202 JOURNAL OF COSMETIC SCIENCE Table $ Clinical study on skin whitening by chromometer Products Change from no treatment (%) Kojic acid I Month p 2 Months p 3 Months p value value value Aqueous base 0.00 0.95 0.9 0.029 0,2 0.73 Anhydrous 0.80 0.09 1.6 0.006 2.6 .0003 base Hydroquinine 1.10 0.05 0.2 0.61 2.8 0.019 product Bold: Significant p values. Patent pending. * Y. L. worked at Mary Kay Holding Co. A SENSITIVE AND QUANTITATIVE SPLIT-FACE CLINICAL PROTOCOL TO DISCRIMINATE COSMETIC FORMULATIONS FOR FACIAL HYPERIGMENTATION REDUCTION EFFICACY Kukizo Miyamoto and Greg G. Hillebrand INTRODUCTION: Procter & Gamble Far East, Inc., Kobe, Japan Facial skin hyperpigmentation, e.g., senile lentigines, freckles, post-inflammatory hyperpigmentation, and melasma, is a problem for all peoples of the world but is particularly problematic for Asiansl, 2. Sound clinical data that unequivocally proves the visible skin benefits afforded by skin lightening and/or hyperpigmented spot reduction formulations is frequently needed for claims support and regulatory approval. Split-face paired comparison testing has proven to be quite useful in dermatology for treatment efficacy assessment for various skin disorders, especially acne 3. In this paper, we present results that demonstrate the sensitivity and efficiency of a split-face clinical design combined with high resolution video imaging to discriminate cosmetic formulations for hyperpigmentation reduction efficacy. CLINICAL METHODS: Design: Randomized split-face double blind paired comparison study. Study Period: February-August, ! 997. Subiects: 120 Japanese females, 25-60 years, with moderate to severe hyperpigmentation on both sides of the face. Test Treatments and Regimen: Subjects (60 per group) were instructed to apply a test formulation (0.26g) to one side of their face and to apply another formulation to the other side twice a day (morning/evening) for 6 months. Test treatment pairs were randomized right/left and were: One Side of Face Other Side of Face Leg I SLM = A commercially marketed Skin Lightening NM = A commercially marketed Norma• Moisturizer containing a mixture of MHW-approved skin vs. Moisturizer without skin lightening active• lightening actives (Max Factor Natique CR White) (Oil of Olay Sensitive Skin Beauty Fluid) 1 Leg 2 E = A commercially marketed Essence containing a vs. V = SK-II Facial Treatment Repair C Vehicl• yeast-conditioned culture medium ,(SK-II Facial (yeast-conditioned medium is replaced b 1 Treatment Repair C) : water) Skin Measures: Video images of the left and right sides of the subject's clean face were collected at baseline, 1, 3 and 6 months of treatment using a Sony DXC-537H 3CCD color TV camera with a Canon J I5x9.5 BKRS lens equipped with a polarizing filter. Facial illumination was provided by two Balcar Flux-Lites equipped with flux tungsten light bulbs (3500øK) positioned above and below the camera to provide even lighting of the side of the subject's face. Accurate repositioning of the subjects was facilitated by superimposing the live image on the digitally-stored image obtained at baseline. A color chart was used as a standard to calibrate the imaging system each study day. Computer analysis of the video images allowed quantification of basal skin color tone (L*a*b*) and hyperpigmcntcd spot area (mm 2) of pigmented maculcs. Reproducibility and Accuracy: The intra-subjcct coefficient of variance (COV) of L*, a*, and b* values and hyperpigmented spot area for five video images of the same subject (images spaced 10 min. apart) were determined over five subjects. Accuracy was determined using Asian and Caucasian mannequin heads (Figure I) to which circular artificial spots of known area were applied the computer image-analyzed area was compared to the true area. Visual Perception System (VPS): Judges (n=7) viewed each subject's pre-trcatment (baseline) rs. post-treatment (1, 3 or 6 month) facial images side by side on a calibrated video monitor (Barco Type 121). The judge was blind to which image was predpost treatment and the identity of the test treatment. The judges were trained to recognize melanin

PREPRINTS OF THE 1998 ANNUAL SCIENTIFIC SEMINAR 203 hyperpigmentation and to make judgments based on pigmentation differences in the upper cheek and eye area only. The judge indicated which of the two images had less noticeable hyperpigmented spots and rated the magnitude of the difference between images on a 1-4 scale. The magnitude rating was assigned a positive (+) or negative (-) sign depending upon whether (+) or not (-) the judge indicated the post-treatment image had less noticeable spots. Thus, positive values of the magnitude rating indicate spot reduction efficacy and values of zero or less indicate no efficacy. Statistical Analysis: The difference between the test treatments' (SLM and E) mean change from baseline and their respective controls' (NM and V) mean change from baseline for L*, a*, and b* values and hyperpigmented spot area were compared using a paired T-test (significance set at p0.05). The mean VPS grades for each treatment and control pair were also compared using a paired T-test. A non-parametric statistic analysis (McNemar's test) was used to compare the test products for the number of 'AFTER' images selected as improved in the VPS analysis. RESULTS: Reproducibility and Accuracy: The within subject COVs for L*, a*, and b* value measurements in the cheek and eye area were 0.6%, 2.1% and 1.8•/• respectively. The within subject COV for the hype•igmented spot area measurement in the cheek and eye area was 0.5%. The imaging system accuracy for measuring changes in hyperpigmented spot area was determined to be less than +/-5% when measurements were made in the region of the cheek and eye area. All image analysis measurements of live subjects were confined to the area of the cheek and around the eye. Subiect Accountability: Of the 120 subjects enrolled, 104 completed the study (118 completed through 3 months). Basal Skin Color Tone: There was no significant difference between treatments for L*, a* or b* value change from baseline (see Table I for L*-valuc change from baseline). Hyperpigmented Spot Area: Both test treatments (SLM and E) showed significant reduction in hypcrpigmented spot area vs. their respective controls at all time points (Figure 2). Visual Perception System {VPS}: The mean VPS grades for the test treatments (SLM & E) were significantly greater than their respective controls at all time points (Figure 3). Judges more often selected the AFTER image (vs. the BEFORE image) for both SLM & E vs. their respective controls (Table II). Figta'e 4 shows hyperpigmentation improvement on one subject treated with E on one side of her face at baseline and 6 months (image overlays are also shown). FIGURE I TABLE I TABLE II Asian Shaped Mannequin Heads L*-value Change From Baseline VPS % AFTER Images Selected Months Months Treatment I 3_ 6 SLM -0.27 0.78 0.04 NM -0.26 0.96 -0.20 E -0.27 0.61 0.22 V -0.19 0.55 0.07 Treatment ! 3_ f SLM 74* 60* 77* NM 45 37 56 E 60* 43* 62* V 40 30 43 *Sig. diff. from control (p0.025), McNemar's Test FIGURE 2 FIGURE 3 FIGURE 4 Hyperpigmented Spot Area VPS Mean Grades Subject Treated With Product E Baseline image (left) with overlay (right) SIg. diff. frown NM (pO.001) ' • ]T•g. diff. frown NI4 (pO.001) i Monks of T•a•ent Monks of T•a•ent 6-mon• image (le•) with overlay (right) Monks of Tm•ent Mon• of T•a•ent