Table I Characteristics of th e Included Studies Author Gender (%)age, Mean y Number of samples (intervention: control) Area Type of control Details of treatment Details of outcome Treatment regimen Treatment detail Duration, w Outcome target Measurement tool Point of measure (at w) Ryu et al. (19) Female (100) 23.3 20–50 (25:25) Lip Placebo 0.2% asiaticoside lipstick Apply twice daily 8 Wrinkle color Five-point visual scoring scale Image analysis 0,4,8 Lee et al. (18) Female (100) 46.6 27 (27:27) Periorbital Vehicle 0.1% asiaticoside cream Apply twice daily 12 Wrinkle Visiometer SV 600 0,4,8,12 Tangsumroengwong (20) Female (100) 32 35 (35:35) Periorbital 0.02% tretinoin gel 5% CAa extract gel 0.25 g at bedtime 12 Wrinkle Visioscan VC98 0,4,8,12 RG5VSS Moisture Corneometer Satisfaction AE Tongsrikeaw (21) Female (100) 41.4 25 (25:25) Peri-orbital 5% Pueraria mirifi ca gel 5% CAa extract gel Apply twice daily 12 Wrinkle Visioscan VC98 0,4,8,12 Satisfaction AE RG5VSS Zhang et al. (22) Female (100) 19–71 35 Lateral angleVehicle Centella triterpenes cream® Applied 3x daily 12 Periorbital wrinkling Visioscan VC98 and dermatologist scoring 0,4,8,12 CA: C. asiaticaFTU: , fi ngertip unit (5 mm. of cream from the tip of the index fi nger to the fi rst top joint of the index fi nger). This study defi ned 1 FTU = 0.5 g. of cream. a CA extract was ordered for Chang Kaew Natural Company, Nontaburi, Thailand, containing deionised water 88.25%, allantoin 0.3, g 1.0, propylene glycol 2.0, carbopol 940 0.4, triethanolamine 0.4, aloe vera extract 1.0, cucumber extract 1.0, Centella extract 5.0, Glydant plus 0.2, Emulsogen HCO 040 0.45, shower fragrance, and green coloring. No description of how Centella extract was produced or standardized. EFFICACY AND SAFETY OF CENTELLA ASIATICA 443

JOURNAL OF COSMETIC SCIENCE 444 Table II Quality Assessment of Included Studies Using the Cochrane Risk of bias v. 2.0 tool Author Bias from randomization Bias due to deviations from intended interventions Bias due to missing outcome data Bias in measurement of the outcome Bias in selection of the reported result Overall bias Ryu et al. (19) Some concern Low Low Low Low Some concern Lee et al. (18) Low Low Low Low Low Low Tangsumroengwong (20) Low Low Low Low Low Low Tongsrikeaw (21) Low Low Low Low Low Low Zhang et al. (22) Some concern Low Low Low Low Some concern Lee et al. (18) provided detailed numerical “roughness” data, but relation to wrinkle pa- rameters was unclear. They measured six skin parameters determined by dermatologists which were bundled along with participant self-assessment into a single overall score. Some parameters were described as statistically signifi cant, but effect sizes were not re- ported. In general, the quality and depth of reporting were poor. INTERVENTIONS Treatments were randomly allocated to either test or placebo/comparator. Three studies applied treatments twice daily, once a day, or thrice daily (Table I). Measurements were made at 4-w intervals beginning at baseline and continuing to 12 w, except Ryu et al. (19) who stopped at 8 w. Two studies (18,19) used 0.1 or 0.2% asiaticoside in their treat- ments. C. asiatica extract was the intervention in the other three studies but did not identify the plant source, strain, growing condition, detail the extraction method nor any standardization for likely active ingredients. All the studies were double blinded and controlled using a comparator similarly formulated to the intervention. Comparators were 0.02% tretinoin (20) or Pueraria mirifi ca extract (21) or placebo. MOISTURISING SKIN This was measured as capacitive impedance of ~20 μm of the corneum read as resonant frequency. C. asiatica gel increase mean moisture value of periorbital wrinkled skin by 26% between baseline and w 12 compared 1% with tretinoin and 12% and -1% for lateral canthus (Table III) (20). L IP WRINKLES A lipstick containing asiaticoside was applied to the lips for 8 w after which visual grad- ing and image analysis of silicon casts showed reduced wrinkling by ~20%. As the depth