Centella asiatica and Butterfly Pea (Clitoria ternatea) in Oral Beauty Supplements and Functional Beverages: A Critical Review of Human Evidence, Bioavailability, Safety and Formulation

1. Introduction

Oral beauty supplements, or nutricosmetics, are among the fastest-growing segments of the supplement and functional beverage markets in Asia. Two botanicals feature prominently. Centella asiatica, long used topically for wound healing and skin care, is now sold as an oral ingredient for skin ageing. Butterfly pea (Clitoria ternatea) flower, a traditional food colour in Southeast Asia valued for its blue anthocyanins, is used in beverages marketed for beauty and wellness, often in combination with other botanicals [1,2].

Two recent papers address these ingredients. A randomised, double-blind, placebo-controlled trial in 112 Korean women reported that 200 mg/day of a standardised Centella extract for 12 weeks reduced periorbital wrinkle depth and improved hydration, barrier function, elasticity and brightness [3]. A review of butterfly pea described its phytochemistry and health-promoting activities and proposed microencapsulation to stabilise its anthocyanins for functional foods and nutraceuticals [2].

Neither paper addresses the questions a manufacturer or regulator must answer before making claims for a finished product: how strong the human evidence is, whether the active compounds reach the skin or circulation at relevant concentrations, what safety issues apply to chronic oral use, and what formulation choices preserve both the colour and the claimed activity. This review addresses those questions for both ingredients and proposes an evidence-based positioning for oral beauty products.

2. Methods

Europe PMC and Crossref were searched to October 2026 for human trials, pharmacokinetic studies, safety reports and regulatory assessments of oral C. asiatica and C. ternatea. The 2026 Centella trial [3] was appraised against the CONSORT 2010 statement [4] and the requirement for prospective trial registration [5], and its reported values were checked for internal consistency. The 2023 butterfly pea review [2] was used as the starting point for formulation studies, and its safety statements were checked against the original regulatory documents. This is a narrative critical review.

3. Centella asiatica

3.1. The 2026 Randomised Trial

The trial randomised 112 women aged 30 to 60 years with dry skin and visible wrinkles to 200 mg/day of a 70% ethanol extract standardised to 70 mg/g asiaticoside, or placebo, for 12 weeks [3]. The prespecified primary endpoint, change in average wrinkle depth measured by three-dimensional imaging, fell by 5.87 µm with the extract and 1.96 µm with placebo (p = 0.018) in the per-protocol population of 105 women. Mean depth of the biggest wrinkle, total wrinkle volume and surface roughness also improved more with the extract, whereas maximum wrinkle depth, wrinkle area and wrinkle length did not differ between groups. Cheek hydration rose by 2.7%, elasticity by 0.7% and brightness by 0.7%, changes the authors themselves described as modest. No serious adverse events occurred.

The trial is well designed in several respects: it was double-blind, used objective instruments, prespecified a primary endpoint and monitored dietary intake and safety laboratory values. Several features nevertheless limit the weight it can bear (Table 1). It was registered in February 2026, about nine months after the last participant finished, so the prespecification of the primary endpoint and analysis plan cannot be verified independently [5]. The main analysis was per protocol; an intention-to-treat analysis is described but not reported [4]. The methods state that wrinkles were measured on the cheek, whereas the results and figures describe periorbital wrinkles. The intention-to-treat ages of 56.45 and 52.59 years, with standard deviations of about 5, cannot produce the reported p-value of 0.207, and the per-protocol extract group, which differed by only three women, had a mean age of 51.51 years; at least one of these values must be wrong. The standard deviations for maximum wrinkle depth differ between text and table. The product was supplied by its manufacturer, which employed three of the six authors.

The size of the primary effect also needs context. The between-group difference of 3.9 µm on a baseline of about 53 to 56 µm is consistent with a modest effect, and the extract group started with shallower wrinkles than the placebo group (53.0 versus 56.1 µm). Whether a difference of this size is visible is not known, because no investigator or participant photographic assessment was reported as an outcome. The trial supports the hypothesis that this extract has a small effect on wrinkle depth; it does not establish visible anti-ageing benefits, and it cannot be extended to other Centella extracts, whose triterpene profiles differ.

3.2. Dose and Bioavailability

The daily dose in the trial supplied 14 mg of asiaticoside; other triterpenes were not quantified [3]. In healthy volunteers given 250 or 500 mg of a standardised extract rich in madecassoside and asiaticoside, plasma concentrations of these parent glycosides were very low, while their aglycones, madecassic and asiatic acids, reached higher levels and accumulated with repeated dosing [6]. Asiaticoside is converted to asiatic acid in vivo, and asiatic acid exposure increased with repeated dosing of a triterpene fraction at 30 or 60 mg twice daily [7]. The trial dose was well below those used in the pharmacokinetic studies, and plasma or skin concentrations were not measured. The trial authors attributed the effects to stimulation of fibroblast collagen synthesis, which has been shown in cell culture with the parent compounds. Whether the low systemic concentrations of metabolites after 14 mg of asiaticoside can stimulate dermal fibroblasts is unknown.

3.3. Safety

Short-term studies show good tolerability [3,6]. However, three women aged 49 to 61 years developed hepatitis with jaundice after 20 to 60 days of oral Centella, with granulomatous hepatitis on biopsy in two and recurrence on rechallenge in one [8]. These women were of the same age and sex as the target market for oral beauty products. A 12-week trial of 53 women cannot detect idiosyncratic liver injury at this frequency. Products should advise discontinuation if symptoms of liver injury occur, and post-marketing surveillance is warranted.

4. Butterfly Pea

4.1. Human Evidence

No human trial of butterfly pea with skin outcomes was identified (Table 2). The available trials are small and mostly acute. In 15 healthy men, 1 or 2 g of flower extract taken with 50 g of sucrose lowered post-meal glucose and insulin and raised plasma antioxidant capacity [9]. In 16 overweight or obese men, the extract did not affect post-meal glucose after a high-fat meal, while 2 g reduced the triglyceride response [10]. A 12-week comparative study in people with type 2 diabetes used 5 or 10 g daily and reported changes in oxidative and serotonin-related markers [11]. These studies support plausible metabolic effects of gram doses but provide no basis for beauty claims.

4.2. Bioavailability of Ternatins

The blue colour of butterfly pea comes from ternatins, polyacylated delphinidin glycosides that are more stable than common anthocyanins [2]. In rats, ten ternatins were absorbed intact and detected in plasma within 15 minutes of oral dosing [12]. No human pharmacokinetic data exist. In humans, a labelled common anthocyanin was largely degraded to phenolic acids and other metabolites, which reached peak serum concentrations of 10 to 2000 nM [15]. Whatever activity butterfly pea has in people is therefore likely to depend on metabolites rather than the intact pigments measured in food studies.

4.3. Safety and Regulatory Status

The 2023 review noted that butterfly pea flowers have a long history of food use in Thailand and are accepted as food ingredients in some Asian jurisdictions, but that EFSA had raised concerns [2]. Those concerns are specific. In 2022, EFSA raised a reasoned safety objection to placing dried butterfly pea flowers on the European Union market as a traditional food, because some of the cyclotides present have haemolytic and cytotoxic activity in vitro and the data on cyclotide intake and toxicity were limited [14]. Cyclotides are stable cyclic peptides abundant in butterfly pea [16]. In the United States, an aqueous butterfly pea flower extract, ultrafiltered and pasteurised, is permitted as a colour additive in beverages and foods, with specifications for pH and heavy metals [13]. That authorisation covers colouring use at amounts consistent with good manufacturing practice, not the gram doses used in health trials. Products should therefore distinguish colouring use from intake as a functional ingredient, and the cyclotide content of extracts used at functional doses should be characterised.

4.4. What Microencapsulation Achieves

The 2023 review summarised studies that encapsulated butterfly pea extract in maltodextrin, gum arabic, gelatin, carrageenan, cyclodextrin or alginate by spray-drying, freeze-drying, oven-drying, microwave-drying or air-drying [2]. The outcomes were anthocyanin retention, encapsulation efficiency, colour, water activity, radical-scavenging antioxidant activity and, in some studies, antimicrobial activity. The review states that microencapsulation improves the bioavailability of delivered compounds, but none of the studies it summarised measured release in digestive conditions, absorption, cyclotide content or any human outcome. Microencapsulation is therefore established as a way to stabilise the colour and anthocyanin content of powders, not as a way to enhance biological effects. Retention of anthocyanins and of antioxidant activity also depended on the wall material and drying method [2], so results from one encapsulation process cannot be assumed for another.

5. Formulating Combination Products

Combining Centella and butterfly pea in a single beverage or powder raises practical issues. Butterfly pea colour depends on pH, appearing blue in near-neutral conditions and shifting towards purple and red as acidity increases; the United States specification sets the colour additive itself at pH 3.0 to 4.5 [13]. A colour-change effect in acidified drinks is attractive commercially, but it also signals structural change in the pigment. Centella triterpene glycosides are bitter, and their stability in beverages has not been established in the trial or review literature. Because the only positive Centella trial used a specific extract in tablet form, at 200 mg/day after a meal [3], a beverage that uses a different extract, dose or matrix cannot rely on that trial. Each finished product should be tested for content uniformity and stability of the defining markers, namely asiaticoside and other quantified triterpenes for Centella and total ternatins for butterfly pea, over its shelf life.

6. Implications for Claims and Research

Table 3 summarises an evidence-based positioning. Centella extract can be presented as an ingredient investigated for skin ageing in one placebo-controlled trial, with claims restricted to the extract and dose tested and to the small magnitude of the effects. Butterfly pea can be used as a natural colour within regulatory specifications, but skin and anti-ageing claims are not supported. Claims should not be transferred from separate ingredient studies to combination products.

The research priorities are clear. For Centella, an independent, prospectively registered trial with intention-to-treat analysis, quantified madecassoside, asiaticoside, madecassic acid and asiatic acid in the product, plasma metabolite measurements and a photographic or clinician-rated visible outcome would establish whether the effect is real and noticeable. For butterfly pea, characterisation of cyclotide content in commercial extracts, human pharmacokinetics of ternatins and a controlled trial with skin endpoints are needed before beauty claims are made. For both, stability studies in the final beverage matrix should accompany any efficacy claim.

6.1. Limitations

This is a narrative review focused on two papers and the evidence they rely on. The appraisal of the Centella trial is based on its published report and may be resolved by data not presented. Some regulatory information was obtained from secondary summaries of the EFSA report. The author’s company manufactures botanical extracts and functional beverages, including beauty products.

7. Conclusions

Oral Centella asiatica has one placebo-controlled trial suggesting a small reduction in wrinkle depth with a specific extract, but the trial was registered retrospectively, analysed per protocol, run by the manufacturer and contains reporting inconsistencies; parent triterpenes are poorly absorbed, and idiosyncratic hepatitis has been reported. Butterfly pea has small acute trials on glucose, lipids and antioxidant status, no skin evidence, permission as a colour additive in the United States and an unresolved safety objection over cyclotides in the European Union. Microencapsulation stabilises its colour but has not been shown to improve bioavailability. Products combining the two should claim no more than the evidence for each ingredient, at the dose and in the form tested, supports.