The Science

In building a case for a cosmetic ingredient, basic science and clinical evidence do different jobs. Basic science establishes plausibility. In the language of a criminal case, it establishes motive and opportunity — a reason the effect should follow, and a route by which the ingredient can reach the tissue where it would have to act. Motive and opportunity do not convict on their own. They establish that the charge is worth taking seriously, and what evidence would settle it. Clinical evidence establishes what actually happened. Without motive and opportunity, the rest of the evidence has to carry the whole case by itself, and be unimpeachable to do it. With them, a result is easier to credit — it fits something that already makes sense. That is what basic science contributes: not proof, but the common sense that makes the rest believable.

This page sets out the basic science case for Kaempferia parviflora root extract, in three parts.

Skin structure depends on two proteins: collagen, which gives it tensile strength, and elastin, which lets it snap back. Fibroblasts, the resident cells of the dermis, make both and replace them continuously.

Inflammatory signals interfere with that maintenance in two ways. They tell fibroblasts to make less collagen and less elastin. And they raise production of enzymes that cut both apart. Immune cells arriving at the tissue add their own degradation enzymes.

Once fibres start breaking, the process sustains itself. Fibroblasts anchor to intact collagen and pull against it. When the fibres fragment, the cells lose that tension, collapse in shape, and respond by making more of the enzyme that caused the fragmentation. The clearest human evidence comes from ultraviolet exposure, where the inflammatory response and the structural damage have been measured in the same skin.

Production falls. TNF-α reduces procollagen α1(I) transcription by about half in human dermal fibroblasts, with no effect on unrelated genes in the same cells [1]. In human dermal fibroblasts the repression has been mapped to an NF-κB site in the COL1A2 promoter, with binding detectable within five minutes of TNF-α exposure and unaffected by blocking new protein synthesis — meaning the repressor is already present in the cell [2]. A third route, through repressive C/EBP proteins, was established in a mouse fibroblast line using the human COL1A2 promoter rather than in human cells [3]. Type I and type III collagen and fibronectin all fall with dose in human dermal fibroblasts, by a mechanism independent of prostaglandin metabolism and not explained by intracellular retention [4].

The net direction matters more than either arm alone. In three lines of cultured adult human dermal fibroblasts, IL-1 and TNF raised collagen production as well as collagenase — but the rise in collagenase activity was more than tenfold greater than the rise in collagen, and those authors concluded that these cytokines act predominantly catabolically in the dermis [65]. Inflammatory signalling drives both building and breaking; breaking wins.

The elastin evidence is stronger than the collagen evidence. In cultured human skin fibroblasts, TNF-α suppressed elastin mRNA by up to 91% and elastin promoter activity by up to 70%. The effect was abolished by an AP-1 oligonucleotide and mapped to a specific site in the promoter [5]. That matters here because the human studies connecting fat to skin structure, in 2c, measure elastic fibres rather than collagen.

Breakdown rises, from two sources. Fibroblasts produce MMP-1, the main collagen-cutting enzyme of human skin, induced by IL-1β through MAP kinase signalling [6, 7]. Immune cells bring different tools. Neutrophils carry MMP-8 ready-made, released within hours rather than transcribed [8]. Macrophages produce MMP-12, macrophage elastase, which cleaves insoluble human skin elastin extensively — 36 cleavage sites have been mapped [9] — and whose expression rises in human skin after ultraviolet exposure — 11.9-fold at the transcript level within 16 hours — with the protein found in sun-exposed skin colocalized with solar elastosis, the degraded elastic material itself, and scarce in sun-protected skin from the same people [10]. That study measured expression and localisation rather than enzyme activity.

The two populations compound each other. Neutrophil elastase degrades decorin, a proteoglycan that coats collagen fibrils and shields them. Once decorin is cleaved, the same collagen fibril is cut far more readily by MMP-1 [11]. The infiltrate doesn’t only add enzyme. It strips the protection off the target.

Degradation has been measured directly in human skin. Human skin explants incubated with TNF-α for eight days lost type I collagen measurably faster than controls. Blocking MMPs abolished the loss. A neutralising antibody identified MMP-1 as the enzyme doing the cutting, and MMP-3, which activates MMP-1, rose in amount and in activity [12]. In mice lacking MMP-3, TNF-α did not accelerate degradation at all [13].

The effect of different exposure times. In human dermal fibroblasts, brief TNF-α exposure produces a signalling response that reverses. Prolonged exposure raises reactive oxygen species and pushes the cells into premature senescence, with a phenotype that persists after the stimulus is withdrawn and keeps secreting inflammatory mediators and matrix-degrading enzymes [14].

Self-perpetuation. Intact collagen fibrils provide a stable platform for fibroblasts to bind and act. Where fibrils are fragmented, binding and mechanical stability fall and the cells adopt a collapsed shape [15]. Reduced spreading and reduced mechanical force activate c-Jun/AP-1, which raises MMP-1 — the enzyme that produced the fragmentation — while procollagen production falls [16].

Ultraviolet exposure, in living human skin. A single irradiation induces MMP-1, MMP-3 and MMP-9, with MMP-1 responsible for the collagen damage [17, 18]. Neutrophils enter the papillary dermis between four and eight hours, and MMP-8 protein rises from their preformed stores [8]. Decorin degradation follows the same time course as the neutrophil influx [11]. Pretreatment with the anti-inflammatory clobetasol blocked the neutrophil infiltration, the MMP-8 rise, the MMP-1 induction and the collagen degradation [8]. A single exposure raises collagenase and gelatinase activity within eight hours, peaking at 24 hours and returning to near baseline by 72 hours. But with four exposures at half the minimal erythema dose, delivered at 48-hour intervals to separate sites, both enzymes remained maximally elevated — 4.4-fold and 2.3-fold — for seven days [17]. Repeated low-level exposure does not let the tissue return to where it started.

Smoking. In 236 current and former smokers, lower forearm skin elasticity correlated with higher plasma soluble TNF receptors, CRP, pentraxin-3 and serum amyloid A, and with higher MMP-1 and TIMP-1, -2 and -4. After adjustment for smoking status, pack-years and airflow obstruction, TNFR2, CRP, SAA, TIMP2 and TIMP4 remained significant; the others did not. Skin elasticity also tracked CT-quantified emphysema, and in emphysema-susceptible individuals elasticity-determined skin age exceeded chronological age by a median 4.6 years [19].

Three things about that study should be said rather than glossed. Plasma TNF-α and IL-6 — the two mediators this section is built on — showed no correlation with skin elasticity or with facial wrinkling. The markers that did correlate are soluble receptors and acute-phase proteins, which index inflammatory tone rather than the cytokines themselves. And the protease inhibitors rose alongside the protease, so no net proteolytic balance can be read from it.

Air pollution. The human evidence here is weaker than it is usually presented. A systematic review found four eligible cohorts. Pooled pigmentary aging rose by a ratio of 1.11 per 10 µg/m³ of PM2.5, but the confidence interval ran from 0.82 to 1.50 — every individual estimate pointed the same way, yet the pooled result is compatible with no effect. Wrinkles rest on a single cohort reporting 3.2% greater severity per 10 µg/m³, while imaging-based wrinkle scores elsewhere showed no significant change. Those authors call the evidence base small, heterogeneous, of low certainty and hypothesis-generating [20]. The mechanism in culture is clearer: particulate matter raises IL-1β, IL-6, IL-8, MMP-1 and MMP-3 and lowers collagen and elastin transcripts in human dermal fibroblasts [21].

The mechanistic work is short-exposure and mostly in isolated cells. Hours of cytokine at laboratory concentrations is not months of low-grade exposure in a living person, and the one study showing that duration changes the outcome [14] suggests the two are different in kind, not only in degree.

The eight-day explant study [12] is the longest experiment in intact human skin we are aware of, and eight days is still short. In that study the epidermis was the main source of MMP-1, so the responding cell was not identified.

Clobetasol is a glucocorticoid. Glucocorticoids suppress collagen synthesis on their own, so that experiment shows the pathway can be blocked but cannot separate blocked inflammation from a broadly suppressed fibroblast.

In addition to stimulating inflammation, UV exposure damages DNA and matrix proteins directly.

The smoking and pollution studies measure skin function and appearance, not matrix composition.

No study has increased or blocked inflammation alone, over a long period, in living human skin, and measured what happened to collagen and elastin.

Enlarged fat cells can weaken skin by two routes: indirectly, through the inflammation described in Section 1, and directly, through what they release onto fibroblasts. Between the two sits the observational evidence: the conditions that enlarge or shrink fat cells move skin structure in the same direction.

Fat cells are not simply storage vesicles. They secrete hormones and signalling molecules, and what they secrete changes with their size. As a fat cell enlarges, it releases more inflammatory mediators and, more importantly, recruits immune cells into the tissue. Much of the inflammatory output of enlarged fat tissue comes from those recruited cells rather than from the fat cells themselves.

The size difference required is smaller than you might expect. A difference in cell size of less than 10% is associated with meaningful changes in inflammation.

Some of this fat – dermal white adipose tissue – sits inside the skin. This fat has a distinct developmental origin from subcutaneous fat tissue, and its cells share progenitors with dermal fibroblasts. The rest sits immediately beneath – subcutaneous fat tissue. Fibroblasts carry receptors for signals that fat cells send.

And the destructive state is self-reinforcing. Inflammation suppresses the formation of new fat cells, which means the existing cells carry more lipid each, which makes them larger still.

Secretion changes with size. When fat cells from the same person are separated by size, the larger cells secrete more. In abdominal subcutaneous adipocytes from 29 donors, sorted into four fractions spanning a fivefold range of mean volume, release of leptin, IL-6, IL-8 and MCP-1 rose with cell size and remained higher in the largest cells after correcting for cell surface area. IL-10 tended to fall. TNF-α rose only in the largest fraction, and not after that correction [22]. Because the comparison is within each donor, body weight cannot explain it. The effect was concentrated in the very largest cells. In human abdominal tissue spanning a wide weight range, mean adipocyte cross-sectional area predicted macrophage content strongly and continuously, rather than only at the obese end [23]. At matched weight and fat mass, cell size correlates with the M1/M2 macrophage ratio and with TNF-α secretion [24], and predicts senescence markers better than BMI does [25].

A small size difference, a large inflammatory difference. The clearest demonstration compares 17 non-obese, glucose-tolerant first-degree relatives of type 2 diabetics with 17 matched controls [26]. BMI was 24.1 against 24.7, fat mass 16.6 against 18.6 kg, percent fat mass 22.9 against 25.3 — none of these significantly different. Adipocyte diameter was 88 µm against 96 µm, a 9% difference, p = 0.016.

Alongside that 9%: TNF-α strongly upregulated in adipose tissue, CD68 raised, MCP-1 doubled, and IL-1β, IL-1RA and IL-10 all significantly increased in both isolated fat cells and whole tissue. Those authors state the inflammatory elevation in these lean subjects is primarily related to adipocyte hypertrophy, since the groups did not differ in BMI or body fat.

At this degree of enlargement, the inflammation is local. In the same study, serum IL-1β and IL-1RA did not differ between the groups, and neither correlated with cell size [26]. These were lean subjects with a 9% difference in cell size. In obesity, where enlargement is far greater, the signal reaches the blood. Genetically higher BMI raises circulating CRP, and in the primary analysis IL-6 [38]. The point for skin is that modest hypertrophy can inflame its surroundings without showing up in a blood test.

Recruited immune cells amplify inflammation. Within adipose tissue, TNF-α comes predominantly from infiltrating macrophages rather than from the adipocytes [23]. Cell size is the upstream variable; recruitment is the step in between. This connects directly to Section 1: the infiltrate is the source of the enzymes fibroblasts don’t make. One recruitment mechanism has been identified in detail. IL-6 does not act alone; it pairs with a soluble form of its own receptor, and that complex is chemotactic for macrophages. An IL-6/soluble-receptor fusion protein drew macrophages as strongly as MCP-1, the classical chemoattractant, and blocking the pathway abolished the effect. Circulating soluble IL-6 receptor correlates with macrophage marker expression in adipose tissue across lean, obese and diabetic subjects [27].

The state locks in. The proportion of precursor cells that can become new fat cells correlates inversely with donor adipocyte size [28]. TNF-α blocks differentiation of human preadipocytes outright [28], and in a mouse preadipocyte line the mechanism has been traced to a β-catenin/TCF4 pathway downstream of the TNF receptor [29]. In older subjects, precursor replication correlates inversely with TNF-α release from the same person’s adipose tissue [30]. Fewer new cells means existing cells each carry more lipid per cell.

This fat is in the skin. Dermal white adipose tissue is developmentally distinct from subcutaneous fat, and its adipocytes arise from progenitors shared with dermal fibroblasts [31]. It is transcriptionally distinct, with an immature preadipocyte population enriched in skin [32]. In mice a muscle layer separates the two depots. In human skin there is no such boundary: dermal adipocytes cluster around the lower hair follicle and run continuously into the superficial subcutaneous fat [33]. It can be measured directly in human skin: size, number and lipid content in place [33]. Adipogenesis in adult skin is active and tied to the hair cycle [34].

Fibroblasts can receive the signal. Human dermal fibroblasts express adiponectin receptors 1 and 2 and the leptin receptor [35]. The signalling network is established.

Adiposity and inflammatory skin conditions. In the Nurses’ Health Study II, 89,886 women followed 14 years produced 5,249 incident cases of rosacea, with risk rising across every BMI category: 11% higher at BMI 25.0–29.9, 21% at 30.0–34.9 and 48% at 35 or above, against a 21.0–22.9 reference. Weight gain after age 18 added 4% per ten pounds, and waist circumference predicted independently of BMI [36]. A Mendelian randomization study — a design that uses genetic variants as a proxy for BMI, so the comparison is much less confounded than one based on measured weight — found that higher BMI causally increases psoriasis risk, across 753,421 individuals from UK Biobank, the Nord-Trøndelag Health Study and published genome-wide association studies [37]. The reverse was not true: psoriasis genetic risk showed no causal effect on BMI. A separate two-sample analysis puts the magnitude at a 57% increase in psoriasis risk per standard deviation of BMI [38].

Rosacea is the more informative of the two here, because the dermis has been examined. In a case-control study of five patients with mild-to-moderate erythematotelangiectatic rosacea and five matched controls, collagen content by picrosirius red was significantly lower in rosacea skin (16.8% ± 7.8% against 19.6% ± 8.8%, p = 0.030), and every one of the five patient-control pairs showed the same direction [63]. Those authors read it the way Section 1 does: where the collagen matrix is degraded, fibroblasts make less procollagen.

A separate study supplies the enzymes. Comparing 26 patients with erythematotelangiectatic rosacea against 20 with telangiectatic photoaging and 11 age- and sex-matched controls, expression of MMP-1, MMP-3 and MMP-9 was significantly higher in rosacea than in controls, with MMP-3 five-fold higher than controls and four-fold higher than in photoaged skin [67]. Those are the same enzymes named in Section 1, raised in a condition defined by chronic cutaneous inflammation, in skin that also carries less collagen.

Almost all of this work is in subcutaneous or visceral fat. Whether dermal fat recruits macrophages the same way has not been tested. This is the largest untested assumption on the page.

The first-degree-relatives study measures gene expression rather than secreted protein, and it is a relatively small study (n = 17).

Both studies at matched body weight were in people with type 2 diabetes or their first-degree relatives [24, 26], so a shared metabolic predisposition could contribute to both the enlargement and the inflammation.

The size-fraction study used isolated fat cells in 24-hour culture, and those authors note that stress during cell isolation could itself have increased release [22].

Hypoxia is often proposed as the link between cell enlargement and inflammation. We are not asserting it: one of these cohorts found no difference in hypoxia markers despite significant hypertrophy [26], while another reports a large increase [39]. Two human studies, opposite results.

IL-6 appears in the list of size-dependent adipokines, but it does not act on dermal fibroblasts the way TNF-α does. IL-6 increases collagen production in adult human dermal fibroblasts [40], does not induce collagenase in them [41], and raises their production of the MMP inhibitor TIMP [42]. We include it here as a recruitment mechanism, not as a signal that damages the dermis directly.

Rosacea and psoriasis are inflammatory diseases with genetic drivers of their own, unrelated to fat cell size. An association with body weight shows that adiposity adds to cutaneous inflammatory burden. It does not show that fat cell enlargement is what damages the matrix.

The rosacea biopsy study [63] has five patients and five controls. It is a comparison of rosacea skin against healthy skin, not across body weight, and no adipose measurement was made in those subjects. It shows that a chronically inflamed dermis carries the matrix changes described in Section 1. It does not connect them to fat.

The things that enlarge fat cells are the same things known to weaken skin: obesity, age, menopause, ultraviolet exposure. The things that reduce fat cell size — exercise, diet, weight loss — are associated with better skin structure.

These are correlations. Each of these exposures changes many things at once, and none of the studies isolates fat cell size as the operative variable.

The strongest single item is a paired-site study in which adipokine production in fat and receptor expression in the dermis above it were both reduced at photodamaged sites, and silencing those adipokines raised MMP-1 and lowered procollagen in the dermal compartment [49].

On the other side: eight weeks of voluntary wheel running in diet-induced obese mice reduced adipose thickness and adipocyte diameter toward normal values, and collagen density and second-harmonic imaging signal rose to control levels alongside [50]. In aged mice fed 25 different diets, skin structure tracked macronutrient ratio rather than energy intake, and adipocyte size correlated inversely with dermal thickness [51]. In overweight and obese young men who reduced to normal BMI, dermal echogenicity improved at the thigh in 83.3% of subjects [52].

One note on how these outcomes are measured, because it changes how to read them. Second-harmonic generation imaging detects fibrillar collagen. Dermal thickness is a gross dimension. In one of these studies the imaging signal fell significantly while thickness did not move [53], which means thickness is insensitive to matrix changes that imaging detects. A thickness result showing no change is weak evidence of no effect.

One note on the menopause study. The markers that rose in subcutaneous fat were IL-6, IL-18, MCP-1, fibrosis and hypoxia markers, rather than the TNF-α that Section 1 runs on; TNF-α expression and CD68 macrophage infiltration did not differ significantly [39]. Those are null results in a small cohort — 13 premenopausal and 15 postmenopausal women, with five from each group excluded from the subcutaneous transcript analyses for lack of tissue, leaving roughly eight against ten. At that size a modest difference would not be detected, so the nulls do not establish that TNF-α and macrophage recruitment are unchanged. The findings that did reach significance did so despite the same limited power.

Exercise, diet and weight loss are independently anti-inflammatory and change mechanical loading. None of these designs isolates adipocyte size. In one of the same studies, a high-fructose arm lowered collagen signal without changing body weight or adipocyte size at all [53] — diet composition can reduce collagen quality by a route that has nothing to do with fat cell size.

Not every human dataset points the same way on body weight. In the smoking cohort described in Section 1, higher BMI was associated with better forearm skin elasticity (β = 0.46, P < 0.0001) and less facial wrinkling (β = −0.15, P = 0.025) [19]. Both are functional or visual measures rather than compositional ones, and fuller subcutaneous tissue can raise a viscoelasticity reading without any improvement in matrix — but the finding runs against the obesity row above and we record it.

Most of this section is mouse work. The one human intervention measured echogenicity at a single site.

When enlarged fat cells are placed near fibroblasts, the fibroblasts make less collagen and less elastin, and more of an enzyme that breaks collagen down. Small fat cells produce no such effect. In that experiment the two cell types shared fluid but could not touch, so the signal was something the fat cells released. It was traced to free fatty acids, and one fatty acid applied alone reproduced the effect. This is a different route from the inflammatory one described in Sections 1 and 2a.

The same pattern appears in a living animal. In mice fed a high-fat diet, the fat layer beneath the skin thickened and the dermis above it thinned, with fewer fibroblasts and less elasticity. The ear, which has no fat layer beneath its dermis, did not thin in the same animals. That is what you would expect if the signal travels a short distance and not through the bloodstream.

In people the evidence shows correlation. A thicker fat layer in the lower cheek is associated with lower skin elasticity and more sagging, measured lying down so that gravity is not the explanation. In abdominal skin from women aged 40 to 59, larger fat cells were associated with fewer elastic fibres in the dermis above them.

Most of this work comes from a single research group (Shiseido) and has not been independently replicated.

In mice fed a high-fat diet, the subcutaneous fat layer thickened and the dermis above it thinned, with fewer fibroblasts and reduced elasticity. In the same animals, the ear — which has no subcutaneous fat layer beneath its dermis — did not thin [54]. A difference between two sites in the same animal is not what a circulating mediator produces.

In a co-culture where fat cells and fibroblasts shared medium but could not touch, enlarged adipocytes reduced fibroblast proliferation to about 68% of control, reduced collagen I transcript to about a third, reduced elastin, and raised MMP13 roughly 3.7-fold. Small adipocytes produced no significant change in any measure. The mediator was identified as free fatty acid release, reproduced by palmitic acid applied directly [55]. Separately, enlarged fat cells raise MMP9 through ERK–AP-1 signalling, and MMP9 is more abundant in the adipose layer of high-BMI human abdominal specimens [43].

In humans: ultrasound thickness of the lower-cheek fat layer correlates negatively with dermal elasticity and positively with graded sagging, measured lying down so that gravity is not the explanation [56]. In abdominal skin specimens from women aged 40 to 59, adipocyte size correlated with BMI at R = 0.739 and with elastic fibre abundance at R = −0.631 [43].

The studies in this section — the ear-pinna experiment, the co-culture, the cheek study and the abdominal specimens — come from one research group, Ezure and Amano at Shiseido. We credit them for the foundational work. This work has not been independently replicated.

The two direct experiments identify non-inflammatory mediators: free fatty acids and an enzyme made by the fat cell itself. They support the link between cell size and matrix damage. They do not run through the inflammatory pathway described in Sections 1 and 2a.

In the abdominal specimens, collagen density did not differ by BMI.  That result was reported without image analysis as was applied to the study of elastic fibres which did show a difference [43].

The human studies are correlations.

KPRE reduces the size of mature fat cells. It does this in part by activating the machinery that releases stored lipid, and, unlike some other agents that shrink fat cells, supports rather than shuts down the formation of new ones. In human 3D cultures it raised cell number whether added during or after differentiation, and kept lipid droplets smaller when added during.

Applied to the surface of human skin explants with the fat layer intact, KPRE reduced the size of fat cells in the superficial fat of the skin, but only where those cells were already enlarged. In skin with small adipocytes it changed nothing.

By mouth in aged mice, KPRE reduced fat across every depot measured and brought raised serum inflammatory markers back to young values. In a separate study in obese mice, a K. parviflora extract given by mouth reduced subcutaneous fat, shifted fat cells toward smaller sizes, and preserved the structure of dermal collagen in the same animals.

The topical studies with KPRE are limited to showing a reduction in fat cell size. No topical study has measured the effect of fat cell size reduction on inflammation, collagen, elastin, or any other property of the dermal matrix.

In mature fat cells. Applied to 3T3-L1 adipocytes after differentiation was complete, KPRE reduced intracellular triglyceride at 10 µg/ml and 30 µg/ml without cytotoxicity. ATGL and HSL, the enzymes that mobilise stored lipids, rose with dose, ATGL protein rose on immunoblot, and glycerol release increased. PPARγ transcript was unchanged and a PPARγ agonist was inert in the same system, so the effect in mature cells does not run through that receptor [57]. That is why reduced size and preserved cell formation can coexist for this agent.

In human 3D cultures, size and number move in opposite directions. Primary human preadipocyte spheroids exposed to KPRE during differentiation formed lipid droplets more slowly and the droplets stayed smaller than in untreated controls. Cell numbers in KPRE-treated spheroids more than doubled against control, while overall spheroid volume did not increase: more cells, each smaller, in the same tissue volume. The preadipocyte marker PDGFRA rose more than tenfold [58].

Exposed instead for seven days after differentiation was complete, KPRE reduced lipid droplet diameter numerically relative to controls although this reduction did not reach statistical significance. KPRE also shifted droplet distribution so that most cells carried a single small droplet where controls were multilocular. Cell number rose here too, to about 1.9-fold of control (n = 3, p < 0.05) [58].

Applied topically to human skin explants. Full-thickness explants (skin with the dermis and fat layer intact, epidermis facing up, KPRE formulation applied to the surface daily for six days) came from two donors: a 50-year-old man with BMI 28.1 and a 31-year-old woman with BMI 24.4. Lipid droplet diameter was measured by PLIN2 immunohistochemistry in the superficial fat of the explant.

KPRE-only formulations had no significant effect on droplet size in explants with small adipocytes. In explants with hypertrophic adipocytes, droplet diameter fell highly significantly. The effect was dose dependent, with twice-daily treatment having a greater effect than once-daily treatment [58].

Where this fat sits matters for the argument. Almost all of the evidence in 2a and 2c comes from subcutaneous fat, and the studies closest to skin measured the fat immediately beneath the dermis [43, 54–56]. The explant measurement covers that same superficial fat.

By mouth, fat and serum inflammation fall together. Aged mice given oral KPRE for eight weeks at 100 or 200 mg/kg lost fat in every depot: abdominal fat volume relative to body weight fell 26.5% and 45.0%, subcutaneous adipose weight 17.7% and 29.7%, visceral 35.5% and 55.9%, perirenal 48.2% and 42.4% [64]. In the same animals, serum TNF-α was 9.94 pg/mL in young mice, 33.19 in aged, and 11.84 and 11.17 in the two treated groups; serum IL-6 was 10.05, 40.37, 13.78 and 15.73. Both returned to roughly young values, p < 0.01 against aged. NF-κB protein fell dose-dependently [64].

Adipocyte size was not measured in that study but it is reasonable to assume that fat mass and fat cell size would move together. A separate oral study in high-fat-diet mice did measure it, and found adipocyte size reduced along with fat mass and adipose tissue weight [66].

By mouth, in the one study with a skin endpoint. Obese TSOD mice were fed a K. parviflora ethanol extract at 1% of diet for twelve weeks from four weeks of age, then given a single dose of UVB [62]. There were nine animals per group. The extract reduced visceral and subcutaneous fat area on CT and thinned the subcutaneous fat layer. It shifted subcutaneous adipocyte sizes toward smaller cells, with fewer above 7,000 µm² (n = 3 per group, p < 0.05). Under scanning electron microscopy, collagen fibres in the upper and lower dermis were thicker and more densely organised than in untreated obese mice. In the upper dermis, treated mice looked better even than untreated obese mice that had not been irradiated, although every treated animal received UVB.

Lipolysis and free fatty acids. KPRE reduces fat cell size by releasing stored lipid, and 2c identifies free fatty acid release from enlarged fat cells as a signal that suppresses fibroblasts [55]. So it is fair to ask whether KPRE raises the exposure it is meant to lower. We think the net effect runs the other way. In that same co-culture, small adipocytes produced no significant change in any fibroblast measure [55]. In human spheroids KPRE more than doubled cell number while droplets stayed smaller when added during differentiation, and raised cell number about 1.9-fold when added afterwards, with a fall in droplet diameter that did not reach significance [58]. If the same shift occurs in skin, the result is more fat cells, each smaller, and on the evidence of [55] that population should expose fibroblasts to less free fatty acid than the enlarged cells it replaces. In the one animal study with a skin endpoint, smaller fat cells and better-preserved collagen were found together [62]. In explants we have measured the reduction in superficial fat cell size over six days. We have not measured free fatty acid levels or fibroblast response in treated skin.

What was tested. The material was a standardized K. parviflora microemulsion from Ava Plant, Thailand, quantified by HPLC against 5,7-dimethoxyflavone as principal marker at >1%, with total specific flavonoid content >3%, expressed relative to the formulated microemulsion rather than a dry extract. Spheroid experiments used 5% v/v KPRE [58].

Delivery. Dermal adipocytes cluster around the base of hair follicles, and the follicular route is accessible to lipophilic molecules below roughly 500 daltons [33, 34, 59, 60]. The KPRE polymethoxyflavones are 268 to 372 daltons, and the permeation of three of them has been measured across porcine skin [61].

These are the most important limitations on the page, because this is where our commercial interest lies.

No matrix endpoint has been measured for our material, or by the topical route. Cell size and cell number are the outcomes throughout [58]. The one K. parviflora study with a collagen endpoint [62] used a different extract, by mouth. Skin elasticity was in fact a prespecified primary endpoint of the clinical work described in [58], and was abandoned because it could not be measured in a way representative of the whole face, a decision taken before any outcome data were analysed.

No inflammatory endpoint has been measured in skin or in fat. By the oral route in aged mice, fat mass and serum inflammatory markers fell together [64]. Serum does not identify the source. These mice were aged, not obese, and neither adipocyte size nor adipose inflammation was measured. NF-κB also fell, so a direct anti-inflammatory action of KPRE, independent of fat, cannot be excluded.

The oral skin study has limits of its own. The collagen result in [62] is descriptive: representative micrographs, with no quantification and no statistics. The design was preventive, so it says nothing about reversing established damage. There was no treated group without UVB, so protection against obesity and protection against ultraviolet cannot be fully separated. Two other fractions of the extract preserved collagen to a lesser degree without shifting adipocyte size, though both reduced fat mass. Smaller fat cells accompanied the best collagen result, but a partial result occurred without them.

Explants are not living skin. The model allows many samples per condition and statistically testable results, which biopsies in living people do not. It does not establish that the same thing happens in everyday use. Six days is short, the tissue is not perfused, and the work used two donors. The contrast between enlarged and small adipocytes is a contrast between those two donors, so it is confounded with age, sex and BMI and has not been tested within a donor. In unperfused tissue a compound applied to the surface can also diffuse further than living dermal circulation would allow.

Follicular accumulation has not been measured. The porcine permeation study used heat-separated epidermis, so the dermis, the compartment containing the target cells, was absent, and the receptor fluid was a sink with no physiological counterpart. Transit was measured. Arrival at the target was not.

The active constituents are not established. In one animal study, a fraction rich in polymethoxyflavones did not reproduce the whole extract’s shift in adipocyte size, and a fraction nearly devoid of them still reduced fat area and partly preserved collagen [62].

Disclosure. The spheroid and explant work [58] was partially financed by Adipeau and is published open access under CC BY. One author is a founder of Adipeau and holds patents involving the compounds analysed.

Part one: inflammation weakens skin structure. Well supported at the level of mechanism, in the right cell type, by independent groups. In intact human skin the support is an eight-day explant study and ultraviolet exposure in living skin, each with the caveats given in Section 1. The gap is exposure regime: nearly all of it is short exposure, and nobody has manipulated inflammation alone in living human skin over a long period with matrix as the endpoint.

Part two: enlarged fat cells are a source of inflammation. Well supported in humans, across independent groups, including within the same donor and at matched body weight and fat mass, which argues against body weight as the explanation. The signals that rise most consistently with cell size are those that recruit immune cells. Within fat, most TNF-α comes from those recruited macrophages, so the link to Part one runs through recruitment. The gap is location: this has been shown in subcutaneous and visceral fat, not in the superficial fat close to the skin.

Part three: a topical that reduces fat cell size in the superficial fat of skin can plausibly help skin structure. Two lines of evidence support it. Independent of KPRE, Sections 1 and 2a describe one route from enlarged fat cells to weaker skin, through inflammation, and 2c describes another, through free fatty acids. Conditions that enlarge or shrink fat cells are associated with worse or better dermal structure in animals and in people, and one study found that enlarged fat cells suppress collagen and elastin production by fibroblasts where small ones do not (2b and 2c). KPRE itself reduces fat cell size in human skin explants under topical application. By mouth, three mouse studies using different K. parviflora extracts point the same way: fat mass and serum inflammatory markers fell together in aged mice, fat cells shrank in diet-induced obese mice, and in the one study with a skin endpoint, smaller fat cells accompanied better-preserved dermal collagen. No single study measured all three, and these are oral, not topical, data. No one has done a topical study with KPRE measuring fat cell size y inflammation or dermal matrix. We are currently planning such a  study.  Hence, plausible, not proven.

This is our grading, and it may be inaccurate. If you think we’ve read any of this wrong, we’d like to hear it.

For the full library of studies, see the Study Index.

1. Solis-Herruzo JA, Brenner DA, Chojkier M. Tumor necrosis factor alpha inhibits collagen gene transcription and collagen synthesis in cultured human fibroblasts. J Biol Chem. 1988;263(12):5841–5. PMID 3258601.
2. Kouba DJ, Chung KY, Nishiyama T, Vindevoghel L, Kon A, Klement JF, Uitto J, Mauviel A. Nuclear factor-kappa B mediates TNF-alpha inhibitory effect on alpha 2(I) collagen (COL1A2) gene transcription in human dermal fibroblasts. J Immunol. 1999;162(7):4226–34. PMID 10201951.
3. Greenwel P, Tanaka S, Penkov D, Zhang W, Olive M, Moll J, Vinson C, Di Liberto M, Ramirez F. Tumor necrosis factor alpha inhibits type I collagen synthesis through repressive CCAAT/enhancer-binding proteins. Mol Cell Biol. 2000;20(3):912–918. PMID 10629048.
4. Mauviel A, Daireaux M, Rédini F, Galera P, Loyau G, Pujol JP. Tumor necrosis factor inhibits collagen and fibronectin synthesis in human dermal fibroblasts. FEBS Lett. 1988;236(1):47–52. PMID 3165348.
5. Kähäri VM, Chen YQ, Bashir MM, Rosenbloom J, Uitto J. Tumor necrosis factor-alpha down-regulates human elastin gene expression. Evidence for the role of AP-1 in the suppression of promoter activity. J Biol Chem. 1992;267(36):26134–41. PMID 1281483.
6. Honda A, Abe R, Makino T, Norisugi O, Fujita Y, Watanabe H, Nishihira J, Iwakura Y, Yamagishi S, Shimizu H, Shimizu T. Interleukin-1β and macrophage migration inhibitory factor (MIF) in dermal fibroblasts mediate UVA-induced matrix metalloproteinase-1 expression. J Dermatol Sci. 2008;49(1):63–72. PMID 18060744.
7. Bauer J, Huy C, Brenmoehl J, Obermeier F, Bock J. Matrix metalloproteinase-1 expression induced by IL-1beta requires acid sphingomyelinase. FEBS Lett. 2009;583(5):915–20. PMID 19302792.
8. Fisher GJ, Choi HC, Bata-Csorgo Z, Shao Y, Datta S, Wang ZQ, Kang S, Voorhees JJ. Ultraviolet irradiation increases matrix metalloproteinase-8 protein in human skin in vivo. J Invest Dermatol. 2001;117(2):219–26. PMID 11511297.
9. Taddese S, Weiss AS, Neubert RHH, Schmelzer CEH. Mapping of macrophage elastase cleavage sites in insoluble human skin elastin. Matrix Biol. 2008;27(5):420–8. PMID 18334288.
10. Chung JH, Seo JY, Lee MK, Eun HC, Lee JH, Kang S, Fisher GJ, Voorhees JJ. Ultraviolet modulation of human macrophage metalloelastase in human skin in vivo. J Invest Dermatol. 2002;119(2):507–12. PMID 12190877.
11. Li Y, Xia W, Liu Y, Remmer HA, Voorhees J, Fisher GJ. Solar ultraviolet irradiation induces decorin degradation in human skin likely via neutrophil elastase. PLoS One. 2013;8(8):e72563. PMID 24023624.
12. Ågren MS, Schnabel R, Christensen LH, Mirastschijski U. Tumor necrosis factor-α-accelerated degradation of type I collagen in human skin is associated with elevated matrix metalloproteinase (MMP)-1 and MMP-3 ex vivo. Eur J Cell Biol. 2015;94(1):12–21. PMID 25457675.
13. Mirastschijski U, Lupše B, Maedler K, Sarma B, Radtke A, Belge G, Dorsch M, Wedekind D, McCawley LJ, Boehm G, Zier U, Yamamoto K, Kelm S, Ågren MS. Matrix metalloproteinase-3 is a key effector of TNF-α-induced collagen degradation in skin. Int J Mol Sci. 2019;20(20):5234. PMID 31652545.
14. Mavrogonatou E, Konstantinou A, Kletsas D. Long-term exposure to TNF-α leads human skin fibroblasts to a p38 MAPK- and ROS-mediated premature senescence. Biogerontology. 2018;19(3-4):237–249. PMID 29582209.
15. Varani J, Dame MK, Rittié L, Fligiel SEG, Kang S, Fisher GJ, Voorhees JJ. Decreased collagen production in chronologically aged skin. Am J Pathol. 2006;168(6):1861–1868. PMID 16723701.
16. Qin Z, Fisher GJ, Quan T. Age-associated reduction of cellular spreading/mechanical force up-regulates matrix metalloproteinase-1 expression and collagen fibril fragmentation via c-Jun/AP-1 in human dermal fibroblasts. Aging Cell. 2014. doi:10.1111/acel.12265. PMID 25201474.
17. Fisher GJ, Wang ZQ, Datta SC, Varani J, Kang S, Voorhees JJ. Pathophysiology of premature skin aging induced by ultraviolet light. N Engl J Med. 1997;337(20):1419–1428. PMID 9358139.
18. Brennan M, Bhatti H, Nerusu KC, Bhagavathula N, Kang S, Fisher GJ, Varani J, Voorhees JJ. Matrix metalloproteinase-1 is the major collagenolytic enzyme responsible for collagen damage in UV-irradiated human skin. Photochem Photobiol. 2003;78(1):43–48. PMID 12929747.
19. O’Brien ME, Chandra D, Wilson RC, Karoleski CM, Fuhrman CR, Leader JK, Pu J, Zhang Y, Morris A, Nouraie S, Bon J, Urban Z, Sciurba FC. Loss of skin elasticity is associated with pulmonary emphysema, biomarkers of inflammation, and matrix metalloproteinase activity in smokers. Respir Res. 2019;20(1):128. PMID 31234847.
20. Tjiu JW, Lu CF. Long-term PM2.5 exposure and clinical skin aging: a systematic review and meta-analysis of pigmentary and wrinkle outcomes. Life. 2026;16(1):61. PMID 41598216.
21. Park SY, Byun EJ, Lee JD, Kim S, Kim HS. Air pollution, autophagy, and skin aging: impact of particulate matter (PM10) on human dermal fibroblasts. Int J Mol Sci. 2018;19(9):2727. PMID 30213068.
22. Skurk T, Alberti-Huber C, Herder C, Hauner H. Relationship between adipocyte size and adipokine expression and secretion. J Clin Endocrinol Metab. 2007;92(3):1023–33. PMID 17164304.
23. Weisberg SP, McCann D, Desai M, Rosenbaum M, Leibel RL, Ferrante AW Jr. Obesity is associated with macrophage accumulation in adipose tissue. J Clin Invest. 2003;112(12):1796–808. PMID 14679176.
24. Acosta JR, Douagi I, Andersson DP, Bäckdahl J, Rydén M, Arner P, Laurencikiene J. Increased fat cell size: a major phenotype of subcutaneous white adipose tissue in non-obese individuals with type 2 diabetes. Diabetologia. 2016;59(3):560–70. PMID 26607638.
25. Gustafson B, Nerstedt A, Smith U. Reduced subcutaneous adipogenesis in human hypertrophic obesity is linked to senescent precursor cells. Nat Commun. 2019;10(1):2757. PMID 31227697.
26. Henninger AM, Eliasson B, Jenndahl LE, Hammarstedt A. Adipocyte hypertrophy, inflammation and fibrosis characterize subcutaneous adipose tissue of healthy, non-obese subjects predisposed to type 2 diabetes. PLoS One. 2014;9(8):e105262. PMID 25148116.
27. Kraakman MJ, Kammoun HL, Allen TL, et al. Blocking IL-6 trans-signaling prevents high-fat diet-induced adipose tissue macrophage recruitment but does not improve insulin resistance. Cell Metab. 2015;21(3):403–16. PMID 25738456.
28. Isakson P, Hammarstedt A, Gustafson B, Smith U. Impaired preadipocyte differentiation in human abdominal obesity: role of Wnt, tumor necrosis factor-alpha, and inflammation. Diabetes. 2009;58(7):1550–7. PMID 19351711.
29. Cawthorn WP, Heyd F, Hegyi K, Sethi JK. Tumour necrosis factor-α inhibits adipogenesis via a β-catenin/TCF4(TCF7L2)-dependent pathway. Cell Death Differ. 2007;14(7):1361–1373. PMID 17464333.
30. Caso G, McNurlan MA, Mileva I, Zemlyak A, Mynarcik DC, Gelato MC. Peripheral fat loss and decline in adipogenesis in older humans. Metabolism. 2013;62(3):337–40. PMID 22999012.
31. Driskell RR, Jahoda CA, Chuong CM, Watt FM, Horsley V. Defining dermal adipose tissue. Exp Dermatol. 2014;23(9):629–31. PMID 24841073.
32. Rivera-Gonzalez GC, Butka EG, Gonzalez CE, et al. Comparative single-cell lineage tracing identifies distinct adipocyte precursor dynamics in skin and inguinal fat. Cell Stem Cell. 2025;32(8):1267–1284.e8. PMID 40744015.
33. Nicu C, Pople J, Bonsell L, Bhogal R, Ansell DM, Paus R. A guide to studying human dermal adipocytes in situ. Exp Dermatol. 2018;27(6):589–602. PMID 29603400.
34. Zhang B, Tsai PC, Gonzalez-Celeiro M, et al. Hair follicles’ transit-amplifying cells govern concurrent dermal adipocyte production through Sonic Hedgehog. Genes Dev. 2016;30(20):2325–2338. PMID 27807033.
35. Ezure T, Amano S. Adiponectin and leptin up-regulate extracellular matrix production by dermal fibroblasts. Biofactors. 2007;31(3-4):229–36. PMID 18997286.
36. Li S, Cho E, Drucker AM, Qureshi AA, Li WQ. Obesity and risk for incident rosacea in US women. J Am Acad Dermatol. 2017;77(6):1083–1087.e5. PMID 29033249.
37. Budu-Aggrey A, Brumpton B, Tyrrell J, Watkins S, Modalsli EH, Celis-Morales C, et al. Evidence of a causal relationship between body mass index and psoriasis: a Mendelian randomization study. PLoS Med. 2019;16(1):e1002739. PMID 30703100.
38. Li Z, Zhao Y. Evidence of a causal relationship between body mass index and immune-mediated and inflammatory skin diseases and biomarkers: a Mendelian randomization study. Clin Cosmet Investig Dermatol. 2024;17:2659–2667. PMID 39606276.
39. Abildgaard J, Ploug T, Al-Saoudi E, et al. Changes in abdominal subcutaneous adipose tissue phenotype following menopause is associated with increased visceral fat mass. Sci Rep. 2021;11(1):14750. PMID 34285301.
40. Duncan MR, Berman B. Stimulation of collagen and glycosaminoglycan production in cultured human adult dermal fibroblasts by recombinant human interleukin 6. J Invest Dermatol. 1991;97(4):686–692. PMID 1940439.
41. Emonard H, Munaut C, Melin M, Lortat-Jacob H, Grimaud JA. Interleukin-6 does not regulate interstitial collagenase, stromelysin and tissue inhibitor of metalloproteinases synthesis by cultured human fibroblasts. Matrix. 1992;12(6):471–4. PMID 1287415.
42. Sato T, Ito A, Mori Y. Interleukin 6 enhances the production of tissue inhibitor of metalloproteinases (TIMP) but not that of matrix metalloproteinases by human fibroblasts. Biochem Biophys Res Commun. 1990;170(2):824–829. PMID 2166509.
43. Ezure T, Amano S. Increment of subcutaneous adipose tissue is associated with decrease of elastic fibres in the dermal layer. Exp Dermatol. 2015;24(12):924–9. PMID 26194659.
44. Brincat M, Moniz CJ, Studd JW, et al. Long-term effects of the menopause and sex hormones on skin thickness. Br J Obstet Gynaecol. 1985;92(3):256–9. PMID 3978054.
45. Brincat M, Kabalan S, Studd JW, et al. A study of the decrease of skin collagen content, skin thickness, and bone mass in the postmenopausal woman. Obstet Gynecol. 1987;70(6):840–5. PMID 3120067.
46. Affinito P, Palomba S, Sorrentino C, et al. Effects of postmenopausal hypoestrogenism on skin collagen. Maturitas. 1999;33(3):239–47. PMID 10656502.
47. Li WH, Pappas A, Zhang L, Ruvolo E, Cavender D. IL-11, IL-1α, IL-6, and TNF-α are induced by solar radiation in vitro and may be involved in facial subcutaneous fat loss in vivo. J Dermatol Sci. 2013;71(1):58–66. PMID 23639700.
48. Kim EJ, Kim YK, Kim JE, et al. UV modulation of subcutaneous fat metabolism. J Invest Dermatol. 2011;131(8):1720–6. PMID 21562570.
49. Kim EJ, Kim YK, Kim MK, et al. UV-induced inhibition of adipokine production in subcutaneous fat aggravates dermal matrix degradation in human skin. Sci Rep. 2016;6:25616. PMID 27161953.
50. Lőrincz K, Haluszka D, Kiss N, et al. Voluntary exercise improves murine dermal connective tissue status in high-fat diet-induced obesity. Arch Dermatol Res. 2017;309(3):209–215. PMID 28180933.
51. Hew J, Solon-Biet SM, McMahon AC, et al. The effects of dietary macronutrient balance on skin structure in aging male and female mice. PLoS One. 2016;11(11):e0166175. PMID 27832138.
52. Matsumoto M, Ogai K, Aoki M, et al. Changes in dermal structure and skin oxidative stress in overweight and obese Japanese males after weight loss. Skin Res Technol. 2018;24(3):407–416. PMID 29377346.
53. Haluszka D, Lőrincz K, Kiss N, et al. Diet-induced obesity skin changes monitored by in vivo SHG and ex vivo CARS microscopy. Biomed Opt Express. 2016;7(11):4480–4489. PMID 27895989.
54. Ezure T, Amano S. Increased subcutaneous adipose tissue impairs dermal function in diet-induced obese mice. Exp Dermatol. 2010;19(10):878–82. PMID 19758317.
55. Ezure T, Amano S. Negative regulation of dermal fibroblasts by enlarged adipocytes through release of free fatty acids. J Invest Dermatol. 2011;131(10):2004–9. PMID 21697886.
56. Ezure T, Amano S. Influence of subcutaneous adipose tissue mass on dermal elasticity and sagging severity in lower cheek. Skin Res Technol. 2010;16(3):332–8. PMID 20637003.
57. Okabe Y, Shimada T, Horikawa T, et al. Suppression of adipocyte hypertrophy by polymethoxyflavonoids isolated from Kaempferia parviflora. Phytomedicine. 2014;21(6):800–806. PMID 24629599.
58. Taebnia N, Murawska M, Kemas AM, Sheng H, Wang Y, Xu J, Zhou T, Galanin I, Lauschke VM. Dual modulation of adipogenesis and lipolysis augments facial adipose tissue restoration. Comput Struct Biotechnol J. 2026;35(1):0213. doi:10.34133/csbj.0213.
59. Otberg N, Teichmann A, Rasuljev U, Sinkgraven R, Sterry W, Lademann J. Follicular penetration of topically applied caffeine via a shampoo formulation. Skin Pharmacol Physiol. 2007;20(4):195–8. PMID 17396054.
60. Bos JD, Meinardi MM. The 500 Dalton rule for the skin penetration of chemical compounds and drugs. Exp Dermatol. 2000;9(3):165–9. PMID 10839713.
61. Tuntiyasawasdikul S, Limpongsa E, Jaipakdee N, Sripanidkulchai B. Transdermal permeation of Kaempferia parviflora methoxyflavones from isopropyl myristate-based vehicles. AAPS PharmSciTech. 2014;15(4):947–55. PMID 24789664.
62. Hidaka M, Horikawa K, Akase T, et al. Efficacy of Kaempferia parviflora in a mouse model of obesity-induced dermatopathy. J Nat Med. 2017;71(1):59–67. PMID 27592007.
63. Thompson KG, et al. The association of photo-induced collagen degeneration and the development of telangiectasias in rosacea. J Anat. 2021. doi:10.1111/joa.13394. PMID 33432575.
64. Kim C, Kang M, Kim Y, Hwang JK. Inhibitory effect of standardized Kaempferia parviflora extract on sarcopenia by improving protein metabolism pathways in aged C57BL/6J mice. J Tradit Complement Med. 2026;16:412–421. PMID 42382057.
65. Duncan MR, Berman B. Differential regulation of collagen, glycosaminoglycan, fibronectin, and collagenase activity production in cultured human adult dermal fibroblasts by interleukin 1-alpha and beta and tumor necrosis factor-alpha and beta. J Invest Dermatol. 1989;92(5):699–706. PMID 2541208.
66. Lee HS, Jeon YE, Awa R, Yoshino S, Kim EJ. Kaempferia parviflora rhizome extract exerts anti-obesity effect in high-fat diet-induced obese C57BL/6N mice. Food Nutr Res. 2023;67. doi:10.29219/fnr.v67.9413. PMID 37691744.
67. Helfrich YR, Maier LE, Cui Y, Fisher GJ, Chubb H, Fligiel S, Sachs D, Varani J, Voorhees J. Clinical, histologic, and molecular analysis of differences between erythematotelangiectatic rosacea and telangiectatic photoaging. JAMA Dermatol. 2015;151(8):825–36. PMID 25798811.