326 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS chromatographic assay on creams and gels calculating the analyte amount in compari- son with a standard preparation. In this paper a procedure is described for routine control of skin-care cosmetics, like creams, gels, lotions, and ointments, able to reveal the possible presence of TRT in very low concentration. Due to the interferences of the complex cosmetic formulations, a treatment was required to simplify the matrix and to enrich the TRT concentration level in the sample. For this purpose, the cosmetic, dispersed in tetrahydrofuran, was treated with a carbon column. After removing excipients among the carbon granules and washing with repeated amounts of chloroform, pyridine was finally used as analyte eluent. TRT determination was performed on the pyridinic solution by UV-derivative spectrophotometry. In the case of cosmetics containing UV filters highly overlapping the UV curve of TRT, a full fractionation with an alternative method was needed. Therefore, a further procedure by HPTLC densitometry was developed, performing the acidification of the pyridinic eluate with hydrochloric acid and the analyte extraction with methylene chloride. It is plain that, in any case, when the UV signals are not clearly ascribed to TRT, due to whatever interference, the analyte presence can be assured by the HPTLC method. Sev- eral attempts to apply an HPTLC procedure without a prior fractionation and analyte enrichment of the samples proved to be unsuccessful. The derivative spectrophotometric method was also suitable for TRT assay in pharma- ceutics, directly on a tetrahydrofuran suspension of the samples without any removal of the other components. EXPERIMENTAL MATERIALS Tretinoin was purchased from Aldrich Chemicals (USA) HPTLC plates (Kieselgel 60, 10 ) 10 cm) from Merck (Germany) and nylon membrane filters, 0.45-1•m-pore size, from Whatman (England). All solvents, supplied by C. Erba (Italy), were of analytical grade. Active carbon types studied are reported in Table I. Ointment and cream bases, used for recovery studies, were marketed by Resriva (Italy) and Schering-Plough (Italy). Table I Carbon Types Investigated Proprietary name Particle size Supplying firms Granular 1.5 mm Merck (Germany) Granular 2.5 mm Merck Darco 4-12 mesh Aldrich (USA) Darco 12-20 mesh Aldrich Darco 20-40 mesh Aldrich Norit RB1 0.6 pellets Aldrich Norit R0 0.8 pellets Aldrich

TRETINOIN ASSAY BY CARBON PHASE EXTRACTION 327 Pharmaceutical forms assayed were: Retin-A cream 0.010%, 0.025%, 0.05%, gel 0.025%, and lotion 0.05 mg/ml (Cilag, Switzerland) Airol cream 0.05% and lotion 0.05% (Roche, Switzerland) and Apsor ointment 0.1% (IDI Farm., Italy). The following commonly used cream and ointment excipients were used: 3-butyl-4- hydroxyanisole, diethylene glycol monoethyl ether, dimethyl polysiloxane, glyceryl be- henate, glyceril monostearate, glycerol, isopropyl myristate, lanolin, lanolin isopropyl esters, oleic acid esters, paraffin, perfluoropoliether, polyoxyethylen 40 stearate, perhy- drosqualene, propylene glycol, saturated fatty acids, triglycerides, saturated polygly- colyzed glycerides, sorbitol 70%, xanthan gum, and white wax. Several formulations with varying concentrations of these substances were prepared with the addition of known amounts of TRT. The following products, variously employed in commercial anti-aging creams, were added to the base creams to study the potential interference: allantoin, p-aminobenzoic acid, benzoic acid, camphor, N-dimethyaminobenzoic acid, retinol, dl-o•-tocopherol, salicylic acid, salicylic esters, sorbic acid, and stearic acid. The UV filters investigated were 2,2'-dihydroxhy-4,4'-dimethoxy-benzophenone, 2- ethylhexyl-p-methoxycinnamate, 1-(4-methoxyphenyl)-3-tert-buylphenyl) propan- 1,3- dione, 3-(4-methylbenzylidene)-camphor, and octyl-dimethyl p-aminobenzoic acid. APPARATUS Spectrophotometry. Spectra were recorded over the wavelength range of 450-250 nm in 10-mm silica quartz cells using a Perkin-Elmer Lambda 16 spectrophotometer scan speed 2 nm/s response (time constant) 1 s for zero-order and 5 s for second- and third- order derivative spectra spectral bandwidth 1 nm A)• 8 for both derivative orders. The spectra were elaborated with PECSS 4.0 software by Perkin Elmer. Densitometry. Measurements were performed with a Shimadzu (USA) model CS930. Ex- perimental conditions were: absorption reading mode at 370 nm scanning speed (lin- ear) 1 mm/s recorder baseline 200 mV beam size 0.4 X 0.4 mm. The development was performed in a linear development chamber (Camag, Switzerland). Carbon columns. Glass tubes of 30 X 0.5 cm (i.d.) were used, with a capillary end of 0.05 cm (i.d.) to optimize the solvent flow. Three to five grams of carbon were washed, refluxing in Soxhelet with 200 ml of chloroform for 12 hours, then vacuum- dried at room temperature until a constant weight. One gram of dry carbon, accu- rately weighed, was transferred and gently packed in one column, obtaining a granu- lar phase height of 14 __+ 0.5 cm. For this column the intra- and inter-particles volumes proved to be 0.2 and 0.8 ml with THF, and 0.7 and 1.5 ml with pyridine, respectively. LABORATORY PRECAUTIONS All the assay procedures were carried out in a dark room provided with a red lamp of 60 W kept at a distance of 2 meters to avoid photodegradation of the retinoic acid (11,12). The carbon columns were protected with tin foil.