Analytical Communications, January 1996, VoE33 (1 5-1 7) 15 Comparison of Supercritical, Subcritical, Hot, Pressurized and Cold Solvent Extraction of Four Drugs From Rodent Food John R. Williams", E. David Morgan"* and Brian Lawb a Department of Chemistry, Keele University, Keele, StafSordshire, UK ST5 5BG Macclesfeld, Cheshire, UK SKI 0 4TG Toxicology Resources Group, Zeneca Pharmaceuticals, Alderley Park, Propranolol, tamoxifen and two experimental drugs (all basic but of differing pK, and polarity) were extracted from rodent food by supercritical carbon dioxide-methanol more efficiently than by a more polar, subcritical mixture of carbon dioxide and methanol.Hot, pressurized methanol alone was as efficient as the supercritical fluid. The efficiency of these methods was similar to conventional solvent extraction with acidified methanol, but conventional solvent extraction gave the best precision, as measured by the relative standard deviation.In recent times, interest has grown in alternatives to conven- tional solvent extraction for the isolation and concentration of organic substances from a variety of matrices. The drive to find alternatives to solvent extraction are several-fold and include the wish to; avoid using and disposing of large volumes of flammable or chlorinated solvents; reduce manipulation (which is costly in manpower and a potential source of errors); and achieve shorter extraction times.Supercritical fluid extraction (SFE)1-5 has come to be the most prominent alternative to solvent extraction, but other methods such as microwave- assisted extractionCl0 and accelerated solvent extraction' are being offered as alternatives.To date, only one direct com- parison has been made between these modem approaches to extraction in which Soxhlet, microwave and SFE were compared for the recovery of polycyclic aromatic hydrocarbons from highly contaminated soils.I2 The highest total amount of polycyclic aromatic hydrocarbons was extracted using SFE, followed by microwave-assisted extraction, and then Soxhlet extraction; but the conditions and solvent were not optimized for the Soxhlet method.Methods of the extraction of pharmaceutical compounds from medicated rodent diet is of interest. Polar drugs can be difficult to extract efficiently using supercritical carbon dioxide, and in normal circumstances, it is necessary to add a polar solvent (e.g., methanol, ether, acetonitrile) to the carbon dioxide to increase its polarity and solvent properties. It is, nevertheless, logical to ask if this is the most efficient extraction medium.What if the fluid is made more polar by adding a greater proportion of the conventional solvent, and the mixture drops below its critical point? Will that be more efficient for the purpose, and how would these compare with simply using a liquid such as methanol either at room temperature and pressure or at high temperature and pressure? Experiments that attempt to answer these questions are described for a group of four pharmaceutical compounds (Fig. 1) that have been incorporated into rodent diet.* To whom correspondence should be addressed. Experimental Apparatus and Reagents SFE, subcritical fluid extraction (SubFE) and hot, pressurized solvent extraction were carried out with a manual, single- sample extractor. The extractor consisted of a Philips Analytical (Cambridge, UK) LC-XP3 pump to deliver liquid carbon dioxide, a Waters (Milford, MA, USA) M-45 pump to contribute methanol, a Philips Analytical Series 104 Chromato- graph oven to regulate temperature and a stainless-steel extraction cell (3 cm X 4.6 mm id) made from high- performance liquid chromatography (HPLC) column parts to contain the sample.Pressure in the system was maintained with a silica capillary (25 cm X 50 pm id; Composite Metal Services, Worcester, UK). The HPLC consisted of a Philips Analytical LC-XP3 pump to deliver the mobile phase, a Rheodyne (Cotati, CA, USA) 7125 injection valve, a Philips Analytical LC/UV detector and a Shimadzu (Kyoto, Japan) C-R3A integrator.Propranolol (as the hydrochloride), tamoxifen (as the base), ZM 95527, ZM 169369 (two candidate drugs without generic names) and the rodent food were obtained from Zeneca Pharmaceuticals (Macclesfield, UK) and used without further purification. The carbon dioxide was industrial grade (98.5%) from BOC (Guildford, UK).The methanol was HPLC grade from Prolabo (Paris, France) and trifluoroacetic acid (99%) was from Avocado (Lancaster, UK). Ammonium formate (analyt- ical-reagent grade) was purchased from Fisons (Loughborough, UK) . General Procedure Sample Preparation The drugs and rodent food were blended together at concen- trations of 5 mg g-1 with a Turbula mixer (Stanmore, UK) for I A I Ropranolol " - O q Tamoxifen ZM 169369 Fig.1 Structures of the four basic compounds used in this study.16 Analytical Communications, January 1996, Vol33 10 min. The resultant mixture was further ground with a pestle and mortar to aid sample uniformity.Supercritical Fluid Extraction Carbon dioxide-methanol (85 + 15 v/v or 83 + 17 mol%) was pumped (2 cm3 min-l) through the extractor at 70 "C and 17.25 MPa. The sample size was 0.3 g. The extracted drugs were depressurized into a glass tube (15 cm X 2.5 cm) containing 5.0 cm3 of methanol. When extraction was complete, the collector tube was emptied and rinsed with methanol and the total volume of liquid adjusted to 25.0 cm3 (with methanol) ready for analysis.Subcritical Fluid Extraction The same procedure and conditions were used as for SFE, except that the extracting fluid was carbon dioxide-methanol (15 + 85 v/v) and flow rates of 1 cm3 min-l and 2 cm3 min-1 were used. Solvent Extraction Hot, pressurized solvent extraction A similar procedure to that used for SFE was followed, except that methanol was used at 1 cm3 min-' at 55 "C and 100 "C, and a flow rate of 2 cm3 min- was used at 70 "C.No methanol was necessary in the glass collector tube at the start of the extraction. Cold solvent extraction For cold (unpressurized) solvent extraction two methods were used, method A was a similar procedure to SFE but the temperature was at 20 "C and no silica capillary or collection solvent was used.For the second, method B, the drugs-food sample (0.3 g) was placed in a sample tube [poly(propylene) 15 cm X 1.5 cm] and the extraction solvent added [5.0 cm3 of methanol or methanol-aqueous 1 mol dm-3 trifluoroacetic acid (96.2 + 3.8 v/v)]. Sample and solvent were then thoroughly mixed by vortexing for 1.5 min, followed by the separation of the solvent from insoluble matter by centrifugation (5 min at 1500 rpm) before analysis.HPLC Quantitation was by isocratic HPLC using a Spherisorb 5 SCX strong cation exchange column (10 cm X 4.6 mm id; Phase Separations, Deeside, UK) and a mobile phase of 0.02 mol dm-3 ammonium formate in methanol-water (80 + 20 v/v) at pH 2.45 (adjusted by addition of trifluoroacetic acid). The flow rate was 1 cm3 min-l and the ultraviolet detector was set at 270 nm.Results and Discussion Propranolol, tamoxifen, ZM 95527 and ZM 169369 represent a range of basicity (as measured by pK,) and polarity (as measured by log P) and these values are given in Table 1. Tamoxifen is the least polar drug. ZM 95527 is the most polar drug and is also the least basic of the drugs.Propranolol is the most basic of the four drugs. There appeared to be a correlation between recovery of the drugs from rodent food and drug basicity when methanol (with and without aqueous tri- fluoroacetic acid) at 20°C was used in three methods as the extraction medium. Within this group, recovery appeared to increase with increasing basicity.However, the correlation did not extend to other similar basic drugs (unpublished results). For the other extraction methods, there was no correlation between either recovery and basicity or recovery and polarity (see Table 1). Some of the extractions were carried out for 3 h. The plots of percentage recovered versus time for propranolol show an initial rapid rise in recovery, followed by a slowing down in the extraction rate leading to a plateau (Fig.2). Similar curves were obtained for the other three drugs. From the results for recovery after 10 min (Table l), a good impression of the relative efficiency of recovery after 3 h is achieved. Recoveries by the different extraction methods were com- pared using the paired t-test13 with a confidence interval of 95%.Hot, pressurized methanol at 2 cm3 min-l and 70 "C was more efficient than SFE (with methanol-modified carbon dioxide) for the recovery of ZM 95527, but not significantly different for the recovery of propranolol and ZM 169369. It was less efficient than SFE for the recovery of tamoxifen. Although one might have expected a subcritical mixture of methanol and carbon dioxide at 2 cm3 min-I and 70 "C to be intermediate between SFE and hot, pressurized methanol, it was, in fact, slightly less efficient.The subcritical method was also relatively insensitive to flow rate; recoveries were similar at 1 and 2 cm3 min-1. Precision, as indicated by the relative standard deviation (Table l), was extremely good for recovery by a routine methanol-aqueous acid procedure (not optimized for this group of compounds).Furthermore, precision was also good for the Table 1 Comparison of SFE and SubFE with the use of hot, pressurized and cold solvent extraction of propranolol, tamoxifen, ZM 95527 and ZM 169369 from rodent food. The results obtained after extraction for 10 rnin are means of six determinations and s, values (%) are shown in brackets. Methanol was pumped through the sample in method A and vortex-mixed in method B.Recovery (%) Extraction fluid Supercritical Subcritical Subcritical Hot, press. MeOH Hot, press. MeOH Hot, press. MeOH Method A MeOH Method B MeOH Method B aq. TFA C02-MeOH (85 + 15) C02-MeOH (1 5 + 85) C02-MeOH (15 + 85) Flow1 cm3 min-1 2 2 1 2 1 1 2 - - Propranolol Temperature1 pK,, 9.42; "C log P , 3.56 70 88.4 (12.9) 70 78.2 (15.2) 70 73.2 (25.5) 70 91.0 (8.1) 55 38.1 (35.8) 100 43.2 (37.0) 20 92.1 (1.8) 20 88.8 (3.9) 20 95.7 (1.4) Tamoxifen pK,, 8.57; log P , 6.63 86.2 (8.1) 70.8 (15.5) 67.1 (24.2) 82.3 (7.8) 35.9 (34.8) 40.4 (39.7) 80.3 (0.7) 77.7 (3.0) 84.5 (0.8) ZM 95527 pK,, 7.93; log P , 1.07 84.5 (12.1) 78.6 (16.7) 69.4 (26.8) 92.1 (9.0) 37.9 (30.1) 36.0 (30.5) 59.6 (3.8) 54.6 (6.9) 72.2 (1.0) ZM 169369 pK,, 8.23; log P , 4.50 81.2 (1 2.2) 68.1 (14.5) 65.8 (24.5) 78.9 (7.6) 37.7 (32.8) 42.1 (41.3) 74.4 ( 1.4) 69.6 (3.8) 76.5 (2.7)Analytical Communications, January 1996, Vol33 17 other two methanol-based solvent extraction methods con- ducted at 20°C.The precision was poor for supercritical and subcritical conditions and hot, pressurized solvent extraction. However, for these latter methods, the precision seemed higher the more efficient the recovery.Hot, pressurized solvent extraction at a low flow rate gave poor extraction and very poor precision. As expected, recovery of all four drugs from rodent food was significantly better when following method A than method B (methanol) and precision was also better (see Table 1).Aqueous acidified methanol produced significantly better recovery of all four drugs than methanol alone. Method A is directly comparable (same extractor, flow rate and extraction time) with the methods of SFE, SubFE and hot, pressurized methanol. Recovery of the drugs by SFE was significantly better than by cold, unpressurized methanol (except for propranolol where the reverse was true) but precision with SFE was poorer (Table 1).In contrast, recovery of the drugs by cold, unpressurized methanol was significantly better than by SubFE (except for recovery of ZM 95527 where the reverse was true) but again precision was superior for the cold, unpressurized solvent. Only the recovery of ZM 95527 and ZM 169369 were significantly better for hot, pressurized methanol than cold, unpressurized methanol; there was no significant difference between the two methods for recovery of propranolol and tamoxifen.In general, precision with cold, unpressurized methanol was better than with hot, pressurized solvent. Conclusions Recovering basic drugs from cereal-based food is not as simple as recovering a compound from an inert matrix.The variable '"1 I 0 100 200 Ti rn e/m i n Fig. 2 Plot of propranolol recovery versus time for several extraction fluids. 1, MeOH, 55 OC, 1 ml min-1; 2, MeOH, 100 "C, 1 ml min-I; 3, C02-MeOH (15 + 85), 1 mi min-1; 4, C02-MeOH (15 + 85), 2 ml min-1; 5, C02-SMeOH (85 + 15), 2 ml min-1; 6, MeOH, 70 OC, 2 ml min-1; and 7, MeOH, 20 OC, 2 ml min-1. results suggest there is some intereaction between the com- pounds and the food.Conditions may be quite different in the extraction of less polar compounds, e.g., extraction of poly- cyclic aromatic hydrocarbons 14-16 or polychlorinated bi- phenyls16.17 from soil or sediment. Newer methods of extraction deserve careful testing and comparison. However, in this case and under these conditions, they appear to offer no clear advantage.For two of these compounds, the supercritical fluid method offered the highest extraction and it was second highest for another. Hot, pressurized methanol may appear hazardous, but the solvent was cold when it emerged from the silica capillary after depressurizing. J.R.W. thanks The Royal Society of Chemistry for the award of a SAC industrial studentship. References 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 Analytical Supercritical Fluid Chromatography and Extraction, ed.Lee, M. L., and Markides, K. E., Chromatography Conferences, Provo, Utah, 1990. Supercritical Fluid Extraction and Its Use in Chromatographic Sample Preparation, ed. Westwood, S. A., Blackie, London, 1992. Analysis With Supercritical Fluids: Extraction and Chromatography, ed.Wenclawiak, B., Springer-Verlag, London, 1992. Hyphenated Techniques in Supercritical Fluid Chromatography and Extraction, ed. Jinno, K., Elsevier, Amsterdam, 1992. Luque de Castro, M. D., Valcircel, M., and Tena, M. T., Analytical Supercritical Fluid Extraction, Springer-Verlag, London, 1994. Ganzler, K., Salg6, A., and Valko, K., J . Chromatogr., 1986, 371, 299. Neilson, R. C., J . Liq. Chromatogr., 1991, 14, 503. Onuska, F. I., and Terry, K. A., Chromatographia, 1993,36, 191. Lopez-Avila, V., Young, R., and Beckert, W. F., Anal. Chenz., 1994, 66, 1097. Barnabas, I. J., Dean, J. R., FowIis, I. A., and Owen, S. P., Analyst, 1995, 120, 1897. Richter, B. E., Ezzell, J. L., Felix, D., Roberts, K. A., and Later, D. W., Int. Lab., 1995, May, 18. Dean, J. R., Barnabas, I. J., and Fowlis, I. A., Anal. Proc., 1995,32, 305. Miller, J. C., and Miller, J. N., Statistics for Analytical Chemistry, Ellis Horwood, Chichester, 1989, ch. 3, pp. 58-59. Lee, H.-B., Peart, T. E., Hong-You, R. L., and Gere, D. R., J . Chromatogr. A , 1993, 653, 83. Reimer, G., and Suarez, A., J . Chromatogr. A , 1995, 699, 253. Langenfeld, J. J., Hawthorne, S. B., Miller, D. J., and Pawliszyn, J., Anal. Chem., 1993, 65, 338. Bowadt, S., Johannsson, B.. Wunderli, S., Zennegg, M., de Alencastro, L. F., and Grandjean, D., Anal. Chem., 1995, 67, 2424. Paper 5l05804J Received September I , I995 Accepted November 10,1995