The Science

Adipeau’s reasoning is one connected chain of evidence. Each step is a discrete, separately sourced link. The headings show how strong each class of evidence is, and the last section states what the chain does not establish.

Scorecard evaluation by Claude, 15 June, 2026.

What this scorecard does not claim. This scores the mechanism – whether the biology holds together and whether our ingredient can reach and move the target. It does not score clinical proof that the product improves skin in everyday use; that evidence lives in our Clinical Evidence section. And one link – that reducing fat-cell size restores the dermal matrix in living human skin – is not yet proven in people. We say so on purpose.

The introduction above describes what dermal fat cells do in broad terms. But how deep can they sit? What exactly do they secrete? And why does even a modest increase in their size lead to inflammatory signaling that research associates with weakened skin structure? The research points to a self-reinforcing cycle — and understanding it is the foundation for the sections that follow.

Dermal white adipose tissue (dWAT) is a specialized adipose compartment embedded within cutaneous architecture. Unlike deeper subcutaneous fat, dermal adipocytes arise from mesenchymal progenitors shared with dermal fibroblasts, positioning dWAT as a compartment with structural and signaling relationships to the surrounding dermis (Driskell et al., 2014; PMID: 24841073). Dermal adipocyte regeneration has been shown to be essential for restoring normal skin architecture after injury: newly formed hair follicles instruct wound myofibroblasts to convert into adipocytes, reconstituting the dermal fat layer in a pattern associated with reduced fibrotic repair and suggesting a potential target for scar modulation (Plikus et al., 2017; PMID: 28059714). Dermal white adipose tissue is anatomically positioned in close proximity to hair follicles, surrounding the lower portions of follicular structures that extend into the mid-to-deep dermis and subcutis (Welle, 2023; PMID: 37272599).

Adiponectin enhances dermal fibroblast collagen and hyaluronic acid synthesis (Ezure & Amano, 2007; DOI: 10.1002/biof.5520310310); hyaluronic acid synthesis specifically involves AMPK-dependent signaling (Yamane et al., 2011; PMID: 22024046). In hypertrophic adipose tissue, TNF-α and IL-6 levels are elevated — secreted by macrophages that accumulate with adipocyte hypertrophy (Weisberg et al., 2003; PMID: 14679176) and by adipocytes in a size-dependent manner (Skurk et al., 2007; PMID: 17164304). Pro-inflammatory cytokines such as TNF-α suppress collagen synthesis (Solis-Herruzo et al., 1988; PMID: 3258601) and upregulate matrix metalloproteinase expression in dermal fibroblasts (Quan et al., 2009; PMID: 19675548), contributing to loss of extracellular matrix characteristic of skin aging.

Experimental work from Shiseido demonstrated that enlarged adipocytes suppress dermal fibroblast proliferation and collagen and elastin gene expression while increasing matrix metalloproteinase expression, whereas small adipocytes do not — an effect traced to free fatty acids released by the enlarged cells (Ezure & Amano, 2011; PMID: 21697886). Conversely, adiponectin secreted by healthy adipocytes supports fibroblast collagen and hyaluronic acid synthesis (Ezure & Amano, 2007; DOI: 10.1002/biof.5520310310). Consistent with this, in human skin greater subcutaneous fat and adipocyte enlargement correlate with reduced dermal elastic-fiber abundance, via an ERK/MMP9-mediated pathway in enlarged adipocytes (Ezure & Amano, 2015; PMID: 26194659); and with aging, subcutaneous fat has been observed to infiltrate the dermal layer itself, a structural change associated with reduced dermal elasticity and an aged facial appearance (Ezure et al., 2022; PMID: 35020969; PMID: 36314382).

In a clinical study of 70 women, lower-face subcutaneous fat thickness correlated with reduced dermal elasticity and increased sagging. Facial fat predicted dermal elasticity about as strongly as chronological age did (fat–elasticity correlations up to R ≈ −0.59; age–elasticity up to R ≈ −0.63), yet fat and age were uncorrelated with each other, indicating two independent contributors to elasticity loss. Elasticity was measured with subjects lying down to remove the effect of gravity, so the association reflects a change in skin quality rather than weight burden alone (Ezure & Amano, 2010; PMID: 20637003). While correlative rather than causal, these findings are consistent with a mechanistic framework linking adipocyte state to dermal structural characteristics.

Adipocyte hypertrophy is closely associated with inflammatory amplification. Adipocyte diameter correlates with inflammatory cytokine production (Skurk et al., 2007; PMID: 17164304). Adipocyte hypertrophy co-occurs with upregulated adipose tissue inflammatory gene expression (IL-6, TNF-α, MCP-1) in individuals with genetic predisposition to type 2 diabetes (Henninger et al., 2014; PMID: 25148116). Defined physiological transitions can shift adipocytes toward this hypertrophic, pro-inflammatory state: following menopause, women show adipocyte hypertrophy and fewer cells per unit mass, with increased adipose inflammation and fibrosis — and without a significant change in total fat mass, indicating enlargement of existing cells rather than net fat gain (Abildgaard et al., 2021; PMID: 34285301).

TNF-α has been shown to suppress adipogenic differentiation in human preadipocytes from individuals with obesity (Gustafson et al., 2009; PMID: 19622783), establishing a reinforcing cycle in which adipocyte hypertrophy promotes TNF-α production, TNF-α may reduce adipogenic renewal, and reduced renewal may increase stress on existing adipocytes. This inverse coupling between renewal and cell size has been described directly: across adipose depots, the rate at which new adipocytes are generated runs inversely to the enlargement of existing ones, such that impaired adipogenesis and hypertrophy reinforce one another (Iacobini et al., 2024; PMID: 38727299).

The basic biology of fat cell hypertrophy, inflammation, and suppression of adipogenesis has been established in cell systems — TNF-α blocks normal adipogenic differentiation in human preadipocytes via Wnt signaling.¹ While its magnitude may differ from depot to depot, this core relationship is not believed to be functionally different across adipose tissues.

¹ Gustafson B, Gogg S, Hedjazifar S, Jenndahl L, Hammarstedt A, Smith U. Inflammation and impaired adipogenesis in hypertrophic obesity in man. Am J Physiol Endocrinol Metab. 2009;297(5):E999–E1003. doi:10.1152/ajpendo.00377.2009.  PMID: 19622783.

Epidemiologic associations between elevated BMI and inflammatory skin diseases such as psoriasis and hidradenitis suppurativa further support a potential link between adipose inflammatory tone and cutaneous biology (Norder et al., 2022; PMID: 34118300; Kromann et al., 2014; PMID: 24577555).

Together, these data suggest that dermal adipocytes function as active contributors to cutaneous structural homeostasis and inflammatory balance.

Human adipose tissue remains regenerative throughout adult life. Using atmospheric radiocarbon dating to retrospectively birth-date adipocytes, Spalding et al. (2008; PMID: 18454136) estimated that approximately 10% of adipocytes in subcutaneous abdominal fat are renewed annually. In contrast, higher turnover estimates have been reported using a metabolic tracer approach: Strawford et al. (2004; PMID: 14600072) measured adipose tissue triglyceride turnover, de novo lipogenesis, and adipocyte proliferation in humans through incorporation of deuterium from ²H₂O, suggesting a more dynamic rate of adipose tissue remodeling.

Adipogenesis is a multistep process involving stem cell commitment, pre-adipocyte proliferation, and PPARγ-driven differentiation into mature adipocytes (Rosen & Spiegelman, 2014; PMID: 24439368). TNF-α inhibits PPARγ activity and suppresses adipogenic differentiation (Cawthorn & Sethi, 2008; PMID: 18037376). IL-6 has been separately shown to impair subcutaneous adipogenesis in preadipocytes from individuals with obesity and insulin resistance, via reduction in PPARγ and C/EBPα expression (Almuraikhy et al., 2016; PMID: 27342408). Thus, adipogenesis appears to remain active but context-dependent, and chronic inflammatory signaling may reduce adipogenic efficiency without eliminating regenerative capacity.

Direct measurements in humans show that this efficiency declines with age. Subcutaneous preadipocytes isolated from older adults replicate and differentiate more slowly than those from younger adults — differentiation roughly 25% lower and independent of donor body fat — alongside greater TNF-α release from their fat tissue, linking the decline to inflammatory tone (Caso et al., 2013; PMID: 22999012). In skin specifically, dermal adipose progenitors lose proliferative and differentiation capacity with age, while an accumulating population of inflammatory regulatory cells suppresses new adipocyte formation (Liu et al., 2024; PMID: 38914551). These findings indicate that adipogenic renewal persists but slows with age, consistent with the reinforcing cycle described in the previous section.

In murine single-cell lineage-tracing and transplantation models, dermal adipocyte precursors showed substantially greater adipogenic commitment than inguinal subcutaneous precursors: skin-associated progenitors gave rise to adipocytes over shorter timeframes and at a markedly higher rate, consistent with a more active adipogenic niche in dWAT relative to deeper subcutaneous depots (Rivera-Gonzalez et al., 2025; PMID: 40744015).

In human in vitro systems, primary facial preadipocytes exhibit greater adipogenic differentiation capacity than abdominal subcutaneous cells under standardized conditions (Chon & Pappas, 2015; PMID: 26167398). These findings address intrinsic differentiation potential rather than direct in vivo turnover kinetics. Together, they indicate that adipogenic capacity varies by anatomical location, although direct measurement of adipocyte replacement rates in human dermal or facial adipose tissue remains unavailable.

Importantly, adipogenic capacity has also been shown to be biologically modifiable under defined pharmacologic conditions. Synthetic PPARγ ligands such as thiazolidinediones promote adipogenic differentiation in preadipocyte and stem cell models, establishing PPARγ as their molecular target for adipogenesis (Lehmann et al., 1995; PMID: 7768881; Tontonoz & Spiegelman, 2008; PMID: 18518822). A randomized controlled trial showed increased adipogenesis in femoral — but not abdominal — subcutaneous adipose tissue after 16 weeks of pioglitazone treatment in human adipose progenitor cells from individuals with type 2 diabetes (Stafeev et al., 2024; PMID: 38493915), and pioglitazone reduced pro-inflammatory gene expression in epicardial adipose tissue in women with obesity (White et al., 2021; PMID: 33001232). This depot-specificity is consistent with the broader evidence of anatomically distinct adipogenic capacities described in this section. GLP-1 receptor agonists have been shown to restore adipogenic differentiation in epicardial adipose tissue in individuals with cardiovascular disease (García-Vega et al., 2024; PMID: 38172989). These findings derive from systemic pharmacologic interventions under clinical or experimental conditions and do not imply equivalence with topical approaches. However, they demonstrate that adipose inflammatory tone and adipogenic competence remain modifiable under defined inflammatory and pharmacologic conditions.

Collectively, available evidence indicates that adipogenesis persists into adulthood, varies by anatomical depot, and remains biologically modifiable under appropriate conditions.

In animal studies, diets that keep fat cells smaller go hand in hand with healthier dermal collagen, while high-fat diets that enlarge fat cells damage it. Exercise can go further: in mice, it has reversed collagen damage that obesity had already caused. And in people, exercise improves skin elasticity and structure — and can make fat cells smaller even when total body weight doesn’t change. None of this yet proves that shrinking fat cells reverses skin aging in humans, but it shows the same biology responds to ordinary, non-pharmacologic inputs.

The previous section established that adipogenic capacity and adipocyte state are modifiable under pharmacologic conditions. A parallel question is whether non-pharmacologic inputs — diet and physical activity — act on the same axis, and whether their effects reach the skin. The evidence is strongest in animal models, with emerging human signals: it supports prevention, demonstrates reversal in one animal model, and stops short of demonstrating reversal in human skin.

The clearest reversal evidence comes from an animal model. In mice with diet-induced obesity, voluntary exercise reduced adipocyte size and was accompanied by recovery of previously degraded dermal collagen structure, which the authors framed explicitly as reversal of connective-tissue deterioration (Lőrincz et al., 2017; PMID: 28180933). This is reversal of established damage rather than its prevention — though in a mouse, and through systemic exercise rather than an isolated reduction in adipocyte size.

Diet shapes adipocyte size and dermal structure in parallel, and the relationship here is one of prevention rather than treatment: in these studies the diet is present throughout, shaping the trajectory of the skin rather than reversing damage already established. In a large survey of aging mice spanning 25 diets, adipocyte size correlated inversely with dermal structure, and the diet producing the smallest adipocytes showed the least dermal attrition (Hew et al., 2016; PMID: 27832138).

Imaging work makes the link more direct and also marks its boundary. In mice kept on different diets for 32 weeks, body weight correlated strongly with adipocyte size, and the dermal collagen signal measured by second-harmonic-generation microscopy fell as body weight rose; high-fat-fed mice developed both enlarged adipocytes and degraded dermal collagen — a diet → adipocyte hypertrophy → collagen-degradation sequence observed directly in skin (Haluszka et al., 2016; PMID: 27895989; DOI: 10.1364/BOE.7.004480). The same study marks where that sequence ends: mice on a high-fructose diet that gained no weight showed no adipocyte enlargement, yet still developed collagen damage — through glycation, a pathway independent of fat-cell size. Diet therefore reaches the dermis by more than one route, and adipocyte hypertrophy is a demonstrated contributor rather than the only one. (The collagen readout here reflects fibre density and organization rather than total collagen mass.)

In humans, the evidence is younger but points the same way. A randomized study of resistance and aerobic training in middle-aged women found that both improved skin elasticity and dermal structure, with resistance training also increasing dermal thickness, accompanied by reductions in circulating inflammatory factors and increased expression of dermal matrix genes (Nishikori et al., 2023; DOI: 10.1038/s41598-023-37207-9). The resistance-training benefit occurred without a measured decrease in body fat. Stable total adiposity does not preclude a shift within the fat tissue itself — from fewer, larger cells toward more, smaller ones — and this kind of shift has been shown directly in humans: exercise training reduced subcutaneous adipocyte size in adults with obesity without accompanying weight loss (Ahn et al., 2022; PMID: 35249225). The anti-inflammatory changes seen with training are consistent with healthier adipocyte signaling, though the skin study did not measure adipocyte size, and its demonstrated mechanism runs through circulating factors rather than through the fat cell directly.

Taken together, these findings show that the adipocyte–dermis relationship responds to ordinary, non-pharmacologic inputs: diet shapes it, exercise reverses obesity-related skin damage in animal models, and human training improves skin while independently reducing fat-cell size. What this body of work does not yet establish is the specific link the rest of this page is careful to bound — that reducing adipocyte size restores the dermal matrix in living human skin. That gap, and the research aimed at it, is taken up in the sections that follow.

Given that hypertrophic adipocytes secrete pro-inflammatory mediators associated with suppression of adipogenesis and altered fibroblast function, a biologically coherent hypothesis is that modulation of adipocyte signaling could influence this balance. Botanical compounds influence adipocyte pathways including AMPK activation, PPARγ transcriptional regulation, lipase activation, and mitochondrial metabolism. Importantly, activation of shared upstream pathways such as AMPK does not uniformly determine adipogenic outcome.

Kaempferia parviflora Extract (KPE)

Methoxyflavones derived from Kaempferia parviflora have been shown to promote adipogenic differentiation in 3T3-L1 preadipocytes (mouse) via upregulation of PPARγ and C/EBPα transcription factors (Horikawa et al., 2012; PMID: 22687402).

In an obese mouse model, oral Kaempferia parviflora extract given alongside the development of obesity helped preserve a smaller adipocyte size distribution and better-retained dermal collagen structure under metabolic and UVB stress (Hidaka et al., 2016; PMID: 27592007). This is a preventive effect under systemic administration. It is not evidence that the extract reverses existing damage, nor that a topical application would reproduce it.

Human studies indicate systemic metabolic activity of KPE constituents (Matsushita et al., 2015; PMID: 25994142), supporting biological relevance beyond isolated cell systems.

Caffeine and Resveratrol

Caffeine has been shown to suppress PPARγ expression and inhibit adipogenic differentiation in preadipocyte models via modulation of the AKT/GSK3 pathway (Kim et al., 2016; PMID: 26350746). Resveratrol activates SIRT1 and inhibits adipogenic differentiation via suppression of PPARγ transcriptional activity (Fischer-Posovszky et al., 2010; PMID: 20463039). These comparisons suggest that lipolytic activity or AMPK activation alone does not necessarily preserve adipogenic competence.

Oleic and Linoleic Acids

Oleic and linoleic acids function as weak PPARγ modulators in lineage-committed preadipocyte systems (Kliewer et al., 1997; PMID: 9113987). In vivo, their biological impact on adipogenesis is less understood.

For topical compounds to influence dermal adipose biology, access to deeper dermal compartments would be required. The pilosebaceous unit provides a recognized transfollicular pathway for small lipophilic molecules. Experimental studies demonstrate dermal permeation of KPE methoxyflavones (Tuntiyasawasdikul et al., 2014; PMID: 24789664), follicular localization of linoleic acid (Raufast & Mavon, 2006; PMID: 18492146), transfollicular penetration of caffeine (Otberg et al., 2007; PMID: 17396054), and measurable dermal delivery of resveratrol (Hung et al., 2008; PMID: 18451526). While penetration depth depends on formulation and exposure conditions, these data indicate physicochemical compatibility with dermal access under defined experimental conditions.

Most published claims about how ingredients affect fat cells are extrapolated from the 3T3-L1 mouse cell line or from systemic, oral or injected, drug studies. Neither speaks directly to the question that matters for a topical product: can a compound applied to human skin reach the dermal adipocytes and change their state? Adipeau’s research program is built to test that question directly. In collaboration with the laboratory of Professor Volker Lauschke at the Karolinska Institutet (Department of Physiology and Pharmacology and Center for Molecular Medicine, Stockholm), the work uses two human models the literature has rarely combined: three-dimensional primary human adipocyte spheroids, which preserve cell-to-cell signaling and tissue-like architecture, and full-thickness human skin explants, which keep the dermis and its associated adipose layer intact.

The first phase of this work describes a dual, cell-state-dependent mechanism (Taebnia et al., 2026). In preadipocytes, the active extract reconfigures the transcription-factor network that governs differentiation — reducing the activity of PPARγ, CEBPβ, and SREBP1 while increasing SOX9, KLF3, and GATA3 — and shifts the population toward more small cells rather than fewer large ones. In mature, enlarged adipocytes, the same extract reduces lipid accumulation and mean droplet diameter. The effect runs in two directions: it supports the formation and maintenance of small, healthy adipocytes where they are depleted, and it reduces lipid load where cells have become hypertrophic. That bidirectional behavior is the property the correlational literature implies a useful intervention would need, and it is difficult to observe in a single mouse cell line.

A separate body of ex vivo work tests finished Adipeau formulations, not just the isolated actives, on full-thickness human skin. Under twice-daily topical application, the formulations significantly reduced adipocyte lipid droplet diameter in the tissue, measured by PLIN2 immunohistochemistry. This is the most direct evidence in the program that a topically applied product can reach the dermal adipose layer and change adipocyte size in human skin, and it is the basis for the twice-daily use protocol.

These are human-relevant model systems, and the program is ongoing; co-culture work examining how adipocyte signaling influences fibroblast collagen output is in progress. What this research establishes is a mechanism and a measurable change in adipocyte size in human tissue. What it does not establish — the translation of that change into lasting improvement in the dermal matrix of living facial skin over months of use — is taken up, to the extent current data allow, in our Clinical Evidence section, and is bounded plainly in “What This Page Does Not Claim,” below.

The mechanism above is causal only at the level of cells and remains correlational in living skin, as we set out in detail in “What This Page Does Not Claim,” below. But even bounded that way, it generates a prediction. If hypertrophic dermal adipocytes weaken the surrounding matrix through the inflammatory signaling described here, then a modality that works against that hypertrophy — by reducing the size of enlarged cells, by supporting regeneration of small healthy ones, or both — should leave the skin structurally sounder. This follows from the facial fat fitness framework we proposed in the peer-reviewed literature (PMID 32914326): that the fitness of dermal fat cells, not their quantity, governs the fullness and the strength of the curves of the face.

By “strength” we mean contour integrity, and we want to be exact about it. We mean the smooth, continuous convex line of the face holding its shape: the arc from the under-eye through the cheek to the jawline, and the transition from the cheek to the mouth, without dips, bulges, or abrupt breaks. That is the whole of what we mean when we say a modality should “strengthen” the skin. We do not mean it has been shown to raise collagen, to restore the mechanical strength of the dermal matrix, or to reverse elastosis. Contour is shaped by more than the matrix, including the volume and distribution of the fat itself, so an improvement in contour is a signal consistent with a sounder structure, not a measurement of one. We are naming what can be seen and being plain about what it does and does not stand for.

This also gives a fair way to judge any modality that claims to act through dermal fat. The question is not whether it changes a fat cell in a dish. It is whether contour integrity improves, judged against an endpoint defined before the study begins, on a scale that can register no change or a worse result as readily as a better one. A test a modality cannot fail is not a test.