samples either did not contain EBT or contained <2 pg of EBTIg.

W should be monitored for this impurity. A quantitative method for EBT has now been developed, with an average recovery of 93% at 1.2 pglg. Approximately 8 h is required for the analysis of eight test samples.

Experimental Section EBT Preparation-This

material was prepared by the method of Smith et al.9 and is available from FDA, Center for Food Safety and Applied Nutrition, Division of Natural Products (HFS-345),200 C Street, SW, Washington, DC 20204 (202-2054355). The purity was determined to be >95% by W absorption and by HPLC. The major impurity is W. EBT Standard SolutionPreparation-Stock solution was prepared by dissolving 1 2 mg of EBT in 100 mL of 0.1 M triethylammonium acetate (TEAA) buffer (pH 9.2) and was stored at 5 "C. An EBT working standard solution of 1.2 ng/pL was prepared daily by adding 20 pL of the stock solution to a 2-mL volumetric flask and diluting to volume with water. Extraction--One W tablet containing >0.5 gitablet or two tablets each containing K0.5 g/tablet were ground. Only the powder of the W capsule was used for analysis. One 0.5-g test portion of the powder was extracted with 100 mL of a 0.1% solution of trifluoroacetic acid (TFA, Catalog no. T6608, Sigma Chemical Company, St. Louis, MO) in water In the fall of 1989, a peculiar illness marked by eosinophilia for 5 min by shaking or stirring. The extract was filtered, and 50 mL and severe myalgia was recognized by several physicians in New of the filtrate was collected in a graduated cylinder. Mexico. They, in collaboration with researchers at the Mayo Cleanup-A cartridge reservoir (Catalog no. 210315, Alltech AssoClinic, National Institutes of Health, National Institute of Mental ciates, Deerfield, IL) was placed on top of a Sep-Pak Clscartridge (Catalog Health, and the Centers for Disease Control (CDC),1-3associated no. 51910, Millipore-Waters, Milford, MA). The cartridge was coupled this illness with the consumption of an L-tryptophan (W) to a vacuum apparatus (Catalog no. 210351, Alltech Associates). A 250preparation. This illness was named the eosinophilia-myalgia mL beaker was placed in the vacuum apparatus under the outlet of the syndrome (EMS). Prompt action by the Food and Drug cartridge. The cartridge was conditioned by washing with 6 mL of Administration (FDA) in removing W from the market rapidly methanol (Catalog no. 230-4, Baxter Diagnostics, Inc., Muskegon, MI) and then 6 mL of water (MilliQ quality; flow rate, 3 mL/min). When ended this epidemic. However, over 1600 cases of EMS with 38 the solvent reached the bottom of the reservoir, the filtrate was added fatalities were attributed to exposure to W. This W was sold to the cartridge (flow rate, 6 mL/min). When the cartridge reservoir primarily as a food supplement, although it was also used to was empty, the graduated cylinder was rinsed with 30 mL of water, and treat insomnia, premenstrual syndrome, obesity, and drug the rinse was added to the reservoir. The cartridge was washed with 6 withdrawal. There had been no previous reports of adverse mL of acetonitri1e:water (6:94; acetonitrile, HPLC and UV grade; Catalog effects. no. 015-4, Baxter Diagnostics, Inc.) and air-dried under reduced pressure Studies by the CDC and other researchers in Oregon and in for 1min. The vacuum was turned off, and a 5-mL graduated cylinder Minnesota found a strong relationship of EMS with W produced was placed under the cartridge. EBT was eluted with 1.5mL of methanol with minimum vacuum so that the cartridge was saturated with methanol by Showa Denko, K.K., during the period from October 1988 to (shown by methanol droplets a t the end of the cartridge). Immediately June 1989.3.4 In early 1990, we developed an HPLC procedure the vacuum was released. When the methanol reached the top of the to screen W products for impurities and provided the procedure cartridge packing, 5 mL of water was added to the reservoir, and the to a number of research groups working on the EMS p r ~ b l e m . ~ vacuum was turned on slowly. When the solvent level reached 3.8 mL, The chromatograms of W products varied, depending on the the vacuum was released and 4.0 mL of eluate was collected. The manufacturing process. Chromatograms were similar for prodgraduated cylinder was removed and the eluate was poured into a 5-mL ucts from the same manufacturer, except for the case in which UniPrep syringeless filter housing (Catalog no. UN113UNYL,Whatman two different manufacturing processes had been used. This Inc., Clifton, NJ). The eluate was poured back and forth between the finding was used to determine the source of an unknown lot. cylinder and the housing several times to ensure proper mixing. The filter was pushed through the housing containing the eluate. The Mayo Clinic, CDC, and the Japanese National Institute Instrumental Conditions-HPLC was used to separate and quanof Hygienic Sciences correlated the impurities found by HPLC titate the EBT in the filtered eluate. The HPLC system consisted of with incidence of EMS3357 and concluded that certain impurities a U6K injector, two model 510 pumps, a model 490 W detector set a t were most strongly associated with the illness. One of these 280 nm, and an 840 data system (Waters, Milford, MA). The HPLC impurities was subsequently identified as 1,l'-ethylenebis(Lconditions are shown in Table 1. A standard curve was prepared daily tryptophan) (EBT).8,9 Recent testing in an animal model has by injecting 50,100,200, and 400 pL of working standard solution onto shown that EBT can produce some,but not all, of the pathological the column. The correlation coefficient of absorbance versus concentration was usually >0.99. A portion of the eluate (50-500pL for unknown abnormalities observed in EMS patients.1° Although the role test solution; 500 pL for recovery study) was injected. The eluate was of EBT in EMS has not yet been determined, preparations of further diluted if the response of the injected test solution was out of the response range of the standard curve. The EBT peak in the @Abstractpublished in Advance ACS Abstracts, February 1, 1994. chromatogram of the test solution was identified by comparing the 720 /Journal of Pharmaceutical Sciences Vol. 83, No. 5,May 1994

This article not subject to U.S. copyright.

Published 1994, American Chemical Society and American Pharmaceutical Association

Table 1-Llquld

Chromatography Condltlons

Parameter

Condition

Column

Mobile phase A Mobile phase B Flow rate Gradient profile

Detector Temperature

Delta Pak CI8 (Catalog no. 98801,Waters) 3.9 mm X 30 cm, 5 pm (particle size), 300 (pore size) 0.1% TFA in water 0.1 % TFA in acetonitrikwater (8:2) 1 mL/min min % A %B Gradient 0 80 20 linear 10 70 30 linear 25 20 80 step 35 20 80 linear UV, 280 nm ambient

a

50 mV

40 30 -

I

EBT

W

-

20

retention time with that of the reference standard (the retention time of EBT was -19.5 min). Calculation-The EBT level was reported as pg of EBT/g of test portion. The level of EBT was calculated as follows: A = 0.5 g X 50 mL/100 mL = 0.25 g, and EBT (pg/g) = B X (T.,/Z.,)X (l/A), where A is the weight of test portion represented by eluate (0.25 g), B is the amount (pg) of EBT in eluate injected (Bwas determined from standard curve), T., is the final volume of eluate (4000 pL), and I., is the volume of eluate injected (500 pL for recovery study).

B

rduOr

3

50

Results and Discusslon EBT is more readily soluble in acidic or basic solution than in water and is more stable in basic than in acidic solution. A pH 9.2 buffer solution, 12 mg of EBT/100 mL of 0.1 M TEAA kept at 5 "C for 1month gave the same absorbance at 280 nm as the original solution before storage. EBT is not very stable in acidic solution. EBT in 0.1% TFA solution kept at 5 OC decomposed completely within 1week.5 The HPLC chromatograms at the different time intervals were identical. Because W is more soluble in 0.1 % TFA solution than in water [pKa values for W are 2.43 and 9.441, spiked test portions were dissolved in TFA solution and the cleanup step was performed immediately. Because base can cause the bonded-phase CIS cartridge to deteriorate, W was not dissolved in a basic solution. The solid-phase CIS cleanup column was used to eliminate most of the W and to concentrate the EBT. EBT was then eluted with the following five solvent systems: 30% acetonitrile in 0.1% TFA (7 mL); 70% acetonitrile in water (7 mL); acetonitrile (1.5 mL); methanol (1 mL, then 3 mL of water); and methanol (1.5 mL, then 2.5 mL of water). Both the 70% acetonitrile and the methanol systems gave >80% recovery of the EBT added to W at the 1.2 pg/g level. However, the eluate with 70 % acetonitrile required an additional step (evaporation to near dryness and dilution to 4 mL final volume) before injection onto the HPLC column. The volume of methanol (1.5 mL) was sufficient. Because the liquid hold-up volume of the cartridge was -0.85 mL, the addition of water to the cartridge was necessary to make the EBT elution more complete. The other eluting solvent systems either gave many broad peaks just before the EBT peaks or gave poor recoveries of the added compound (<50%).

HPLC analysis must be performed immediately after the solidphase extraction step. EBT is not stable in dilute methanol solution at low concentration (<1ng/pL). The cartridges could be stored at 5 "C overnight after the washing with 6 % acetonitrile and drying. Results of analyses were the same for eluates collected from cartridges both with and without storage in a refrigerator.

rnV 40

30

20 10 II

00"

I

5

10

15

20

25

30

4 35

Retention Time (min) 50 mV 40

30 20

10 I

OC 0

I

5

-I 15

10

20

25

30

35

Retention Time (min) Figure 2-Liquid chromatogramsfrom analyses of (A) control L-tryptophan powder and (B) spiked L-tryptophan powder (EBT at 4.8 pg/g).

The CIS cartridge cleanup step can be omitted if the test sample contains EBT at >74 pg/g. When a portion of the eluate containing an equivalent of -31 mg of the original test sample spiked at 1.2 pg of EBT/g of W (-37 ng of EBT) was injected onto the HPLC column for separation and quantitation, the signal-to-noise ratio was 1 0 1 and the recovery of the added EBT was -90%. Similar recovery (96%) and signal-to-noise ratio were obtained when 100 pL of the 0.1 % TFA extract from a test sample spiked at 74 pg/g was subjected to HPLC analysis. The limit of detection without the CIScartridge cleanup is -37 pglg. HPLC mobile phase starting with 80% A and 20% B linear gradient to 30% B in 10 min was used to separate W and

-

Journal of pharmaceutical Sciences / 721 Vol. 83, No. 5, May 1994

0

5

I

I

10

15 20 25 Retention Time (min)

30

35 I

I

B 40

Figure 2 shows the determination of EBT in W powder (with and without added EBT) by the described method (with Cla cartridge cleanup). W obtained from Tanabe U.S.A., Inc., contained no EBT and was used as the control for the spiking study. Average recoveries, based on peak heights, for replicate test samples spiked a t 1.2, 2.4, and 4.8 pglg were 93, 91, and 90%,respectively. The standard deviations (SDs) of the three levels were 10.4, 4.0, and 1.7; the corresponding relative SDs were 11.2, 4.4, and 1.9%. Average recoveries based on peak areas were 79,96, and 92%. At the limit of determination (0.6 pglg), recovery was >130%; a t such a low level, the slight background interference gave higher results. Recovery based on peak height a t this level was -50 % . Figure 3 shows the chromatograms obtained from a patientrelated test sample and a nonpatient-related test sample. All of the 10 patient-related test samples contained EBT a t >70 pg/g. The three nonpatient-related test samples either did not contain EBT or contained EBT a t <2 pg/g.

References and Notes 0

5

10

15 20- ---25 Retention Time (rnin)

30

35

Figure 3-Liquid

chromatograms from analyses of (A) nonpatient-related L-tryptophan test sample (equivalent to 31 mg of original compound/ injection)and (B) patient-related L-tryptophan test sample (equivalent to 0.5 mg of original compoundlinjection). other impurities from EBT. An isocratic elution at 30 % B for an additional 15 min was necessary to separate EBT from interferences. A step gradient to increase B to 80% for 10 min was used to flush all the late-eluting components off the column. Figure 1 shows the chromatograms of a test sample (containing no EBT) spiked a t 74 pglg and a test sample containing EBT at 124 pg/g analyzed by the direct injection method without the CIS cartridge cleanup step (with the CIScartridge cleanup step, the same contaminated test sample was found to contain EBT at 143 pglg). W was the major peak. A larger injection volume would determine EBT a t a lower level but would also increase the noise level due to matrix interference.

722 / Journal of pharmaceutical Sciences Vol. 83, No. 5, May 1994

1. Hertzman, P. A.; Blevins, W. L.; Mayer, J.; Greenfield, B.; Ting, M.; Gleich, G. J. N. Engl. J. Med. 1990,322, 869-873. 2. Silver, R. M.; Heyes, M. P.; Maize, J. C.; Quearry, B.; VionnetFuasset, M.; Sternberg,E. M. N. Engl. J. Med. 1990,322,874-881. 3. Belongia, E. A.; Hedberg, C. W.; Gleich, G. J.; White, K. E.; MacDonald, K. L.; Osterholm, M. T. N. Engl. J. Med. 1990,323, 357-365. 4. Slutsker, I,.; Hoesly, F. C.; Miller, L.; Williams, L. P.; Watson, J. C.; Fleming, D. W. J. Am. Med. Assoc. 1990,264, 213-217. 5. Trucksess, M. W. J. Chromatogr. 1993,630, 147-150. 6. Centers for Disease Control Morbid. Mortal. Weekly Rep. 1990, 589-591. 7. Toyooka, T.; Yamazaki, T.; Tanimoto, T.; Sato, K.; Sato, M.;

Toyoda, M.; Ishibashi, M.; Yoshihira, K.; Uchiyama, M. Chem. Pharm. Bull. 1991,39,820-822. 8. Mayeno, A. N.; Lin, F.; Fodte, C. S.; Loegering, D. A.; Ames, M. M.; Hedberg, C. W.; Gleich, G. J. Science 1990, 250, 1707-1708. 9. Smith, M. J.; Mazzola, E. P.; Farrell, T. J.; Sphon, J. A.; Page, S. W.; Ashley, D.; Sirimanne, S. R.; Hill, R. H., Jr.; Needham, L. L. Tetrahedron Lett. 1991,32,991-994. 10. Love, L. A.; Rader, J. I.; Crofford,L. J.; Raybourne,R. B.;Principato, M. A.; Page, S. W.; Trucksess, M. W.;Smith, M. J.; Dugan, E. M.; Turner, M. L.; Zelazowski, E.; Zelazowski, P.; Sternberg,E. M. J. Clin. Invest. 1993, 91, 804-811.