Deciphering the cellular landscape of pathological scars

Abstract

Pathological scars (PSs) are characterized by excess dermal collagen deposition and primarily manifest as hypertrophic scars and keloids. The microenvironment of PSs is shaped by a diverse array of cells, collectively forming a profibrotic and proangiogenic niche. Recent advances in high-throughput sequencing have enhanced our understanding of the cellular heterogeneity, transcriptional changes, and intercellular communication within the PS microenvironment. This review comprehensively describes the cellular landscape of PSs, including mesenchymal cells, endothelial cells, epithelial cells, and immune cells. Additionally, we evaluate precision medicine strategies ranging from biomarker-driven diagnostics to targeted biological therapies and innovative drug delivery systems. We aim to provide a mechanistic framework that accelerates the translation of basic experimental research into effective and personalized treatments for pathological scarring.

Highlights

• Single-cell ribonucleic acid sequencing and spatial transcriptomics redefine pathological scars (PSs) as spatially organized, multicellular fibrotic ecosystems.

• Mesenchymal, immune, vascular, epithelial, and neural cell populations interact through coordinated signaling networks to sustain fibrosis, inflammation, angiogenesis, and scar progression.

• Emerging biomarkers, targeted biological therapies, and innovative delivery systems provide a framework for more precise and individualized management of PSs.

Background

Pathological scars (PSs) are common fibrotic disorders of the skin caused by abnormal wound healing processes [1]. PSs include hypertrophic scars (HTSs) and keloids, which are characterized by enlarged, protrusive, and reddish skin lesions accompanied by marked itching and pain. These lesions not only compromise esthetic appearance but also impose a significant physical burden, severely impairing patients’ quality of life. Historically, an understanding of PS pathogenesis has been constructed based on tissue-level observations and bulk molecular analyses. Extensive research has revealed six hallmark features (Figure 1): fibrosis-promoting inflammation [2], epigenetic reprogramming [3], persistent fibrotic signaling [e.g. hyperactivation of transforming growth factor β (TGF-β)/small mothers against decapentaplegic (Smad)] [4], induction of angiogenesis [5], dysregulation of cellular energetics [6], and activation of invasion [7]. These macroscopic biological behaviors reflect intricate interactions at the cellular and molecular levels within the PS microenvironment, including cellular heterogeneity, dynamic molecular networks, and spatial coordination.

Figure 1.

Pathological scar hallmarks. ^*Activation of invasion occurs in keloids. Created at https://BioRender.com

However, these traditional approaches suffer from a fundamental resolution limit. Bulk ribonucleic acid (RNA) sequencing and proteomic studies treat scar tissue as a uniform mixture, averaging gene expression signals across millions of cells. This averaging conceals the critical roles of rare cell subpopulations and obscures the intricate intercellular communication networks that drive fibrosis. The advent of single-cell RNA sequencing (scRNA-seq) and spatial transcriptomics (ST) has overcome these limitations, enabling a paradigm shift in scar research from tissue-level description to high-resolution cellular mapping [8].

Current clinical management for PSs relies heavily on surgical excision [9], intralesional corticosteroids [10], laser therapy [11, 12], and radiotherapy [13, 14]. However, the efficacy of these conventional modalities is often limited. Surgical excision alone is associated with recurrence rates approaching 100%, thus necessitating adjuvant postoperative radiotherapy, albeit with inherent risks of complications such as hyperpigmentation [9]. Intralesional injections and laser therapy are suitable for small PSs but are accompanied by treatment-related pain and potential recurrence. The persistent recalcitrance of PSs to these treatments highlights a critical gap: current therapies fail to target the true drivers of the disease. PSs are not merely collections of overactive fibroblasts but also complex, self-sustaining ecosystems composed of heterogeneous mesenchymal, immune, vascular, and epithelial components.

The concept of cellular heterogeneity is central to this new understanding. Recent data reveal that not all fibroblasts are profibrotic. One of the fibroblast subtypes, mesenchymal fibroblasts (MFs), serves as the central architect of the scar niche [15], while others may play regulatory roles. Similarly, the immune microenvironment is not just inflamed but also polarized into specific macrophage phenotypes that stimulate fibrosis signaling [16]. By mapping this cellular landscape, we move beyond broad anti-inflammatory or antiproliferative strategies toward identifying precise therapeutic targets that can disrupt the fibrotic niche while minimizing the compromise of normal tissue homeostasis.

This review aims to integrate recent single-cell breakthroughs with histopathological and functional studies to construct a comprehensive cellular atlas of PSs. In this framework, the mesenchymal ecosystem refers to the stromal and matrix-remodeling core, which includes fibroblasts, mesenchymal stem cell (MSC)-like progenitors, and cells in a transitional mesenchymal state. The epithelial compartment includes mainly keratinocytes and melanocytes. The endothelial system comprises vascular and lymphatic endothelial cells (ECs), and the immune system encompasses innate and adaptive immune cell populations. Moving beyond isolated signaling pathways, we characterize the scar microenvironment as a dynamic ecosystem. Our objective is to elucidate the spatiotemporal coordination driving fibrosis and critically evaluate how this high-resolution landscape accelerates the transition toward precision medicine.

Review

Methodological foundations and technological advances

Single-cell ribonucleic acid sequencing

Although scRNA-seq has significantly advanced our understanding of pathological scarring, its application to this specific tissue type presents unique technical challenges [15]. As PSs are characterized by an exceptionally dense collagen matrix, they require prolonged enzymatic digestion to achieve cellular dissociation [15, 17]. This extended processing can inadvertently trigger cell stress and change gene expression profiles [17]. Furthermore, the mechanical and chemical rigors of the workflow can lead to the loss of more fragile cell populations [18]. Single-nucleus sequencing (snRNA-seq) offers a way to avoid harsh dissociation treatments; however, it excludes cytoplasmic RNA, thereby limiting insights into posttranscriptional regulation [17–19].

Researchers must also carefully consider the trade-offs between droplet- and plate-based sequencing platforms, as this choice impacts data quality. Droplet-based methods, such as 10× Genomics and Drop-seq, enable large-scale cellular atlasing and are therefore well suited for defining broad cellular heterogeneity, comparing scar subtypes, and identifying major disease-associated cell states. In contrast, plate-based approaches, such as Smart-seq2, provide deeper transcript coverage per cell and may be preferable when the research goal is to characterize selected or rare populations in greater molecular detail, although they are more costly and less scalable [20, 21]. No single method is optimal for every goal. Careful selection and rigorous validation are crucial for advancing research objectives [17].

Rigorous quality control is essential for ensuring the reliability of scRNA-seq findings. Key considerations include monitoring dissociation-induced stress, doublets, batch effects, cell recovery statistics, and reproducibility across samples. To improve transparency and comparability, experimental protocols, analytical parameters, and codes should be clearly documented and, when possible, made publicly available. In this context, Table 1 provides a structured comparison of representative primary scRNA-seq studies, summarizing their sample composition, sequencing platforms, and major cellular and molecular findings to help readers assess the strength and scope of current single-cell evidence.

Table 1.

Comprehensive single-cell studies comparison. Only original studies that collected new clinical samples and performed primary single-cell RNA sequencing analysis are included

Study	Platform	Sample detail	Cellular and molecular findings
Deng et al., 2021 [15]	10× Genomics	3 keloids 3 normal scars	• Expanded MFs: crucial for collagen overexpression in keloids. • Specific markers: MFs exhibit a unique CD266+/CD9- profile and overexpress osteogenic genes. • Fibrotic signaling: MFs drive surrounding collagen synthesis via POSTN-integrin interactions and enhanced TGF-β1 signaling.
Vorstandlechner et al., 2021 [8]	10× Genomics	3 HTSs 3 normal skins mouse scar models (normal skin, 6 weeks scar and 8 weeks scar)	• Serine proteases as key drivers: DPP4 and PLAU are critical mediators driving TGFβ1-induced myofibroblast differentiation and ECM overproduction.
Direder et al., 2022 [16]	10× Genomics	• 4 keloids • 1 normal skin	• Novel Schwann cell Population: discovery of abundant, axon-independent SCs exhibiting a repair phenotype (NES+, SOX10+). • Pro-fibrotic crosstalk: SCs secrete factors (TNFAIP6, CCN3) driving stromal accumulation of M2-polarized macrophages.
Shim et al., 2022 [20]	10× Genomics	• scRNA-seq: 2 keloids • ST: 2 keloids	• Fibrovascular spatial niche: spatial transcriptomics reveals disease-associated fibroblasts physically co-localize with ECs in deeper regions. • Endothelial transition: keloid ECs undergo mesenchymal activation driven by dysregulated TGF-β/Smad signaling.
Liu et al., 2022 [22]	10× Genomics	• 4 keloids • 4 adjacent normal skins	• Key regulators: identified TWIST1, FOXO3, and SMAD3 as critical regulators of fibroblast fibrogenesis. • Pathogenic signaling: overactive TGF-β and eph-ephrin signaling pathways drive simultaneous fibrogenesis and angiogenesis.
Xu et al., 2022 [23]	10× Genomics	PBMCs from 2 keloids, 2 healthy donors	• CTL downregulation: identified a significant reduction of cytotoxic CD8+ T cells (CTLs) as a key immune signature in peripheral blood and local lesions in keloid patients. • NKG2A-sHLA-E axis: upregulation of the NKG2A/CD94 complex on CTLs interacts with high serum sHLA-E, leading to CTL exhaustion.
Yeo et al., 2024 [24]	10× Genomics	• 6 active keloids • 4 inactive keloids • 3 mature scars	• Neuro-immune enrichment: notable enrichment of Mast Cells (MCs) and Neural Cells (NCs) in active keloids, positively correlating with disease activity. • Neuro-immune crosstalk: intricate cell–cell communication between MCs and NCs synergistically drives fibrotic progression and sensory symptoms.
Cheng et al., 2024 [25]	10× Genomics	• 1 keloid and 1 HTS derived from the same patient	• Mechanoresponsive fibroblasts: the dominant fibroblast subgroup in keloids is highly mechanoresponsive, showing enhanced mechanotransduction and migration capacity that confers aggressive growth. • Mesenchymal transitions: both ECs and keratinocytes in keloids actively undergo EndoMT and EMT transitions.
Oh et al., 2024 [26]	10× Genomics	• 3 active keloids (center and periphery), • 3 inactive keloids (center and periphery), • 3 mature scars • 1 normal skin	• Activity-dependent fibroblasts: pro-inflammatory fibroblasts are significantly increased in active keloids compared to inactive ones. • Spatiotemporal EndoMT: VECs at the active periphery are immature/tip cells, while VECs in the active center undergo profound mesenchymal activation.
Shi et al., 2025 [27]	10× Genomics	• 6 keloids and 6 matched adjacent normal skins	• Melanocyte-driven fibrogenesis: overactive pigmentation pathways in keloid melanocytes drive excessive melanin secretion. • Iron overload and ferroptosis resistance: increased melanin induces iron overload in neighboring fibroblasts, rendering them resistant to ferroptosis.
Akita et al., 2025 [28]	10× Genomics	• 3 keloids • 1 normal skin	• Mechanosensing upregulation: significantly increased expression of the PIEZO2 mechanoreceptor, specifically localized to vascular/lymphatic ECs and a subgroup of fibroblasts, converting physical tension into fibroproliferative signals.
Zhao et al., 2025 [29]	10× Genomics	• 3 keloids • 3 normal skins	• Pro-fibrotic POSTN+ subpopulation: confirmed POSTN+ MFs are highly enriched in keloids. • Anti-fibrotic IGFBP2+ subpopulation: discovered an abundant IGFBP2+ fibroblast population in normal skin that is insensitive to TGF-β/Periostin and possesses anti-fibrotic potential. • EndoMT crosstalk: identified an endothelial cell subpopulation exhibiting mesenchymal activation that synergizes with fibroblasts to drive fibrosis.

HTS hypertrophic scar, MFs mesenchymal fibroblast, ECs endothelial cells, AMT adipocyte-to-mesenchymal transition, CCN3 cellular communication network factor 3, CTL cytotoxic T lymphocyte, EC endothelial cell, ECM extracellular matrix, EMT epithelial-to-mesenchymal transition, EndoMT endothelial-to-mesenchymal transition, GPX4 glutathione peroxidase 4, HTS hypertrophic scar, IGFBP2 insulin-like growth factor-binding protein 2, MC mast cell, MF mesenchymal fibroblast, NC neural cell, NES nestin, NKG2A natural killer group 2 member A, PBMC peripheral blood mononuclear cell, PIEZO2 piezo-type mechanosensitive ion channel component 2, PLAU urokinase-type plasminogen activator, SC Schwann cell, sHLA-E soluble human leukocyte antigen E, ST spatial transcriptomics, TGF-β transforming growth factor beta, TNFAIP6 tumor necrosis factor alpha-induced protein 6, VEC vascular endothelial cell

Spatial transcriptomics

ScRNA-seq is a powerful investigational tool; however, it inherently loses the spatial context of the tissue. Advances in ST have successfully bridged this gap, demonstrating that cellular function is regulated by precise anatomical location [20, 30, 31]. Therefore, ST represents a critical frontier in PS research, as it enables the investigation of anatomical niches, proximity-dependent signaling, and spatial lesion organization.

Currently, ST has developed diverse technical platforms, each with distinct trade-offs. Array-based methods provide expansive whole-genome coverage but often lack true single-cell resolution; conversely, imaging-based tools offer subcellular precision but are typically limited to a preselected gene panel [32–34]. Consequently, platform selection must be meticulously aligned with specific research goals, although integrating multiple strategies often yields the most robust insights [17, 20]. Notably, when dense collagenous tissues are analyzed, rigorous protocol optimization remains essential to ensure data reliability [18, 20].

Validation strategies

Experimental validation is needed to support computational discoveries. Confirming cell locations and marker expression is essential [20]. Multiplexed immunofluorescence and RNA in situ hybridization can be used to visualize cell types within intact tissue [35–37]. Any discrepancies between predicted cell abundance and their actual physical locations must be thoroughly resolved to ensure the integrity of the findings [20, 38]. To move beyond descriptive correlations, validation must integrate functional assays that establish mechanistic causality [38, 39]. Patient-derived three-dimensional cultures and animal models are valuable for testing cell functions, and targeted perturbations can confirm complex cellular interactions [30, 39, 40]. Employing these advanced tools strengthens the robustness of the experimental findings. Finally, navigating the inherent biological and technical noise in single-cell data is paramount for reproducible research [41, 42]. High-impact studies increasingly rely on integration with standardized reference datasets, transparent benchmarking, and open-source ecosystems (Figure 2) [41, 42].

Figure 2.

Conceptual framework. A synergetic workflow utilizing scRNA-seq and spatial transcriptomics combined with traditional experimental platforms to investigate pathological scars. Created at https://BioRender.com

The mesenchymal ecosystem: key drivers of fibrosis

Heterogeneity and functional specialization of fibroblasts

The advent of scRNA-seq has facilitated the high-resolution clustering of dermal fibroblasts. Deng et al. classified fibroblasts in PSs into secretory–papillary, secretory–reticular, mesenchymal, and proinflammatory subgroups. Among these subclusters, a specific MFs subpopulation is markedly expanded in keloids compared with normal scars [15]. This mesenchymal subset is defined by a unique transcriptional signature, including the upregulation of periostin (POSTN), cartilage oligomeric matrix protein (COMP), and collagen type XI alpha 1 chain (COL11A1), as well as membrane proteins such as syndecan 1 (SDC1), a disintegrin and metalloproteinase 12 (ADAM12), and CD266. Gene Ontology (GO) enrichment analysis indicated the activation of pathways related to extracellular matrix (ECM) organization, cell adhesion, and chondrogenesis. Recent investigations employing paired keloid and HTS samples have further characterized them as a dominant mechanoresponsive subgroup [25]. The results of functional assays further demonstrated that the secretome of these MFs is sufficient to induce collagen synthesis in quiescent fibroblasts and that this effect can be attenuated by anti-POSTN antibodies [15]. The enrichment of MFs in PSs was widely confirmed in later scRNA-seq studies [20, 29, 43]. ScRNA-seq analysis of scleroderma also revealed an increase in MFs in lesion tissues, which demonstrated that MFs play a crucial role in skin fibrosis [15].

Myofibroblasts: apoptosis resistance and metabolic reprogramming

Myofibroblasts act as the primary effector cells driving contracture and excessive ECM deposition [44]. In physiological wound healing, these cells undergo apoptosis upon tissue re-epithelialization [45]. However, in PSs, they exhibit aberrant longevity and resistance to apoptosis, potentially sustained by mechanical tension [46, 47] or a chronic inflammatory niche [48, 49]. Immunohistochemical evidence confirms the dense accumulation of α-smooth muscle actin (α-SMA) cells in keloid and HTSs [50, 51], whereas these cells are rare in normal mature scars. The overactivation of myofibroblasts and the consequent excessive accumulation of ECM lead to PS formation. The increased number of myofibroblasts in keloids may partly originate from endothelial-to-mesenchymal transition (EndoMT) [52] or epithelial-to-mesenchymal transition (EMT) [53], whereas the mainstream view is that their main source is the resident fibroblasts in the wound [54, 55]. Crucially, recent investigations have revealed that this pathological persistence is related to metabolic reprogramming toward aerobic glycolysis (the Warburg effect) [56–58]. Key regulatory nodes, such as acyl-CoA synthetase short chain family member 3 (ACSS3) and vestigial like family member 3 (VGLL3), have been implicated in coupling wingless-related integration site (Wnt)/β-catenin signaling to glycolytic flux, providing the requisite adenosine triphosphate (ATP) and biosynthetic intermediates to sustain continuous collagen production and cell survival under hypoxic conditions [59, 60]. Beyond ECM synthesis, keloid myofibroblasts exhibit an invasive phenotype similar to that of benign tumors, a behavior potentially driven by the overexpression of specific cytoskeletal regulators such as the KN motif and ankyrin repeat domain-containing protein 4 (KANK4) [61].

Schwann cells: the neural–fibrotic interface

Emerging evidence places the peripheral nervous system as an integral component of the scar microenvironment [16, 62, 63]. A comparative scRNA-seq study revealed a keloid-associated Schwann cell state that was expanded in the analyzed keloid samples relative to control skin [16]. Morphologically, these cells possess a spindle-shaped architecture resembling the repair Schwann cells observed during nerve regeneration [16, 64–66]. They are molecularly defined by the expression of nestin (NES), elastin, cellular communication network factor 3 (CCN3), and insulin-like growth factor-binding protein 5 (IGFBP5) [16]. Far from being passive structural supports, these glial cells actively modulate the microenvironment by upregulating the expression of genes associated with the production of ECM, such as collagen type I alpha 1 chain (COL1A1) and collagen type III alpha 1 chain (COL3A1), and the secretion of neurotrophic factors [e.g. semaphorin 3C (SEMA3C)] [67] that promote fibroblast proliferation. Furthermore, keloidal Schwann cells contribute to immune dysregulation through the secretion of tumor necrosis factor alpha-induced protein 6 (TNFAIP6) and CCN3, which are known to drive macrophage polarization toward a profibrotic M2 phenotype [16]. Directly targeting the neural component via botulinum toxin type A (BoNT-A) [68] offers a mechanism-based strategy to alleviate symptoms by disrupting this pathological neuro-fibrotic crosstalk.

Mesenchymal stem cells and cellular plasticity

MSCs are progenitor cells that can self-renew and undergo multipotent differentiation. The maintenance of the fibrotic mass requires continuous replenishment of the mesenchymal pool. Progenitor populations identified in keloids express classical MSC markers (CD73, CD90, and CD105) and possess multipotent differentiation potential [69]. However, under the influence of inflammatory cytokines such as interleukin-6 (IL-6), these progenitors are biased toward a fibrogenic lineage [70]. This mesenchymal population is further augmented by cellular plasticity mechanisms, particularly EndoMT and EMT [52, 71]. Recently, keloid fibroblasts have been successfully reprogrammed into induced pluripotent stem cells (iPSCs) in vitro, providing a novel platform for dissecting the plasticity of keloid cells and exploring patient-specific strategies for treating skin fibrosis [72].

Vascular and lymphatic networks: the circulation-fibrosis nexus

Heterogeneity and angiogenic dysregulation in endothelial cells

Vascular endothelial cells (VECs) in PSs are not merely passive structural components but also active drivers of disease progression [73], exhibiting profound heterogeneity and functional dysregulation. VEC-derived paracrine mediators further contribute to this pathogenic role. Previous studies have shown that VECs can produce proangiogenic and profibrotic factors, including vascular endothelial growth factor (VEGF), TGF-β, platelet-derived growth factor (PDGF), basic fibroblast growth factor (bFGF), and endothelin-1 (ET-1), which may collectively promote angiogenesis, fibroblast activation, and ECM remodeling [74, 75]. ScRNA-seq and quantitative histological analyses have consistently demonstrated a significant expansion of the endothelial compartment in keloids compared with normal skin [15, 20, 22]. Specifically, scRNA-seq in keloids has identified heterogeneous endothelial subpopulations, including a heme oxygenase 1 (HMOX1)-^high subset, an atypical chemokine receptor 1 (ACKR1)-^high subset, a C-X-C motif chemokine ligand 12 (CXCL12)-^high subset and a CXCL3⁺ subset [22]. This aberrant vascular expansion is potentially molecularly sustained by the activation of the Eph-ephrin signaling axis and vascular endothelial growth factor receptor (VEGFR) pathways, which drive active sprouting angiogenesis [22]. However, this hypervascularity presents a pathological paradox in that the newly formed vessels are structurally immature and functionally incompetent. As PSs progress, the continuous accumulation of ECM ensues, thereby increasing tissue stiffness and tension. Consequently, microvascular lumens are narrowed or even occluded, which predominantly occurs in the central regions of PSs [76–79]. This structural collapse engenders a chronic hypoxic microenvironment that stabilizes hypoxia-inducible factor 1α (HIF-1α). Acting as a master regulator, HIF-1α orchestrates a maladaptive response by promoting reactive oxygen species (ROS) generation and activating downstream profibrotic cascades, including the TGF-β/Smad and toll-like receptor 4/nuclear factor kappa-light-chain-enhancer of activated B cells (TLR4/NF-κB) pathways [80–83]. This establishes a vicious cycle in which endothelial dysfunction fuels fibroblast-mediated collagen synthesis, which in turn further compromises vascular perfusion.

Macroscopically, this vascular dysfunction manifests as a distinct spatial heterogeneity reminiscent of solid tumors. Although the overall blood perfusion is elevated [74, 84, 85], it is unevenly distributed: the peripheral margins of keloids display high vascular density and active perfusion, supporting lateral invasion [77, 86], whereas the central core is predominantly characterized by vascular collapse and ischemia due to mechanical compression. Moreover, compared with deeper regions, superficial dermal layers exhibit greater vascular density [84]. This specific “ischemic center, hypervascular periphery” architecture not only sustains the metabolic demands of the expanding lesion but also creates a gradient of hypoxia that continuously drives the aggressive, invasive behavior of PSs.

Endothelial-to-mesenchymal transition

Beyond resident fibroblast proliferation, accumulating evidence supports EndoMT as a critical, albeit likely supplementary, source of the myofibroblast pool in pathological scarring. During EndoMT, VECs lose apical–basal polarity and endothelial identity, downregulating the expression of endothelial markers while upregulating the expression of mesenchymal markers such as fibroblast-specific protein 1 (FSP-1), α-SMA, vimentin (VIM) and N-cadherin. This transition ultimately leads to alterations in the cytoskeletal structure of VECs, leading to a loss of adhesion to neighboring cells and subsequent acquisition of invasive and migratory properties [87]. Some immunofluorescence studies have demonstrated the colocalization of endothelial and mesenchymal markers within the same cell in keloid tissues, providing static evidence of this transition [20, 88]. Transcriptomic data further support these findings by identifying intermediate cell clusters that express a dual gene signature [20]. However, it is crucial to interpret these findings with caution. Although colocalization suggests EndoMT, definitive proof requires genetic lineage tracing, which is challenging to perform in human clinical samples. Therefore, while EndoMT is likely to contribute to fibrosis, its quantitative contribution to resident fibroblast proliferation remains to be definitively established.

Lymphatic endothelial cell alterations and potential plasticity

Although lymphatic endothelial cells (LECs) represent a rare population within the keloid microenvironment, scRNA-seq analysis identifies them as a biologically active source of TGF-β1, indicating that they are involved in the upstream regulation of fibrogenesis rather than merely serving as passive drainage conduits [22]. The functional maladaptation of this lineage is further hypothesized to involve the lymphatic-endothelial-to-mesenchymal transition (Ly-EndoMT). In analogous fibrotic pathologies such as systemic sclerosis, this transition is well documented and characterized by the colocalization of the lymphatic marker lymphatic vessel endothelial hyaluronan receptor 1 (LYVE-1) and the mesenchymal marker α-SMA and is inducible in vitro by profibrotic stimuli [89]. These findings suggest that LECs in PSs may similarly abandon their homeostatic roles in fluid balance and immune trafficking to directly fuel the myofibroblast pool; however, definitive histological or lineage-tracing evidence within the keloid niche remains lacking. Consequently, unlike the well-characterized vascular component, the lymphatic contribution to scar pathogenesis represents a critical knowledge gap, precluding the current establishment of lymphangiogenesis-targeted therapeutics.

Epithelial components and barrier function

Keratinocyte activation and epithelial–mesenchymal crosstalk

Beyond the dermal pathology, the epidermis in PSs displays a distinct hyperactive transcriptional state [30, 90], characterized by the upregulation of both proliferation and terminal differentiation markers [20]. This reprogramming is related to aberrant Notch signaling [91]. Emerging evidence suggests that keratinocytes in PSs undergo partial EMT, with significantly elevated expression levels of EMT-related genes and increased motility capacity. While definitive evidence for complete lineage conversion is lacking, these cells functionally downregulate adhesion proteins (e.g. E-cadherin) and upregulate mesenchymal markers (e.g. VIM) via the Janus kinase/signal transducer and activator of transcription (JAK/STAT) axis in keloids [92]. The reduced epidermal hydration status in HTSs partially contributes to EMT through the downregulation of caveolin-1 expression [93]. Crucially, this epithelial plasticity fuels pathogenic epithelial–mesenchymal crosstalk. Activated keratinocytes release proinflammatory proteins such as IL-18, high mobility group box 1 (HMGB1) and S100 calcium binding protein A2 (S100A2), which directly stimulate fibroblast collagen synthesis via receptors for advanced glycation end products (RAGE) and Toll-like receptor 2/4 (TLR2/4) [94, 95]. Recent scRNA-seq analysis further identified expanded FOS-like antigen 1 (FOSL1)-positive keratinocyte subpopulations with partial EMT features in both HTSs and keloids. These cells promote fibroblast activation and inflammatory signaling through matrix metallopeptidase 3 (MMP3) [96]. Reciprocally, fibroblasts reinforce epidermal activation by secreting keratinocyte growth factor (KGF) and exosomes, thereby establishing a self-perpetuating epithelial–mesenchymal inflammatory loop. In keloids, fibroblast-derived KGF has been shown to induce keratinocytes to secrete oncostatin M (OSM), which in turn promotes fibroblast activation through STAT3 signaling, representing a double-paracrine mechanism that sustains cutaneous fibrosis [97, 98].

Melanocyte contributions and pigmentation abnormalities

The incidence ratio of keloids between darker-skinned people and white individuals ranges from 5:1 to 15:1. There have been no reported cases of PSs in patients with albinism of any race. Thus, the activity of melanocytes is concluded to be related to the formation of keloids. When a wound occurs, the basement membrane ruptures, allowing fibroblasts to migrate to the wound edge and communicate with melanocytes, thereby impacting the healing process. However, very few investigations have investigated the role of melanocytes in keloids. ScRNA-seq data revealed that the expression of CCN3, a protein related to M2 polarization of macrophages, was elevated in keloid melanocytes [16], indicating their potential involvement in immune regulation. Shen et al. further demonstrated that melanocyte-derived exosomes promote TGF-β/Smad signaling and ECM synthesis in fibroblasts, partly through exosomal miR-7704-mediated inhibition of Smurf1 [99, 100]. In addition, α-melanocyte-stimulating hormone (α-MSH), produced by dermal melanocytes, can increase the secretion of TGF-β [99]. Recent single-cell evidence further suggests that keloid melanocytes exhibit enhanced pigmentation-related programs. Increased melanocyte-derived melanin promotes fibroblast proliferation, migration, and collagen synthesis by inducing iron overload and ferroptosis resistance [27]. This melanocyte–fibroblast axis not only provides a molecular basis for the racial predisposition to scarring but also suggests that targeting pigment-related signaling pathways could offer novel therapeutic avenues for high-risk populations.

Immune microenvironment: inflammation and resolution

Macrophage polarization and tissue remodeling

Macrophages are derived from circulating monocytes and are activated by the local microenvironment during inflammation. Macrophages are responsible for phagocytosis, antigen presentation and cytokine secretion and thereby play essential roles in tissue repair and immunological response [101, 102]. Recent studies revealed that the number of macrophages in PSs was significantly increased and that their dysfunction was closely related to pathological scarring [103–105].

Macrophages exhibit phenotypic plasticity, polarizing into proinflammatory M1 and anti-inflammatory M2 states [103]. Ideally, the relative abundances of M1 and M2 macrophages vary throughout the progression of wound healing, at which time they reach an adaptive balance. Ultimately, both types of macrophages will decrease to baseline levels. In the early phase of wound healing, an elevated M1/M2 (CD86/CD200R) ratio was observed in the blood of patients with HTS [106]; however, increased M2 infiltration and suppressed proinflammatory cytokines were observed in local HTS tissue [105]. These findings are not directly contradictory because they were obtained from different patient populations. One possible interpretation is that impaired M1 recruitment and a dampened initial inflammatory response may contribute to fibrogenesis. Nevertheless, this interpretation remains hypothetical, and longitudinal studies simultaneously profiling circulating and tissue-resident macrophages are needed. In the proliferative phase of HTSs and keloids, M2 macrophages are significantly enriched, as confirmed by immunohistochemistry [104], immunofluorescence staining [107] and scRNA-seq [16]. Aggregated M2 macrophages secrete pro-fibrotic and pro-angiogenic cytokines such as TGF-β, PDGF, and VEGF, stimulating myofibroblast formation, ECM production and angiogenesis [108]. The persistent presence and activation of M2 macrophages contribute to fibrosis and excessive vascularization [109]. Multiple mechanisms are involved in the formation of a PS microenvironment prone to M2 polarization [110], including the regulation of various ILs [111], mechanical stimulation [101], and intercellular communication [16, 43, 108].

The spatial distribution of M2 macrophages tends to correlate with the invasive characteristics of PSs, with M2 macrophages being more abundant at the margins and surfaces of lesions and less prevalent in the central and deeper parts [103]. Although macrophage infiltration is associated with PS progression, quantitative evidence linking macrophage polarization states to specific clinical parameters remains limited and requires further validation [112].

The positive loop between macrophage polarization and fibroblast stimulation plays an important role in PS progression. In a coculture system of keloid fibroblasts and M0 macrophages, fibroblasts secrete chemotactic cytokines to aggregate M0 macrophages [104] and skew their differentiation toward the M2 macrophage phenotype [108]. M2 macrophages subsequently upregulate the secretion of TGF-β and PDGF, which inversely stimulate the activation of myofibroblast differentiation [113]. Specifically, M2 macrophages induce fibroblasts to upregulate the expression of ubiquitin C-terminal hydrolase L1 (UCHL1), which increases the proliferation, invasion and collagen expression of fibroblasts through the insulin-like growth factor 1 (IGF-1)/protein kinase B (PKB, also known as AKT)/mechanistic target of rapamycin (mTOR)/HIF-1α and Wnt/β-catenin pathways [2]. At the resolution of scRNA-seq, secreted phosphoprotein 1⁺ (SPP1⁺) macrophages in keloids stimulate the ECM-related gene expression of MFs, thereby contributing to the progression of keloids [112]. Disrupting these positive loops represents a mechanism-based strategy to halt excessive remodeling in pathological scarring.

Adaptive immune responses and autoimmunity

T cells play an essential role in cellular immunity. Generally, PSs are characterized by a high CD4/CD8 T-lymphocyte ratio with distinct functional alterations [114, 115]. T helper cells, particularly T-helper 2 (Th2) cells, are closely linked to fibrogenesis. Th2 cell cytokines (IL-4 and IL-13) can promote fibroblast proliferation and ECM accumulation through interleukin-4 receptor alpha (IL-4Rα)/JAK/STAT and TGF-β signaling [116]. IL-4 and IL-13 also stimulate M2 polarization of macrophages [2]. Th2 cells and their cytokines are significantly elevated in keloids and HTSs, indicating their important roles in pathological scarring [117, 118]. Investigators have identified four Th2-associated chemokines [CC motif chemokine ligand (CCL) 2, CCL4, CCL5, and C-×3-C motif chemokine ligand 1 (CX3CL1)] that maintain high expression levels throughout the process of proliferative scar formation, revealing the potential mechanism through which Th2 cells are overrecruited to promote fibrosis [119]. Similarly, T helper (Th17) cells are abundantly infiltrated in the peripheral growing margins of keloids. They produce IL-17, which not only upregulates the expression of profibrotic genes but also induces fibroblasts to secrete CXCL12, establishing a positive feedback loop that promotes further Th17 infiltration [120]. Conversely, the number of cytotoxic CD8⁺ T cells is decreased in both blood and lesions [23]. Chronic antigenic stimulation in keloids upregulates the natural killer group 2A (NKG2A)/CD94 inhibitory receptor on these cells, suppressing their function via the soluble HLA-E (sHLA-E) ligand [106]. Additionally, memory T cells display altered cytokine patterns (high levels of interferon-γ (IFN-γ) and low levels of tumor necrosis factor-α (TNF-α)) and defective phenotypes, potentially contributing to the uncontrolled growth of keloids [121].

Regulatory T cells (Tregs), which are characterized by high expression levels of forkhead box P3 (FOXP3), play a crucial role in maintaining immune hemostasis. They prevent the overreactivity of other immune cells via direct cell-to-cell contact mechanisms and through the release of inhibitory cytokines such as IL-10 and TGF-β. The proportion of local infiltrating regulatory Tregs in keloid tissue is significantly elevated [113, 117]. Coculture studies have indicated that activated Tregs promote COL1A1 and COL3A1 expression in fibroblasts, a mechanism dependent on the upregulation of TGF-β expression rather than that of IL-10 [122]. Owing to limited evidence, the precise role of Tregs in keloid pathogenesis remains uncertain, as their increased presence may indicate either a pathogenic role or a response to inflammation.

B lymphocytes are integral cells of the adaptive immune system. Immunohistochemistry (IHC) staining demonstrated increased infiltration of CD19⁺ [123] and CD20⁺ [123, 124] B lymphocytes in PSs. One Gene Expression Omnibus (GEO) data mining study [125] revealed a statistically significant correlation between six differentially expressed genes (DEGs), namely, CCL2, cell division cycle associated 7 (CDCA7), atypical chemokine receptor 4 (ACKR4), plexin domain containing 2 (PLXDC2), forkhead box F2 (FOXF2) and cell adhesion molecule 1 (CADM1), and B naïve cells in HTSs. These DEGs may be involved in the regulation of naïve B cells in HTSs as either a cause or an effect. It has been hypothesized that B cells can secrete various immunoglobulins, which can bind to antigens associated with PSs, forming immune complexes that trigger local inflammatory reactions and thereby influencing the onset and progression of PSs [126]. However, no experimental studies have investigated and validated the role of B lymphocytes in keloids and HTSs.

These adaptive immune alterations offer distinct avenues for clinical intervention and diagnosis. Targeting the Th2 axis with dupilumab (a biologic targeting IL-4Rα) has shown promise in the treatment of PSs, although the efficacy varies and requires further validation through high-quality trials [111, 127–129]. Furthermore, the immune checkpoint ligand sHLA-E is highly sensitive and specific for distinguishing patients with keloids. Notably, sHLA-E levels decrease following intralesional therapy and are correlated with recurrence risk, highlighting its potential utility as a biomarker for diagnosis and prognosis prediction [23].

Innate immune cells and tissue homeostasis

Innate immune cells, comprising mast cells, dendritic cells (DCs), and natural killer (NK) cells, exhibit quantitative and functional dysregulation in PSs. Mast cells can influence neural regeneration and sensitivity through the release of neuropeptides and nerve growth factors [130]. In addition, they can secrete histamine and immunoglobulin E (IgE), inducing local inflammatory responses [131]. Intercellular communication analysis confirmed the significant enrichment between neural cells and mast cells [24], indicating that they are relevant to pain and itching in PSs [132]. ScRNA-seq revealed a significant increase in the number of mast cells in keloids compared with normal skin and mature scars [24, 43]. These results are consistent with those of previous studies on PSs [132–134]. In addition to quantitative changes, alterations in the phenotype and functionality of mast cells have been reported. Multiple studies have shown that mast cells exhibit increased activity, characterized by a high proportion of degranulated mast cells and increased expression of activated markers and inflammatory cytokines [24, 133, 135]. The expression levels of gene markers associated with activation and degranulation, such as bruton tyrosine kinase (BTK) and GATA binding protein 2 (GATA2), were significantly increased in the mast cell population of keloids compared with the levels in mature skin. Furthermore, mast cells are potentially involved in fibrosis through the secretion of proteases. Tryptase acts as a potent mitogen, activating MFs via the upregulation of protease-activated receptor 2 (PAR2). Chymase can promote the proliferation and collagen secretion of fibroblasts in proliferative HTSs through activating the TGF-β1/Smad pathway, although reports regarding its expression levels remain controversial, with conflicting evidence of upregulation versus downregulation in different cohorts [136, 137].

DCs are potent antigen-presenting cells that play crucial roles in activating specific T-cell responses and in establishing immune tolerance [138, 139]. Compared with those in mature scar areas, the dermis in keloid regions contains a significantly greater number of coagulation factor XIII A chain (FXIIIa)-positive dermal DCs. These findings indicate that FXIIIa-positive dermal DCs, which are involved in dermal–epidermal interactions, actively contribute to PS formation [140]. Langerhans cells are a specialized subset of DCs that reside in the epidermis or mucosal epithelia. The number of Langerhans cells in HTSs is significantly greater than that in normal scars, indicating the potential role of Langerhans cells in the formation of PSs [141].

NK cells detect and eliminate infected cells, stressed cells, and tumor cells without specificity. They can secrete diverse immunomodulatory cytokines and chemokines including IFN-γ and TGF-β [142]. The change in the amount of NK cells has been controversial in previous studies. A study exploring immunologic characteristics of active HTS samples with IHC showed that fewer NK cells were present [115]. Flow cytometry results showed a significant enrichment of NK cells, and another study using scRNA-seq analysis revealed a low enrichment in keloids [143]. This inconsistency could be related to individual differences and technological disparities. NK cells express the inhibitory receptor NKG2A/CD94, whose ligand is sHLA. Abundant keloid fibroblasts secrete sHLA-E into the circulation and the microenvironment, which binds to the inhibitory receptor and may suppress NK cell activity [23]. Additionally, because both NK cells and T cells are cytotoxic, there is a need for strict coordination and cooperation between the two cell types [144]. Dysfunctional NK cells in the keloid environment may be associated with the exhaustion of cytotoxic T cells [145]. Overall, the role of NK cells in PSs is still poorly understood. Further analysis is needed to explore the cellular heterogeneity, molecular landscape, and pathogenic role of NK cells (Figure 3; Table 2).

Figure 3.

Cellular landscape and intercellular communication networks in pathological scars. The schematic summarizes major epithelial, mesenchymal, endothelial, lymphatic, and immune cell populations and highlights representative cytokines, growth factors, signaling pathways, and cell–cell interactions that contribute to fibroblast activation, myofibroblast differentiation, ECM deposition, angiogenesis, inflammation, and scar progression. This figure was created by the authors.

Table 2.

Cellular biomarkers and clinical correlations

Biomarker	Sample type	Clinical correlations	Tiera
sHLA-E [23]	Serum	Levels decrease after therapy and correlate with recurrence risk	1
CILP1 [146]	Serum	Sensitive indicator for HTS progression	2
PIM1 [147]	Tissue	Biomarker for skin pan-fibrosis; inhibition induces myofibroblast ferroptosis	3
miR-31-5p [148]	Tissue	Prognostic value for head and neck keloid resections	2
FAP [149]	PET/CT imaging	Visualizes activated fibroblasts; expression peaks at invasive periphery, guiding surgical margins	2

^aDefinitions: Tier 1, supported by direct clinical associations and longitudinal observations; Tier 2, supported by preliminary patient-level or pilot clinical evidence; Tier 3, supported primarily by mechanistic or preclinical evidence

Signaling networks and pathway integration

Hierarchical orchestration of core pathological pathways

The progression of PSs is driven not by isolated molecular events but rather by a hierarchical regulatory network. Within this ecosystem, canonical pathways converge to override normal healing resolution, establishing a persistent, self-sustaining fibrotic phenotype across the dermal landscape.

Transforming growth factor beta/small mothers against decapentaplegic signaling

In addition to functioning as a general profibrotic trigger, TGF-β1 serves as a key driver of fibrosis. ScRNA-seq has fundamentally redefined the sender and receiver dynamics of this signaling axis. Specifically, this technology has revealed that POSTN⁺ MFs located in the reticular dermis effectively function as a TGF-β reservoir [4, 15, 150]. These specialized cells exhibit pathological overexpression of TGF-βR1 and TGF-βR2, which initiate downstream signaling through Smad proteins, ultimately resulting in excessive deposition of ECM [151]. Crucially, this pathogenic axis is further stabilized by USP11-mediated deubiquitination of TGF-β receptors [152].

Wingless-related integration site/β-catenin pathway

The Wnt pathway acts as the primary master switch for cellular plasticity and lineage determination in PSs. Specifically, within the central mesenchymal core, sustained Wnt/β-catenin activation drives adipocyte-to-myofibroblast transition. By antagonizing proadipogenic transcription factors such as peroxisome proliferator-activated receptor γ (PPARγ) and promoting lipolysis, this signaling cascade forces dermal progenitors and resident adipocytes to abandon their regenerative identity in favor of a highly contractile actin alpha 2, smooth muscle⁺ (ACTA2⁺) myofibroblast phenotype [19, 151]. Furthermore, this crucial lineage reprogramming is anchored by the transcription factor zinc-finger E-box-binding 1 (ZEB1). This transcription factor binds directly to Wnt promoter regions to maintain a persistently activated fibrotic gene expression program. Consequently, this molecular anchoring facilitates the relentless expansion of keloids beyond their original wound margins [153].

Janus kinase/signal transducer and activator of transcription signaling

The JAK/STAT pathway links inflammatory cues to fibrotic remodeling. At the fibroblast level, miR-190a-3p–mediated downregulation of the complement regulator CUB and Sushi multiple domains 1 (CSMD1) enhances migration and fibronectin production in HTS fibroblasts and is accompanied by activation of the JAK/STAT pathway [154]. In keloid fibroblasts, loss of the decoy receptor interleukin-13 receptor alpha 2 (IL-13RA2) activates IL-13/STAT6 signaling, promoting collagen accumulation, whereas restoration of IL-13RA2 expression or pharmacologic STAT6 inhibition mitigates keloid fibrosis [104]. Beyond fibroblasts, Wnt5A-driven IL-6 secretion from keloid fibroblasts activates JAK/STAT3 signaling and EMT in adjacent keratinocytes, providing a mechanism through which stromal–epithelial crosstalk contributes to the locally invasive behavior of keloids [92].

Phosphoinositide 3-kinase/protein kinase B pathway

The PI3K/AKT pathway integrates growth factor and hypoxic signals with metabolic reprogramming. In keloid fibroblasts, hypoxia-induced PI3K/AKT activation drives a metabolic shift toward aerobic glycolysis and suppresses mitochondrial oxidative phosphorylation [155, 156]. This metabolic reprogramming enhances fibroblast proliferation, migration and invasion and operates in a positive feedback loop with HIF-1α [155, 156]. Regulators such as N-Myc downstream regulated gene 2 (NDRG2) [157], angiopoietin-2 (ANGPT2) [158], and the USP37–spalt-like transcription factor 4 (SALL4) [159] axis further potentiate PI3K/AKT signaling in HTS and keloid fibroblasts.

Intercellular communication

Pathological scarring is increasingly viewed as a spatially organized microenvironment in which cellular proximity, rather than absolute abundance, dictates signaling outcomes. ST has revealed discrete cellular niches defined by highly coordinated molecular interactions.

Within these niches, spatial mapping has redefined macrophages as structural organizers rather than transient mediators. Specifically, SPP1⁺ macrophages are spatially tethered to POSTN⁺ MFs in the deep dermis [30]. This proximity facilitates a paracrine relay in which macrophage-derived SPP1 binds to CD44 receptors on adjacent fibroblasts. This engagement triggers intracellular cascades that increase fibroblast adhesion and metabolic survival under hypoxic conditions, ultimately driving their differentiation into contractile myofibroblasts and creating localized fibrotic hotspots [30, 112].

In the deep reticular layers, a specialized fibrovascular unit is sustained by intense crosstalk. ECs undergoing EndoMT secrete PDGF and VEGF, which act as potent mitogens for perivascular fibroblasts. These signals, complemented by TGF-β3–TGF-βR2 and Eph–Ephrin interactions, synchronize angiogenesis with collagen synthesis [19, 74, 152]. Simultaneously, a neuro-fibrotic axis has been identified: fibroblasts secrete midkine (MDK) to promote Schwann cell proliferation and their transition toward a repair phenotype [160]. This activation triggers neuronal axonogenesis and the release of neuropeptides such as substance P, providing a molecular basis for the chronic pain and pruritus characteristic of active keloids.

The physical stiffness of the ECM is converted into biochemical persistence via the Yes-associated protein/transcriptional coactivator with PDZ-binding motif (YAP/TAZ) mechanosensing axis [161]. Mechanical tension at the scar periphery promotes the nuclear translocation of YAP, which acts as a mechanosensitive rheostat [162]. Crucially, nuclear YAP upregulates the expression of TGF-β receptors, lowering the threshold for fibrotic activation. This mechanofibrotic feedback loop ensures that the tissue’s physical rigidity remains a self-sustaining driver of fibrosis long after the initial inflammatory stimulus ceases [28, 163].

Clinical translation and therapeutic implications

Precision medicine and multimodal diagnostics

Corticosteroids and antineoplastic drugs remain the main pharmacological treatments for PSs [163]. Both inhibit fibroblast proliferation through G0/G1-phase arrest, and corticosteroids further suppress inflammation by inhibiting NF-κB and mitogen-activated protein kinase (MAPK) signaling [26, 146, 164, 165]. However, these empirical approaches do not account for the substantial cellular and molecular heterogeneity of PSs. With advances in single-cell and spatial profiling, clinical management is gradually shifting toward biomarker-driven precision medicine. Rather than viewing scars as simple localized overgrowths, PSs are increasingly recognized as a spectrum of autoinflammatory fibrotic disorders with distinct cellular states and anatomical identities [48, 164]. For example, recent evidence suggests that keloid activity is correlated with the balance between proinflammatory fibroblasts and SPP1⁺ macrophages [26, 30, 112]. Site-specific transcriptomic programs, such as homeobox (HOX) gene imprinting in earlobe versus trunk keloids, further indicate that the therapeutic response and resistance may be anatomically programmed, supporting the need for site-tailored treatment strategies [165–167].

These advances in cellular and molecular profiling have identified several candidate biomarkers with potential clinical relevance. At present, no biomarker has sufficient multicenter prospective validation for routine decision-making. Among the currently available candidates, systemic soluble HLA-E (sHLA-E) has relatively stronger clinical support because of its association with postoperative recurrence risk [23, 164], whereas serum cartilage intermediate layer protein 1 (CILP1), which may indicate HTS progression, and tissue miR-31-5p, which has shown prognostic value in head and neck keloid resections, remain exploratory biomarkers requiring further validation [146, 148]. Their utility should be further assessed in prospective, biomarker-stratified cohorts to determine whether they can guide treatment selection and predict recurrence.

In addition to molecular markers, imaging technologies provide tools for assessing scar activity in vivo [168]. For example, 68Ga-labeled fibroblast activation protein inhibitor-04 (68Ga-FAPI-04) positron emission tomography/computed tomography (PET/CT) can be used to visualize activated fibroblasts in vivo. By revealing that fibroblast activation protein (FAP) expression peaks at the invasive periphery, this technology directly guides surgeons in planning more precise margins [149]. For nonsurgical monitoring, multimodal ultrasound, including shear wave elastography (SWE) and angio-planewave imaging, together with optical coherence tomography (OCT), can be used to quantify tissue stiffness and microcirculatory changes. These tools provide the necessary feedback to calibrate laser parameters and identify early responders to therapy [168, 169]. Complementing these in vivo monitoring tools, patient-derived 3D keloid spheroids provide an ex vivo platform for simulating drug responses before clinical intervention, thereby bridging molecular profiling with individualized treatment selection [170].

Targeted therapeutic strategies

Identifying cellular interaction hubs has shifted the therapeutic focus from broad tissue destruction toward the precise interception of fibrotic cascades. A primary frontier is targeting immune–stromal crosstalk, especially for refractory keloids [111]. For instance, dupilumab, an anti-interleukin-4 receptor alpha (IL-4Rα) antibody, flattens steroid-resistant lesions by blocking Th2-driven signaling [111, 127]. Because variable clinical responses can result from IL-4/IL-13 fluctuations or Th17 differentiation, these therapies necessitate rigorous patient monitoring [129, 171–173]. Beyond Th2 blockade, oral JAK inhibitors (e.g. tofacitinib and upadacitinib) have emerged as viable systemic options [174, 175].

To minimize systemic side effects, localized gene silencing via siRNA is in active clinical development. The LEM-S401 system delivers siRNA targeting connective tissue growth factor (CTGF) through mesoporous silica nanoparticles. The safety and stability of this drug have already been validated in human trials (NCT04707131), which have shown localized silencing without detectable systemic exposure [176]. Moreover, drug repurposing offers a faster route to clinical use [177]. Multikinase inhibitors such as nintedanib block p38, c-Jun N-terminal kinase (JNK), and extracellular signal-regulated kinase (ERK) phosphorylation to suppress fibroblast invasion [178, 179]. Similarly, targeting the TNF-related apoptosis-inducing ligand receptor 2/death receptor 5 (TRAIL-R2/DR5) axis selectively induces apoptosis in hyperproliferative fibroblasts while sparing healthy tissue [180].

Newer targets are also expanding the therapeutic landscape. Proviral integration site for Moloney murine leukemia virus 1 (PIM1) has been identified as a biomarker for skin panfibrosis. Inhibiting it with small molecules such as bruceine D induces ferroptosis in myofibroblasts [147]. Finally, for symptomatic relief, blocking the MDK–Schwann cell axis represents a novel strategy to alleviate chronic pain and pruritus [160].

Innovative delivery systems and combination strategies

To overcome the dense ECM of the mesenchymal core, research is shifting toward the development of biomaterial-guided delivery systems. The physical rigidity of scars often leads to uneven drug distribution. To address this challenge, 3D-printed personalized microneedle arrays have been developed. These arrays use finite element simulations to conform to high-tension areas such as the thorax, resulting in a significant increase in drug loading capacity [39, 176]. With respect to long-term management, dissolving microneedle patches (e.g. loading 5-fluorouracil) have demonstrated clinical noninferiority to injections while significantly reducing pain and improving patient adherence [161, 181]. A transformative strategy employs sitagliptin-loaded patches to induce fibroblast-to-adipocyte conversion. By inhibiting dipeptidyl peptidase 4 (DPP4) and enhancing IGF-1 signaling, these patches redirect MSCs toward a regenerative lineage, offering a nondestructive alternative to corticosteroids [182, 183].

Furthermore, engineered nanocarriers such as exosomes are being repurposed from pathological messengers as precision delivery vehicles [184]. Exosomes loaded with antifibrotic miRNAs (e.g. miR-26b-5p) specifically target hypoxic regions to modulate the macrophage-to-myofibroblast transition [159, 185]. Given the inherent heterogeneity of PSs, multimodal combination protocols are becoming the standard of care. Sandwich therapy combines radionuclide applicators with injections to simultaneously target superficial vessels and the deep mesenchymal core [186]. Additionally, laser-facilitated drug delivery (fractional CO₂ with topical triamcinolone) provides a safer alternative to traditional injections, offering comparable volume reduction with lower risks of skin atrophy [187–189]. Finally, to address systemic discrepancies, the international adoption of standardized core outcome sets is essential to ensure that future trials utilize uniform assessment frameworks, integrating both molecular readouts and patient-reported outcomes [190] (Figure 4; Table 3).

Figure 4.

Integrated therapeutic landscape of pathological scars. (a) Emerging precision diagnostic and monitoring strategies. (b) Representative targeted therapies for pathological scar. (c) Advanced local drug-delivery strategies. FGFR fibroblast growth factor receptor, IFN-γ interferon gamma, IFNGR interferon gamma receptor, IL interleukin, PDGFR platelet-derived growth factor receptor, TGF-β1 transforming growth factor beta 1, TNF-α tumor necrosis factor alpha. Created at https://BioRender.com

Table 3.

Precision treatment targets for pathological scars

Cellular target	Representative agent	Stage	Therapeutic rationale
Multiple tyrosine kinases	Nintedanib [178, 179]	Preclinical	Suppresses collagen synthesis.
IL-4Rα	Dupilumab [111, 127–129]	Phase I/II	Blocks Th2-driven immune-fibroblast crosstalk, softening steroid-resistant scars.
JAKs	Tofacitinib, Upadacitinib [174, 175]	Preclinical	Inhibits signal transduction of multiple pro-fibrotic cytokines.
CTGF(CCN2)	LEM-S401 (siRNA) [176]	Phase I (NCT04707131)	Localized gene silencing via nanoparticle delivery.
DPP4	Sitagliptin Microneedles [182]	Preclinical	Inhibits DPP4 and enhances IGF-1 signaling, inducing fibroblast-to-adipocyte conversion.
MDK	iMDK (small molecule) [160]	Preclinical	May alleviate chronic pain and pruritus associated with scars.
PIM1	Bruceine D [147]	Preclinical	Inhibits PIM1 kinase, inducing ferroptosis in myofibroblasts.

Future directions and emerging technologies

The integration of single-cell and spatial omics has fundamentally reshaped our understanding of PSs, transitioning from a fibroblast-centric view to a multicellular ecosystem. To translate these molecular maps into clinical cures, several frontiers must be addressed. Future research should focus on four-dimensional multimodal mapping, adding a temporal dimension to existing spatial data [17]. Capturing transcriptional inflection points during the acute-to-chronic transition is essential for identifying intervention windows to prevent fibrosis [17, 190]. Furthermore, next-generation platforms such as Stereo-seq and Xenium will enable the simultaneous detection of RNA, proteins, and metabolites, revealing how mechanical cues are transduced into biochemical signals with unprecedented resolution [36].

To manage the resulting data complexity, advanced computational frameworks are needed. However, a significant barrier remains in that current scRNA-seq and ST datasets are frequently confounded by batch effects and interpatient heterogeneity. Biological noise arising from differences in anatomical sites, patient ethnicities, and even the mechanical rigors of tissue dissociation can obscure true pathological signals. Automated annotation tools, such as scExtract, are being developed to leverage large language models to ensure reproducibility across clinical centers by mitigating these technical variances [42]. Beyond annotation, machine learning algorithms can generate “digital twins” of patient scars, enabling in silico drug simulations to personalize therapeutic selection [191]. However, bridging the gap between computational predictions and biological reality requires functional validation in human-centric models. To this end, microvascularized skin-on-a-chip platforms provide a biomimetic environment to decipher cell fate reprogramming [38]. Harnessing these mechanistic insights will facilitate a shift from broad tissue destruction toward targeted cellular editing [182]. For instance, inducing fibroblast-to-adipocyte conversion offers a refined approach to restore skin function while minimizing adverse effects and recurrence [192].

Finally, addressing methodological heterogeneity through the international adoption of standardized core outcome sets and unified data processing pipelines is essential for clinical translation [190, 193]. Collaborative efforts toward an organ-scale wound healing atlas will provide a unified reference for distinguishing normal from pathological trajectories across diverse ethnicities and anatomical sites [17]. In parallel, future reviews and primary studies should also explicitly distinguish single-cell findings that introduce new pathogenic mechanisms from those that validate earlier histopathological or immunohistochemical observations.

Conclusions

In conclusion, the single-cell and spatial era has provided a cellular and molecular atlas for mechanism-based therapies. By integrating mechanobiology, metabolism, and immune–stromal crosstalk into a spatiotemporal framework, the field is advancing toward personalized, regenerative solutions that promise to overcome the persistent challenge of pathological scarring.

Abbreviations

ACKR1

Atypical chemokine receptor 1

ACKR4

Atypical chemokine receptor 4

ACSS3

Acyl-CoA synthetase short chain family member 3

ACTA2

actin alpha 2, smooth muscle

ADAM12

A disintegrin and metalloproteinase 12

AKT

Protein kinase B (also known as PKB)

ANGPT2

Angiopoietin-2

ATP

Adenosine triphosphate

bFGF

Basic fibroblast growth factor

bHLH

Basic helix–loop–helix

BoNT-A

Botulinum toxin type A

BTK

Bruton tyrosine kinase

CADM1

Cell adhesion molecule 1

CCL

CC Motif chemokine ligand

CCN3

Cellular communication network factor 3

CD

Cluster of differentiation

CDCA7

Cell division cycle associated 7

CILP1

Cartilage intermediate layer protein 1

COL1A1

Collagen type I alpha 1 chain

COL3A1

Collagen type III alpha 1 chain

COL11A1

Collagen type XI alpha 1 chain

COMP

Cartilage oligomeric matrix protein

CSMD1

CUB and sushi multiple domains 1

CTGF

Connective tissue growth factor

CXCL3

C-X-C motif chemokine ligand 3

C×3CL1

C-×3-C motif chemokine ligand 1

CXCL12

C-X-C motif chemokine ligand 12

DCs

Dendritic cells

DEGs

Differentially expressed genes

DPP4

Dipeptidyl peptidase 4

DR5

Death receptor 5

ECM

Extracellular matrix

ECs

Endothelial cells

EMT

Epithelial-to-mesenchymal transition

EndoMT

Endothelial-to-mesenchymal transition

ERK

Extracellular signal-regulated kinase

ET-1

Endothelin-1

FAP

Fibroblast activation protein

FAPI-04

Fibroblast activation protein inhibitor-04

FOSL1

FOS-like antigen 1

FOXF2

Forkhead box F2

FOXP3

Forkhead box P3

FSP-1

Fibroblast-specific protein 1

FXIIIa

Coagulation factor XIII A chain

GATA2

GATA binding protein 2

GEO

Gene expression omnibus

GO

Gene ontology

HIF-1α

Hypoxia-inducible factor 1 alpha

HLA-E

Human leukocyte antigen-E

HMGB1

High mobility group box 1

HMOX1

Heme oxygenase 1

HOX

Homeobox

HTS

Hypertrophic scar

IFN-γ

Interferon gamma

IGF-1

Insulin-like growth factor 1

IGFBP5

Insulin-like growth factor-binding protein 5

IHC

Immunohistochemistry

IgE

Immunoglobulin E

IL

Interleukin

IL-4R

Interleukin-4 receptor

IL-4Rα

Interleukin-4 receptor alpha

IL-13RA2

Interleukin-13 receptor alpha 2

iPSCs

Induced pluripotent stem cells

JAK

Janus kinase

JNK

c-Jun N-terminal kinase

KANK4

KN motif and ankyrin repeat domain-containing protein 4

KGF

Keratinocyte growth factor

LECs

Lymphatic endothelial cells

Ly-EndoMT

Lymphatic-endothelial-to-mesenchymal transition

LYVE-1

Lymphatic vessel endothelial hyaluronan receptor 1

MAPK

Mitogen-activated protein kinase

MDK

Midkine

MMP3

Matrix metallopeptidase 3

MSCs

Mesenchymal stem cells

mTOR

Mechanistic target of rapamycin

MyD88

Myeloid differentiation primary response 88

NDRG2

N-Myc downstream regulated gene 2

NES

Nestin

NF-κB

Nuclear factor kappa-light-chain-enhancer of activated B cells

NK

Natural killer

NKG2A

Natural killer group 2A

OCT

Optical coherence tomography

OSM

Oncostatin M

PAR2

Protease-activated receptor 2

PDGF

Platelet-derived growth factor

PET/CT

Positron emission tomography/computed tomography

PI3K

Phosphoinositide 3-kinase

PIM1

Proviral integration site for moloney murine leukemia virus 1

PKB

Protein kinase B (see AKT)

PLXDC2

Plexin domain containing 2

POSTN

Periostin

PPARγ

Peroxisome proliferator-activated receptor gamma

PS

Pathological scars

RAGE

Receptor for advanced glycation end-products

ROS

Reactive oxygen species

SALL4

Spalt-like transcription factor 4

S100A2

S100 calcium binding protein A2

scRNA-seq

Single-cell RNA sequencing

SDC1

Syndecan 1

SEMA3C

Semaphorin 3C

sHLA-E

Soluble human leukocyte antigen-E

Smad

Small mothers against decapentaplegic

snRNA-seq

Single-nucleus RNA sequencing

SPP1

Secreted phosphoprotein 1

ST

Spatial transcriptomics

STAT

Signal transducer and activator of transcription

SWE

Shear wave elastography

TGF-β

Transforming growth factor beta

TGF-βR1

Transforming growth factor beta receptor 1

TGF-βR2

Transforming growth factor beta receptor 2

Th

T Helper cell

TLR

Toll-like receptor

TNFAIP6

Tumor necrosis factor alpha-induced protein 6

TNF-α

Tumor necrosis factor alpha

TRAIL-R2

TNF-related apoptosis-inducing ligand receptor 2

Tregs

Regulatory T cells

UCHL1

Ubiquitin C-terminal hydrolase L1

USP

Ubiquitin-specific protease

VECs

Vascular endothelial cells

VEGF

Vascular endothelial growth factor

VEGFR

Vascular endothelial growth factor receptor

VGLL3

Vestigial like family member 3

VIM

Vimentin

Wnt

Wingless-related integration site

YAP/TAZ

Yes-associated protein/transcriptional coactivator with PDZ-binding motif

ZEB1

Zinc-finger E-box-binding 1

α-MSH

Alpha-melanocyte-stimulating hormone

α-SMA

Alpha-smooth muscle actin

β-catenin

Beta-catenin

Author contributions

Yixin Sun (Conceptualization [equal], Investigation [equal], Writing—original draft [equal]), Sijing Yan (Conceptualization [equal], Writing—original draft [equal]), Mengdi Zhang (Investigation [equal], Writing—original draft [equal]), Wangfei Mo (Investigation [equal]), Catharina Tao (Writing—review & editing [equal]), Jingyu Li (Funding acquisition [equal], Methodology [equal], Validation [equal], Writing—review & editing [equal]), Jiuzuo Huang (Writing—review & editing [equal]), Nanze Yu (Conceptualization [equal], Writing—review & editing [equal]), Yi Sun (Conceptualization [equal], Writing—review & editing [equal]), Xiao Long (Conceptualization [equal], Writing—review & editing [equal])

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Conflicts of interest

The authors declare no potential conflicts of interest with respect to the research, authorship, and publication of this article.

Funding

The work was supported by the National Natural Science Foundation of China (82472565, 82503017), Peking Union Medical College Hospital Talent Cultivation Program (Category B) No. UGG10110, Peking Union Medical College Hospital Talent Cultivation Program (Category C) No. UBJ11557, Zhejiang Provincial Medical and Health Science and Technology Program (2025HY0098), and Postdoctoral Fellowship Program of China Postdoctoral Science Foundation (No. GZC20251546).

Data availability

All data generated or analyzed during this study are included in this published article.