1514
Anal. Chem. 1991, 63,1514-1519
Determination of Reactive Nitrogen Mustard Anticancer Drugs in Plasma by High-Performance Liquid Chromatography Using Derivat ization Jeffrey Cummings,* Alexander MacLellan, and John F. Smyth
Imperial Cancer Research Fund, Medical Oncology Unit, Western General Hospital, Edinburgh, EH4 2 X U U.K. Peter B. Farmer
Medical Research Council, Toxicology Unit, Carshalton, Surrey, SM5 4EF U.K.
A hlgh-performance llquld chromatography (HPLC) method Is described for the determhratbnof reacthe nllrogen mustard anticancer drugs In plasma after derlvatlzatlon wlth dlethyldlthkcarbamlc acM (DDTC). mee compounds were studied: two reactlve specks (mechlorethamlne (HN2) and galactose 6-mustard (0-6-M)) and a less reactlve species (melphalan (L-PAM)) Included for vaHdaHon experiments. M a s s and NMR spectrometry confirmed that one molecule of DDTC reacts wlth each arm of the mustard, dlsplaclng a chlorine atom to form a stable dlsubstltuted adduct. Wlth the reactlve mustards a 30-mln Incubation at 37 OC Is recommended for >90 % derlvatlzatlon efflclency. Gradient elutlon was employed to analyze all three compounds uslng the same condltlons with a pBondapak C,, 1 0 - m particle slze colmn (30 cm by 3.8 mm 1.d.). The retention time (tR) of HNP-DDTC, was 13.1 mln f 1.5% wlthln day CV; tR of L-PAM was 7.6 mln and L-PAM-DDTC, was 14.6 mln f 0.8% CV. 0-6-MDDTC, ylelded a double peak, tR = 10.7 mln and 10.9 mln f 2.9% CV. The llmlt of detectlon on column was 0.5 ng for HN2, 1 ng for L-PAM, and 5 ng for G-6-M. A solld-phase sample preparatlon technique uslng “Bond Elut ” phenyl Is descrlbed that extracts from plasma Gd-M-DDTC, wlth >74% efflclency and HN2-DDTC, wlth >90% efflclency. When the drugs were derlvatlzed In plasma, recovery remalned hlgh for G-6-M (>84%) but dropped to 50% for HN2. The chemlcal half-life In plasma was 9 mln for G6-M and 37 mln for HN2. For phannacoklnetk studles, lt Is recommended that 10 mg of DDTC/mL of blood is added to specknen contalners In order to trap the reactlve mustard In sltu prlor to analysis by HPLC.
INTRODUCTION The nitrogen mustards represent the first group of synthetic chemical reagents to demonstrate clear anticancer activity in man ( I ) , and clinical trials with mechlorethamine (Figure 1, nitrogen mustard, HN2), the lead compound in the series, dating back to the 40s reported dramatic responses in lymphoma patients (2-4). Due to high chemical reactivity and a presumably short half-life in man, HN2 exhibits limited therapeutic activity against the major solid tumours but remains highly toxic to the bone marrow. Out of the many analogues that have been synthesised only a small number of compounds have proved to have a better therapeutic index than HN2 with three notable examples being melphalan (Figure 1, L-PAM), chlorambucil, and cyclophosphamide (along with its isomeric form ifosfamide). These three all have electron-withdrawing groups substituted on the nitrogen atom,
* Corresponding author. 0003-2700/91/0363-15 14$02.50/0
which reduces its nucleophilicity and results in a considerably less reactive compound. Whilst the trend has been toward less reactive nitrogen mustard species, no correlation has been shown between alkylating activity and biological efficacy (5)and it is clear that equally important to chemical reactivity is the nature of the carrier group. Attachment of nitrogen mustard to the C6 position of glucose or galactose produces compounds with activity greater than HN2 and equal to L-PAM in animal models but with greatly reduced bone marrow toxicity ( 6 ) . Addition of the sugar moeity is unlikely to confer stability to the nitrogen mustard group. One of these compounds, galactose 6-mustard (Figure l, G-6-M), has recently entered clinical trials in both the USA and the U.K. Since the publication of colorimetric assays employing nitrobenzyl pyridine (NBP) for the estimation of alkylating compounds in 1955 (3,and for the estimation of HN2 (8-IO), little advance has been made in developing chromatographic techniques for the determination of reactive nitrogen mustards. NBP assays in their own right are not sufficiently sensitive or selective to be applied to plasma samples from pharmacokinetic studies (11). A number of chemical reagents have been tried unsuccessfully in order to trap HN2 in a stable form that can then be further analyzed (12). Recently, it has been demonstrated that diethyldithiocarbamic acid (Figure 2, DDTC) a strongly nucleophilic derivatizing agent, reacts efficiently with HN2 liberated during microsomal incubation of nitromin, and the resultant disubstituted adduct can be subjected to both HPLC and GC (12). Independently, we have been investigating DDTC as a means of developing a routine chromatographic assay for reactive nitrogen mustards. In this paper, we present the first sensitive HPLC method, along with a reproducible and efficient sample preparation technique, for the determination of two reactive nitrogen mustards, HN2 and G-6-M, in plasma based on derivatization with DDTC.
EXPERIMENTAL SECTION Materials. L-PAM (free base) and DDTC (sodium salt) were from Sigma Chemical Co. (Poole, U.K.), HN2 (hydrochloride salt) was from Aldrich Chemical Co. (Poole, U.K.), and G-6-M (hydrochloride salt) in 50-mg vials as a freeze-dried powder was supplied by Dr. Harold J. Tenoso, Unimed Inc., (Sommerville, NJ). Methanol and acetonitrile were HPLC reagent grade (Rathburn Chemicals, Walkerburn, Scotland), orthophosphoric acid and sodium hydroxide were AnalaR grade (BDH, Poole, U.K.). Water was deionized and bidistilled in a quartz glass still. All other chemicals were of the highest grade available commercially. Safety Precautions and Handling of Nitrogen Mustard Solutions. HN2, G-6-M,and L-PAM are all extremely hazardous substances and carcinogens. Solid material, never greater than 20 mg, was weighed out in a safety cabinet and transferred directly into a volumetric flask, which was then stoppered. The stoppered flasks were transferred to the open bench and the solids were only 0 1991 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 63, NO. 15, AUGUST 1, 1991
-
CH3 N
0.999 in each case, and the data are shown in Table 11. On a molar basis the
(In,
curves for L-PAM and HN2 were almost superimposible, reflecting their e q d y high derivatization efficiency of 90%+. G-6-M, on the other hand, yielded a standard curve of only 26% the value of the other two compounds and hence a higher limit of detection (Table 11). This may be due to instability of the galactopyranose ring, which is capable of undergoing polymerization reactions. Table I1 also shows the interday CV in calibration curves. In most casea the CV was less than 20% with a tendency for a higher value at lower concentrations where a greater number of dilutions had been performed and where there had been a longer time delay before adding DDTC. These data highlight the need to work as quickly as possible with underivatized nitrogen mustard solutions. Extraction Efficiency and Assay Accuracy and Precision. The extraction efficiency of different concentrations of preformed HN2 and G-6-M disubstituted adducts from aqueous standard solutions and plasma is shown in Table III. In each case recovery was high. Accuracy and precision data for the whole assay (derivatization, sample preparation, and HPLC determination) are also contained in Table 111. With G-6-M both in plasma and whole blood theae values remained consistently good with between-day CV's never greater than 17%. However, with HN2 the accuracy was 50% of the expected value in plasma, even though precision was good (always less than 10% for within- and between-day Cv's), and accuracy was significantly greater than 100% of the expected value in whole blood. The latter value may be high since, even though 10 pg of HN2 was added to 1mL of whole blood, the concentration in the plasma that was eventually analyzed could have been much higher because HN2 has been shown not to be significantly taken up into blood cells (21). Addition of 1pg/mL of both compounds simultaneously to plasma followed by derivatization resulted in no significant loss of recovery (49.2% f 7.1 SD for HN2 and 66% 12.3 SD for G-6-M), showing that one drug could be used with the other as an internal standard. However, one would have to
*
1518
ANALYTICAL CHEMISTRY, VOL. 63, NO. 15, AUGUST 1, 1991 HN2
1
I, 50
01nw-L'/
,
5
10
15
-
20
Time (minutes)
Flgure 5. Chromatograms of 10 pg/mL nitrogen mustard added to plasma, derivatized, and extracted on "Bond Elut" phenyl as described in the Experimental Section: (A) 1-mL plasma plus 10 mg of DDTC, no nitrogen mustard: (B) galactose 6-mustard; (C) mechlorethamine.
take into account the large difference in expected recovery between the two. The solid-phase method did not extract any endogenous component from plasma that could interfere with the identification of HN2 and G-6-M (see Figure 5). This permitted plasma samples to be analyzed a t high sensitivity, with the limit of detection being 1 ng/mL for HN2 and 10 ng/mL for G-6-M. Chromatograms of plasma extracts containing either HN2 or G-6-M are shown in Figure 5. The underestimation of HN2 in plasma is probably due to a fraction of the drug reacting preferentially with cellular nucleophiles. Cisplatin, which is less reactive than HN2, binds irreversibly to plasma proteins with a half-life of 2 h (22)and when added to plasma disappears with a half-life of 56 min (18). It also binds to a series of small molecule nucleophiles such as glutathione, cysteine, and methionine, many of which DDTC is also unable to displace (16). When L-PAM is added to plasma only a small fraction of the drug disappears at 23 "C (22% after 6 h), and this is believed to be due to hydrolysis rather than covalent binding to plasma proteins (23). This is consistent with the low intrinsic alkylating activity of LPAM compared with HN2. Indeed, if HN2 is incubated with DDTC in phosphate buffer, only low levels of the disubstituted adduct are formed due to the preferential reactivity of the mustard with phosphate (12). The very high reactivity of G-6-M with DDTC, as exemplified in the kinetic studies reported in this work, may explain why recovery of this drug from plasma was high. Stability of Mechlorethamine (HN2) a n d Galactose 6-Mustard (G-6-M) in Plasma at 37 OC. The stability of 100 or 1 pg/mL HN2 and 100 pg/mL G-6-M in plasma was studied. The decay curves are shown in Figure 6 for 100 rg/mL G-6-M and 100 pg/mL HN2. The curve for 1 pg/mL HN2 was identical with that for 100 pg/mL and is not shown in Figure 6. G-6-M decayed rapidly from plasma through a first-order exponential decline with a half-life of 9 min, and by 1 h >98% of the drug had disappeared. The decay curve for HN2 was more complicated but fitted well a first-order exponential decline with a half-life of 37 min. Handling of Clinical Specimens for the Determination of Reactive Nitrogen Mustards by HPLC. At present, clinical trials with G-6-M are ongoing in both the U.S.A. and the U.K. with view of performing pharmacokinetic studies. Considering the instability of the drug in plasma, this would appear to be a daunting task. Nevertheless, if DDTC can be added to plasma without delay, then in excess of 90% of the native drug can be trapped. The problem with blood specimens collected in the clinic is the unspecified time that samples will be delayed prior to analysis in the laboratory. Even a 10-min delay would invalidate HPLC analysis because in excess of 50% of the drug would have disappeared due to
Time [minulei)
Flgure 8. Chemical stability of 100 pglmL mechlorethamine and galactose 6-mustard in plasma, at 37 "C: (0) mechlorethamine ( n = 3); (0)galactose 6mustard (n= 3). The drugs were added to plasma, and at serial time points DDTC was added: the concentration was then determined by HPLC.
either hydrolysis or covalent binding. In order to determine whether it would be possible to trap the native drug immediately after samples were taken in the clinic, a study was performed with whole blood. Here DDTC was added to fresh blood, and then the drug was added. Under those conditions in excess of 90% of the native drug was trapped. Thus, we believe it is feasible to perform pharmacokinetic studies by using specimen bottles that contain 10 mg of DDTC in solid form for every 1 mL of blood to be collected. Once a blood sample is taken, it is immediately placed in the vessel containing DDTC, mixed by two to four inversions, and allowed to sit at room temperature for 10-15 min, by which time greater than 90% of G-6-M will have reacted to form the stable disubstituted DDTC adduct. For G-6-M no water bath is required, but for HN2 it is recommended to incubate the whole blood for 1 h at 37 "C. Thereafter, samples can be spun down and stored prior to extraction on phenyl minicolumns and HPLC. The data presented in this paper form a basis to enable a class of compounds, reactive nitrogen mustards, which hitherto have proven impossible to measure by HPLC, to be determined in both aqueous solutions and plasma. Each of the three compounds studied was structurally different, and each behaved in a characteristic manner. Nevertheless, all three were amenable to analysis using the same basic HPLC conditions. The two reactive mustards yielded unexcepted findings. G-6-M appeared to react with DDTC with low efficiency (26%), and when HN2 was added to plasma, analytical recoveries dropped to 50%. These observations indicate caution when attempting to determine another nitrogen mustard from the compounds studied in this paper and highlight the problems working with reactive compounds. ACKNOWLEDGMENT We are grateful to J. Lamb, MRC Toxicology Unit, for mass spectrometry and Dr. D. E. V. Wilman and L. Griggs, Institute of Cancer Research, Sutton, U.K., for NMR spectrometry. LITERATURE CITED (1) Einhorn, J. Int. J . Radiat. Oncol., Biol, Phys. 1985, 1 7 , 1375-1378. (2) Goodman. L. S.;Wlntrobe. M. M.; Dameshek, W.; Goodman, M. J.; Gilman, A. JAMA. J . Am. Med. Assoc. 1848, 132, 126-132. (3) Jacobson, L. 0.;Spurr, C. L.; Barron, E. S. G.; Smith, T.; Lushbaugh, C.; Dick, G. F. JAMA, J . Am. Med. Assoc. 1946, 132, 263-271. (4) Rhoads, C. P. JAMA, J . Am. Med. Assoc. 1946, 131, 656-658. (5) Godenuche, D.; Madelmont, J.C.; Moreau, M.-F.; Plagne, R.; Meyniel. G. Cancer Chemother. pharmacal. 1980, 5 , 1-9. (6) Cantrell, J. E.; Green, D.; Schein, P. S. Cancer Res. 1986, 4 6 ,
2340-2343.
Anel. Chi" 1991, 63,1519-1523 (71 Epsteln, J.; Rownthal, R. W.; Ess, R. J. AM/. chsm. 1951, 2 7 , 1435- 1439. (8) Klett, D.; CHtfkr, A. C.; S W n , J. S., Jr. Proc. Soc, Exp. Bkl. Msd. 1960. 104, 829-631. (9) A m n , R. K.; Crevar, (3. E.; HagdOrn, H.; Bardos, T. J.; Ambrus, J. L. JAMA,'J. Am. A M . Assoc. 1901, 178. 735-738. (10) Frbdman, 0. M.; Bogor, E. AM/. Chum. 1901, 2 3 , 906-910. (11) Skbbe. J. L.; cdln~. F. (3. J . h - c d . Me1980, 4 , 155-163. (12) Whke, 1. N. H.; Suzenger. M.; Mattocks. A. R.; Baby, E.; Farmer, P. 0.: Conners, T. A. Cercif"sls 1989, 10. 2113-2118. (13) Munger, D.; Sternson. L. A.; Repta, A. J.; Higuchi, T. J . Chrometogr. 1977, 143, 375-382. (14) Bannister, S. J.; Stemson, L. A.; Repta, A. J. J . chrometogr. 1979. 173, 333-342. (15) Borch. R. F.; Markovltz, J. H.; Pleasants, M. E. AM/. Lett. 1979, 12. 917-926. (18) Andrews, P. A,; Wung, W.; Howell, S. 0. AM/. Blochem. 1984. 143, 46-56.
1518
(17) Drummer, 0. H.; Roudfoot, A.; Hoves, L.; Louls, W. J. C h . Chem. ACte 1904, 136, 65-74. (18) Reece, P. A.; McCaH, J. T.; Powls, 0.; ~lchardwn,R. L. J . clnwne1904, 306, 417-423. (19) BosanqWt, A. C e W h ~ c d1985, . 14, 83-95. (20) Hlncal, A. A.; Long, D. F.; Repta, A. J. J . Parenter. Dnrg Assoc. 1979, 33, 107-116. (21) Sklpper, H. E.; Bennett. L., Jr.; Langham, W. H. Cancer 1951, 4 , 1025-1 027. (22) LeRoy, A. F.; Thompson, W. C. J . Ne#. Cencw Inst. 1989, 8 1 , 427-436. (23) Chang, S. Y.; Alberts, D. S.; Melnlck, L. R.; Walson, P. D.; Salmon, S. E. J . phenn. Scl. 1979, 67, 679-682.
m.
.-
RECEIVFSfor review September 12,1990. Revised manuscript received March 26,1991. Accepted April 24, 1991.
Factors Affecting Direct Control of Electroosmosis Using an External Electric Field in Capillary Electrophoresis S.Lee,* Douglass McManigill,' Chin-Tiao Wu, a n d Bhisma Pate1 Department of Chemical and Biochemical Engineering, University of Maryland Baltimore County Campus, Baltimore, Maryland 21228 Cheng
The change In the direction and flow rate of e l e c t r m o d s in ccrpllrrry e h b p b m k wlth the appllcatkn of an external electric field has been demonstrated and measured by the currentmonltorkrgnnthod. Factonsuchassolutkncondl#on and capillary dimendon affecting the direct control of eiectroounork have been measured and analyzed In detail wlth theproposedcapsdtormodd. ~cspsdtormodelpredlcted reasonably well the trends of experimental results at variow solution PHI, electrolyte concentrations, and capillary dlmenslonr.
INTRODUCTION We have recently proposed and demonstrated the direct control of electroosmotic flow in capillary electrophoresis by using an additional electric field applied from outside the capillary ( I ) . This technique vectorially couples the externally applied potential with the potential across the buffer solution inside the capillary. This electric potential gradient across the capillary wall controls the polarity and magnitude of the t potential on the interior surface of the capillary wall. The direction and flow rate of electroosmosis are dependent upon the polarity and magnitude of the potential (2). The electrooamosis in capillary electrophoresis at the solution condition of 1mM phosphate buffer at pH 5 was directly controlled by simply varying the external electric potential (1). To investigate the fundamentals of the use of an external electric potential for controlling the electroosmotic flow, a capacitor model as shown in Figure 1is proposed in this study. Under normal aqueous conditions with small binary electrolytes, the silica surface has an excess of anionic charge resulting from ionization of surface functional groups. The cationic counterions to these anions are in the diffuse layer adjacent
r
*Towhom all corres ondence should b addressed. lHewlett-Packard Lagoratories, 1501 Page Mill Rd,Palo Alto,CA
to the capillary walls. The potential across the diffuse layer is termed the t-potential(2). The capacitance of the electrostatic diffuse layer at the inner capillary/inner aqueoue interface, Cei in faradays (F)a t 25 O C is given by Cei = (7.23 X F cmaS6 mola*s)C'/2cash [(19.46 V-'){]?rDi.d.l (1) where C is the concentration of ions in the solution in mol/cm3, b is the f potential at the capillary interface in volts, Did. is the inner diameter of inner capillary tubing in centimeters, and 1 is the length of capillary tubing in centimeters (2). The capacitance of the inner capillary tubing, Cc is given by
cc = %2rl/[ln (Do.d./Di.d.)l
C , = (7.23 x F cma.6
mola.6)C'/2 cash [(19.46 V-')f]rD,d.l (3)
where Do.&is the outer diameter of inner capillary tubing in centimeters (2). The total capacitance of three capacitors in series, CT, is given as (3) (4) The capacitance of the capillary tubing and the electrostatic diffuse layer studied in this project are in the range of 10-l' F and 10-6-104 F, respectively. Thus, the reciprocal of the capacitance of the capillary tubing is much greater than the reciprocal of the capacitance of the electrostatic diffuse layer. The reciprocal of the total capacitance, (C&', and the total capacitance, C T , can then be simplified as
(CT)-' =
94304.
0003-2700/~1/0363-1519$02.50/0
(2)
where ec is the electrical permittivity of silica surface and Dod is the outer diameter of the inner capillary tubing (3). The capacitance of the electrostatic diffuse layer at the inner capillary/outer aqueous interface (in the annular space between the outer and inner capillaries), C , in faradays at 25 O C is given by
0 1991 American Chemlcal Society
(cc)-'
(5a)
