Anal. Chem. 1997, 69, 5006-5010
In Vivo Determination of Ultratrace Amounts of Prostaglandin in Plasma by High-Performance Liquid Chromatography/Laser-Induced Fluorometry/Ultrasensitive Laser Spectrometry under Severe Conditions Reiko Tsutsumiuchi, Hiroshi Saito, and Takashi Imagawa
Central Research Laboratories, The Green Cross Corporation, 2-25-1, Shodai-Ohtani, Hirakata, Osaka 573, Japan Takehiko Kitamori, Tamao Odake, and Tsuguo Sawada*
Department of Applied Chemistry, School of Engineering, The University of Tokyo, 7-3-1 Hongo, Bunkyo, Tokyo 113, Japan
We succeeded in determining ultratrace prostaglandin amounts in plasma, at the femtomolar level, using laserinduced fluorometry through a complete redesign of the analytical procedures. Practical samples, especially plasma, contain large amounts of admixtures, and prostaglandin in plasma (pg/mL) has been considered to be difficult to detect because the samples and reagents supplied by conventional procedures are neither pure nor stable enough to get good results by ultrasensitive laser spectrometry. We completely redesigned the analytical procedures after careful investigations of the reagent purification and the column separation conditions based on a newly found behavior of the reagent and derivatized prostaglandin in a small quantity of ethanol in the mobile phase. A lower determination limit of 23 pg/mL (65 fmol) was achieved, the variance was 12% at 25 pg/mL, and the recovery rate was 88-89%. This method was applied to in vivo analysis of the concentration of prostaglandin E1 administered as a prodrug of prostaglandin E1 (∆8-9O-butyryl prostaglandin F1 butyl ester, AS-013) by intravenous infusion to beagle dogs. A clear correlation between the change of blood pressure and the prostaglandin E1 concentration was confirmed. Highly sensitive laser spectrometries, such as laser-induced fluorometry and photothermal spectrometry, have been proposed and shown to be promising for ultratrace and ultramicro analyses. In most of these reports, analyses were made of standard samples, and detection at sub-ppt or subpicogram levels was possible. Such successful results from methodological studies gave an expectation of practical applications, though to date, there are few reports for practical samples. This is because, in biological samples in particular, amounts of admixtures are quite large compared with the ultratrace amount of a substance to be determined, and as is well-known, these admixtures often cause difficulty in the ultratrace analysis. Furthermore, most of the derivatization reagents used in the conventional analysis are usually neither stable nor pure enough to be applied to the ultratrace region; even if their decomposition products and/or byproducts are negligibly small in the ordinary trace region, they make the background fluctuate, * To whom all correspondence should be addressed: e-mail, tsawada@hongo. ecc.u-tokyo.ac.jp; fax, +81-3-3815-6543.
5006 Analytical Chemistry, Vol. 69, No. 24, December 15, 1997
and they interfere with the signal from the ultratrace analytes. Moreover, it is difficult to find a suitable reagent that is stable enough and the excitation wavelength of which coincides with the laser radiation. Therefore, practical applications of ultrasensitive laser spectrometries have been considered to be difficult, though their superior potential is known. This difficulty is completely different from that encountered in studies on spectrometric methodologies, so by complete redesign of the analytical procedures, even a conventional reagent can be applied to laser spectrometry, and practical applications of laser spectrometry become possible in various fields. Analysis of prostaglandins (PGs) in plasma, on the other hand, is of crucial importance in the fields of medicine and biochemistry, and a successful method for their microanalysis has been eagerly awaited. Radioimmunoassay (RIA)1,2 and mass spectrometric detection combined with gas chromatographic separation (GC/ MS)3-7 have achieved high sensitivity in the several picogram per milliliter region. Laser-induced fluorometric detection combined with high-performance liquid chromatographic separation (HPLC/ LIF) is also expected to be a good candidate for sensitive detection of PGs, and several groups have reported microanalysis methods of PGs using HPLC/LIF. McGuffin and Zare8,9 used 4-(bromomethyl)-7-methoxycoumarin (Br-MMC) as a fluorescence-labeling reagent and a He-Cd laser as the light source, and detected PGs in pharmaceuticals as a real sample matrix on the femtomole level. Goto et al.10,11 used 9-anthryldiazomethane (ADAM) as a fluorescence-labeling reagent and a He-Cd laser as the light source to analyze PGs in human gastric mucosa, and they successfully (1) Hotter, G.; Rosello´-Catafau, J.; Closa, D.; Bioque, G.; Gelpı´, E.; Javerbaum, A.; Gonza´lez, E.; Gimeno, M. A. F. J. Chromatogr., A 1993, 655, 85-88. (2) Kominami, G.; Nakamura, M.; Mizobuchi, M.; Ueki, K.; Kuroda, T.; Yamauchi, A.; Takahashi, S. J. Pharmacol. Biomed. Anal. 1996, I5, 175182. (3) Cawello, W.; Schweer, H.: Mu ¨ ller, R.; Bonn, R.; Seyberth, H. W. Eur. J. Clin. Pharmacol. 1994, 46, 275-277. (4) Obata, T.; Nagakura, T.; Kammuri, M.; Masaki, T.; Maekawa, K.; Yamashita, Y.; Ishibashi, K. J. Chromatogr., B 1994, 655, 173-178. (5) Mizugaki, M.; Hishinuma, T.; Yu, G. S. P.; Ito, K.; Nishikawa, M.; Ohyama, M.; Nakagawa, Y.; Harima, N. J. Chromatogr., B 1994, 658, 11-19. (6) Nourooz-Zadeh, J.; Gopaul, N. K.; Barrow, S.; Mallet, A. I.; A ¨ nggård, E. E. J. Chromatogr., B 1995, 667, 199-208. (7) Margalit, A.; Duffin, K. L.; Isakson, P. C. Anal. Biochem. 1996, 235, 7381. (8) McGuffin, V. L.; Zare, R. N. Appl. Spectrosc. 1985, 39, 847-853. (9) McGuffin, V. L.; Zare, R. N. Proc. Natl. Acad. Sci. U.S.A. 1985, 82, 83158319. S0003-2700(97)00821-4 CCC: $14.00
© 1997 American Chemical Society
detected the PGs at subpicomole per milligram tissue levels. In our previous report, we developed a method for microanalysis of PGs with a YAG laser using ADAM reagent and detected urinary PGs at subnanogram per milliliter levels.12 In these cases, however, amounts of derivatized PGs were present at nanogram or milligram per milliliter levels and were considered able to override interference by contaminants in the fluorescence-labeling reagent and sample matrix. In contrast, the amount of a plasma sample that can be used for analysis (usually 1-2 mL) is smaller than that of gastric mucosa or urine, and expected plasma concentrations of PGs are on the level of picogram per milliliter plasma. Since the plasma is the sample matrix most likely to be affected by the problems mentioned above, it appears to be extremely difficult to detect ultratrace amounts of PGs in plasma. Moreover, most of these fluorescence reagents are highly reactive and also highly degradable. Contaminants contained in the fluorescence-labeling reagent and in the catalytic reagent used with it in some cases, an excess of reagents remaining after derivatization, and degradation products are the major factors that increase background fluctuations during analysis. When the HPLC/LIF method is applied to analysis of practical samples, the analysis is susceptible to interference from the listed contaminants. Even if the effects of these factors are small in ordinary microanalysis, they raise serious problems for laser ultramicroanalysis. In vivo analysis of PG reported here is an exemplary case in which such problems occur. PGE1, used in this report, is a compound with strong vasodilating effects, and it also suppresses platelet aggregation. It has been clinically used for treatment of ischemic ulcer associated with chronic arterial occlusion and peripheral circulatory disorders associated with health hazards due to vibration.13-15 Since PGE1 has very strong physiological activities at very small doses, a method able to determine plasma PGE1 concentrations at the picogram level is required for studies of its pharmacokinetics. As mentioned above, however, the amounts of contaminants are much larger than the amount of PGE1 in the plasma. ADAM, which is used as the fluorescence-labeling reagent in the current study, is a superior reagent for fatty acid analysis. However, ADAM is highly degradable based on the criterion of ultratrace analysis. Therefore, an analytical combination of plasma PGE1 and laser spectrometry has been considered difficult. In this study, we were able to overcome the problems resulting from severe conditions by redesigning the analytical procedures. We carried out laser-induced fluorometry, in the femtomolar region, in plasma, which is one of the most heavily admixed samples, using an ordinary fluorogenic reagent which contains byproducts and is easily decomposed. By improving the conditions for purification of the ADAM reagent and derivatization, as well as procedures including separation of derivatized PGE1, we were able to set the conditions which enable analysis of plasma (10) Goto, H.; Sugiyama, S.; Kawabe, Y.; Kuroiwa, M.; Ohta, A.; Tsukamoto, Y.; Nakazawa, S.; Owaza, T. Biochem. Int. 1990, 20, 1119-1125. (11) Goto, H.; Sugiyama, S.; Ohara, A.; Hoshino, H.; Hamajima, E.; Kanamori, S.; Tsukamoto, Y.; Owaza, T. Biochem. Biophys. Res. Commun. 1992, 186, 1443-1448. (12) Sakae, S.; Harata, A.; Kitamori, T.; Sawada, T.; Okubo, A.; Toda, S.; Shimizu, T. Microchem. J. 1994, 49, 355-361. (13) Carlson, L. A.: Eriksson, I. Lancet 1973, (Jan 20), 155-156. (14) Carlson, L. A.; Olsson, A. G. Lancet 1976, (Oct 9), 810. (15) Takeda, J.; Uemichi, S.; Ohshiro, T.; Okada, M.; Shimomura, T.; Yamada, T.; Yamamoto, M. Jpn. J. Clin. Exp. Med. 1986, 63, 2423-2432.
Figure 1. A block diagram of the experimental setup (HPLC/LIF system).
PGE1 to the level of 10 pg/mL. We also administered a prodrug of PGE1 (∆8-9-O-butyryl prostaglandin F1 butyl ester, AS-013) to beagle dogs and determined plasma PGE1 concentrations in an in vivo study. EXPERIMENTAL SECTION Apparatus. The apparatus used at each stage of the procedures are discussed separately. The following instruments were used for all the procedures in the current study. A laboratoryconstructed system was used as the HPLC/LIF system, the structure of which is shown in Figure 1. The two systems used for HPLC were made by Tosoh. The excitation light source for the LIF system was a UV He-Cd laser (Omnichrome) with a wavelength of 325 nm and output of ∼40 mW. The flow cell was of our own design, in which disturbance by wall-scattered light was minimized using an optical fiber for fluorescence collection. Details of construction of the flow cell were given previously.12 Fluorescence at 410 nm (fluorescence peak wavelength of ADAM reagent) was detected by a photomultiplier tube (Hamamatsu Photonics) through an optical band-pass filter. Its intensity was amplified by reducing the noise with a lock-in amplifier. An isolation amplifier was used according to need. The normal-phase HPLC column (PGpakB, 4.6 mm o.d. × 250 mm, Japan Spectroscopic Co.) was used for HPLC/LIF analysis, and the reversedphase column (ODS-80Ts 4.6 mm o.d. × 150 mm, Tosoh) was used for separation of ADAM-PGE1 and for fractionating the analytes. Reagents and Materials. ADAM reagent made by Funakoshi was used. ADAM reagent was purified by solvent extraction using an n-pentane/water/ethanol mixture (1:1:1). The mixing ratio was our own selection, based on reasons discussed later. n-Pentane was added to the ADAM reagent (10 mg/mL), and the mixture was shaken gently. Then, a mixture of water and ethanol (1:1), twice as much as the amount of n-pentane, was added and the resultant mixture was again shaken gently. The supernatant n-pentane layer was filtered using a membrane filter (pore diameter: 0.5 µm) and dried under reduced pressure in the dark. The purified ADAM was stored in the dark at -20 °C. It was dissolved in 0.2% ethanol (dehydrated with anhydrous 10% w/v Na2SO4) immediately before use. PGE1 produced by the Green Cross Co. was used as sample. PGF1R produced by Funakoshi was employed as the internal standard. Chloroform (containing no ethanol as a stabilizer), nhexane, methanol, isooctane, and ethanol of fluorometric grade Analytical Chemistry, Vol. 69, No. 24, December 15, 1997
5007
Figure 2. Scheme of the analytical procedures.
(Kanto Chemical Co.) and ethyl acetate of spectrometric grade (Kanto Chemical Co.) were used. All other reagents were also of special grade. The cartridge columns used for solid-phase extraction were normal-phase Sep-Pak silica, reversed-phase Sep-Pak C18 columns, and Sep-Pak silica light, made by Waters Japan. Kieselgel 60F254 (Merck) was used as the thin-layer plate for pretreatment of the plasma. Analytical Procedures. An outline of the basic analytical procedures is shown in Figure 2. These steps are common through the study, but the important parameter setting and the design we developed are described in the next section. The pretreatment method of the plasma was designed from reports by Powell16 and Kurimoto and Sakurai.17 The plasma was subjected to ethanol extraction, solid-phase extraction in the SepPak C18 and Silica columns and normal-phase TLC fraction, and the PGE1 fraction was obtained. Then, 500 µL of 0.2% ADAM was added to the PGE1 fraction after pretreatment of the plasma or standard PGE1, which was left for 1 h in the dark at room temperature for derivatization. After derivatization, the substance thus obtained was dried under reduced pressure, and the unreacted ADAM and other degradation products were removed by solid-phase extraction using the Sep-Pak silica light. In addition, ADAM-PGs were separated from admixtures and contaminants by reversed-phase HPLC. After being dried under reduced pressure, the sample was analyzed by normal-phase HPLC/LIF. Details are described in the Results and Discussion section. (16) Powell, W. S. Prostaglandins 1980, 20, 947-957. (17) Kurimoto, F.; Sakurai, H. J. Clin. Exp. Med. (Igaku No Ayumi) 1987, 143, 323-325.
5008
Analytical Chemistry, Vol. 69, No. 24, December 15, 1997
RESULTS AND DISCUSSION In conventional microanalysis of fatty acids and carboxyl acids using ADAM, ADAM reagent was purified by n-pentane extraction, and a mixture of methanol and ethyl acetate was used as the solvent for derivatization. Separation of ADAM-fatty acids and ADAM-PGs after derivatization was performed by solid-phase extraction using the Sep-Pak silica light.12 When this method was applied to plasma samples, however, the PGE1 peak was obscured by background fluctuations and determination was impossible. This is because a large amount of ADAM is necessary for derivatization due to consumption of ADAM by coexisting carboxyl acids in a biological sample, although the amount of PGE1 is extremely small. And so, a large amount of decomposition products and byproducts of ADAM for the LIF detection are present in the sample. Accordingly, we attempted to improve these procedures. In investigation of the method of solid-phase extraction, in particular, we succeeded in identifying subtle differences in polarity of solvents as well as in behaviors of ADAMPGE1 and contaminants (surplus reagents) toward the alcoholic OH group. We were therefore able to establish the following appropriate conditions for separation even for severe conditions for ultrasensitive LIF detection. For purification of the ADAM reagent, we added a polar solvent to the conventional extraction solvent and used n-pentane/ ethanol/water (1:1:1), though only a nonpolar solvent has been reported to be satisfactory for the submicrogram per milliliter region. We found that polar contaminants present in small amounts could be removed with ethanol and water. With respect to solvents for ADAM derivatization, ADAM exhibits better reactivity in a solvent with stronger polarity such as methanol, but it is more degradable in such a solvent. If a large amount of PGE1 was present in a sample to be determined, degradation of ADAM would not cause any interference problems. However, in the case when PGE1 is present in only ultratrace amounts, degradation products may significantly interfere with determination. Ethyl acetate, chloroform, or acetone was used as a solvent to suppress the reactivity of the ADAM reagent in accordance with the previous example. But reactivity was too low to obtain sufficient recovery. Moreover, the acetic acid contained in ethyl acetate as a contaminant has a carboxyl group, which, it has been suggested, interferes with derivatization. After investigating both the reactivity and degradability of ADAM for several solvents, we chose ethanol as the best solvent. Considering separation of ADAM-derivatized PGE1, we were able to identify reliable conditions for careful separation in ultratrace analysis. These conditions were based on the newly found behavior of ADAM-PGE1, toward the alcoholic OH group, especially that in ethanol. Details of the behavioral investigation are being carried out and will be reported subsequently. Since derivatization of PG by ADAM is an esterification, the PG carboxyl group which causes the polarity disappears after derivatization, and ADAM-PGE1 is suggested to have weaker polarity. In any case, adsorption and desorption behavior of ADAM-PGE1 toward a silica column showed a drastic change for only a few percent of ethanol in the charge and irrigation solvents as described next. We first tested the solid-phase extraction method using the Sep-Pak silica light. The separation effects in solid-phase extraction changed depending on the small amount of ethanol added to the charge and irrigation solvent. On the basis of results of preliminary examinations, chloroform/ethanol (50:1; dehydrated
Figure 4. Chromatogram of PGE1 standard (50 pg/mL). PGF1R is the internal standard (500 pg/mL). HPLC/LIF conditions are the same as described for Figure 3. Table 1. The Recovery Rate of PGE1 in This Method PGE1 amt to be added (pg/mL plasma)
recovery amta (pg)
recovery ratea (%)
100.2 501.0
88.86 ( 0.51 442.4 ( 3.7
88.69 ( 0.51 88.31 ( 0.73
a
Figure 3. The effect of fractionation by the reversed-phase HPLC. Chromatograms of PGE1 in human plasma. PGF1R is the internal standard. (a) Reversed-phase separation HPLC: column, ODS-80Ts (4.6 mm o.d. × 150 mm); mobile phase, 75% methanol; column temperature, 50 °C; flow rate, 1 mL/min. (b) Normal-phase HPLC/ LIF analysis after fractionation: column: PGpakB (4.6 mm o.d. × 250 mm); mobile phase, isooctane/ethyl acetate/ethanol/acetic acid (75:20:5:2); column temperature, 40 °C; flow rate, 1.2 mL/min. LIF conditions: The light source is a He-Cd laser (output, 40 mW; wavelength, 325 nm). Detection wavelength, 412 nm. Chopper, 80 Hz. Photomultiplier voltage, 650 V. Lock-in amplifier, 50 mV full scale.
with absolute 10% w/v Na2SO4 after mixing of the solvent) was used as the charge solvent, and chloroform/n-hexane/ethanol (200:200:1; dehydrated with absolute 10% w/v Na2SO4 after mixing of the solvent) was used as the irrigation solvent. Moreover, since a large amount of ADAM reagent was necessary for derivatization of PGE1 in the plasma, solid-phase extraction alone was insufficient to completely eliminate the effects of admixtures. Accordingly, steps for separating ADAM-PGE1 and ADAM-PGF1R from admixtures using reversed-phase HPLC conditions (column, ODS-80Ts; mobile phase, 75% methanol) were introduced to the solid-phase extraction procedure. The chromatogram obtained during reversedphase separation from 1 mL of a plasma specimen to which 2 ng of PGE1 (and 2 ng of PGF1R as the internal standard) had been added and the chromatogram obtained by analysis with HPLC/ LIF after separation are shown in Figure 3. Figure 3 indicates a significant improvement in separation of ADAM-PGE1 and admixtures and foreign contaminants as a result of fractionation by reversed-phase HPLC. The peaks of admixtures that occur randomly along the retention time axis and peak heights around the PGE1 and PGF1R as shown in Figure 3a caused an unstable baseline. Such a baseline fluctuation always obstructs peak identification of the analytes in ultratrace quantities. Therefore, introducing this separation procedure was quite effective. As described above, we identified a procedure that allows highly sensitive LIF. Details will be reported elsewhere.
Mean ( SD (n ) 3).
The calibration curve obtained using a PGE1 standard solution of 25-500 pg/mL showed good linearity (correlation coefficient was 0.994). The lower limit of detection (LOD) defined as double the standard deviation was 23 pg/mL. The CV value at 25 pg/ mL was 12%. The chromatogram obtained using 50 pg/mL PGE1 is shown in Figure 4. The performance of the assay was favorable, although the procedure required many purification and separation steps. The LOD value was ∼3 orders superior to the conventional HPLC/UV-visible detection. Moreover, it is significant that our results were obtained for plasma, which is one of the most difficult matrixes for ultratrace analysis. For comparison, PGE1 standard solution was used in this study, and results were obtained using a procedure similar to that for cases in which plasma was used. No significant differences were observed between the chromatograms of standard and plasma samples. Accordingly, the results we obtained are considered applicable to plasma samples. To examine the plasma, a recovery test after addition of PGE1 was done: PGE1, at 100 or 500 pg, was added to 1 mL of the plasma, and the recovery test was performed. The results are shown in Table 1. The recovery rate was ∼88%, with only small variations in results (standard deviation of 1% or less). The amount of endogenous PGE1 measured in the plasma blank (1 mL) was less than the detection limit18 because aspirin, a cyclooxygenase inhibitor, was added immediately after blood sampling. We applied this method to the determination of plasma PGE1 concentrations in an in vivo study of Lipo AS-013. AS-013 is a new compound designed as a prodrug of PGE1 which has efficient conversion to PGE1 in the body (Figure 5). Lipo AS-013, i.e., AS013 incorporated into a fine lipid emulsion is a preparation expected to eliminate the instability problem of PGE1 and to provide good efficacy and safety.19 An in vivo experiment was performed in beagle dogs. Under general anesthesia induced by pentobarbital, Lipo AS-013 (5 µg/mL) was continuously administered through the femoral vein using an infusion pump (dose (18) Morris, H. G.; Sherman, N. A.; Shepperdson, F. T. Prostaglandins 1981, 21, 771-788. (19) Igarashi, R.; Mizushima, Y.; Takenaga, M.; Matsumoto, K.; Morizawa, Y.; Yasuda, A. J. Controlled Release 1992, 20, 37-46.
Analytical Chemistry, Vol. 69, No. 24, December 15, 1997
5009
Figure 5. Chemical structures of AS-013 and PGE1. Figure 7. Plasma PGE1 concentration and mean blood pressuretime curve before, during, and after intravenous infusion of Lipo AS013: (b) plasma PGE1 concentration; (O) mean blood pressure (mean ( SE, n ) 3).
Figure 6. Chromatograms of beagle plasma in vivo study of Lipo AS-013. HPLC/LIF conditions are the same as described for Figure 3, except as follows: chopper, 380 Hz; photomultiplier voltage, 640 V; isolation amplifier (band-pass filter: HPF 300 Hz, LPF 460 Hz).
of AS-013, 0.5 µg kg-1 min-1 for 45 min; infusion speed, 1 mL/ min). Blood samples, 5 mL each, were collected using heparinized syringes 5 min before infusion; 5, 10, 15, 30, and 45 min after start of infusion; and 5, 10, 30, and 60 min after completion of infusion. To suppress hydrolysis of AS-013 after blood collection, 100 µL of 0.05 M diisopropyl fluorophosphate, an esterase inhibitor, and 50 µL of 1% aspirin, a cyclooxygenase inhibitor which suppresses generation of endogenous PGE1, were added immediately after blood sampling. Immediately after addition of these agents, the tube containing the blood sample was placed upside down in an ice bath to mix the agent. Then the sample was centrifuged (10 000 rpm, 2 min, 4 °C) to obtain the plasma. 5010 Analytical Chemistry, Vol. 69, No. 24, December 15, 1997
The blood pressure in the beagle dogs was chronologically recorded throughout the administration experiment period. We used 1 mL of plasma for analysis. Chronological changes in chromatograms of the dog plasma are shown in Figure 6. Increases and decreases of PGE1 peaks were clear in the chromatograms. The plasma PGE1 concentrations and mean blood pressures in dogs are shown in Figure 7. AS-013 was promptly converted to PGE1 after the start of infusion, and the plasma PGE1 concentration increased rapidly. This concentration tended to increase gradually, beginning 15 min after the start of infusion. The plasma concentration at the end of infusion was ∼1000 pg/mL. The plasma PGE1 concentration gradually decreased after the end of infusion and 60 min later was ∼80 pg/ mL. PGE1 has both vasodilatory and hypotensive effects. When changes in blood pressure in beagle dogs were recorded during the study period, changes in the plasma PGE1 concentrations determined were found to be correlated with decrease in, and recovery of, blood pressure. In the current study, we used 1 mL samples of plasma for analysis. The absolute amount of PGE1 used for inducing ADAM derivatives was only 101-102 pg (subpicomole level). Moreover, derivatized PGs in amounts at the nanogram level were induced and were diluted for injection in HPLC/LIF. Therefore, it was easier to make the analysis more sensitive. This method is also expected to be useful for simultaneous analysis of other arachidonic acid metabolites. Our success in developing the analytical procedures demonstrates the usefulness of laser spectrometry to enhance conventional analytical procedures to allow ultramicroanalyses and ultratrace analyses without developing a new fluorogenic reagent. ACKNOWLEDGMENT Part of this work was financially supported by Seikagaku Co. (Tokyo, Japan). Received for review July 29, 1997. Accepted September 24, 1997.X AC970821V X
Abstract published in Advance ACS Abstracts, November 1, 1997.
