Keloids are a benign fibroblast proliferative skin disorder characterized by an overabundance of fibroblasts beyond the boundaries of the original trauma (1–3). Several keloid treatments, including intradermal glucocorticoid injections, surgical excision, and radiation therapy, are available. However, therapeutic effects are usually unsatisfactory, with high recurrence rates (4). Notably, early identification of keloid-prone individuals is significantly important; this can minimize the occurrence of keloids by changing daily behaviors and modifying medical practices to avoid wounds as much as possible. However, it also helps timely intervention in the early stages of keloid formation to avoid serious adverse consequences.

Currently, the recognition of keloids relies heavily on the clinical presentation of keloids that typically extend beyond the original wound border. In addition, histopathological features of keloid collagen and α-smooth muscle actin contribute to diagnosis (1, 5). However, these diagnostic methods are not ideal owing to the need for invasive procedures, such as biopsy, because trauma could directly contribute to the formation of keloids. Consequently, there is an urgent need to explore new non-invasive methods and biomarkers for the early and accurate identification of keloids. For instance, liquid biopsies extracted by non-invasive methods from various biological fluids are performed to analyze the components, cells, or metabolites (6). With the advantages of being non-invasive, rapid, easy, and reproducible, it has become a useful tool for precision medicine, especially in the early diagnosis and individualized therapy of diseases.

Metabolites are intermediate products of metabolic reactions, including cellular biochemical and physiological processes, and are sensitive to internal and external environmental stimuli (7). Metabolomics, which is defined as the study of the collection of all small-molecule metabolites, could delineate characteristic metabolic “fingerprints” or “footprints” through targeted or non-targeted strategies, which contributes to the discovery of biomarkers with diagnostic, predictive, and therapeutic values (8). More importantly, several studies have suggested that keloid development is related to multiple metabolic pathways, such as glucose and lipid metabolism (9, 10). However, studies on body fluid biomarkers in keloid patients are lacking.

This study reports the landscape of circulating metabolic fingerprinting for keloids. Ultra-high-performance liquid chromatography coupled with Q Exactive mass spectrometry (UHPLC-QE-MS) platform was applied for untargeted metabolomic analysis in 15 keloid patients and healthy volunteers in a discovery cohort. Subsequently, the least absolute shrinkage and selection operator (LASSO) was used to screen four differentially expressed metabolites associated with the risk of keloids, including palmitoylcarnitine, sphingosine, phosphocholine, and phenylalanylisoleucine. These were validated through targeted metabolomic analysis in an expanded training and test cohort. Based on a 4-metabolite fingerprint classifier, we described the development and calculation of a risk score for recognizing keloids using a linear and logistic regression model. We propose that the characteristics of circulating metabolic fingerprinting could provide novel and convenient non-invasive testing targets and methods for the early identification of keloid-prone individuals.

To date, the accurate identification of keloids relies on the clinical presentation and histopathology of the biopsied tissue of the formed or mature keloids. Current keloid therapies, such as intralesional corticosteroid injections, surgical excision, and postoperative radiation therapy, have limited efficacy and a high risk of recurrence. However, given the difficulty in treating keloid scars, more emphasis should be placed on the early identification and prevention of keloid management. Therefore, developing non-invasive, specific, and sensitive tools that can be used for early keloid identification is extremely important. Metabolomics has emerged as a promising technology for non-invasive evaluation using biofluids such as blood and urine. In addition, it is cheaper than other omics methods such as proteomics and transcriptomics, making it easier to translate into clinical practice (14).

This study investigated and identified the landscape of metabolic fingerprinting for the identification and risk scoring of keloids. We performed an untargeted metabolomic analysis based on the UHPLC-QE-MS platform to characterize the metabolic profiles of patients with keloids. Thirty statistically different metabolites were identified, most of which were fatty acyl groups, steroids and steroid derivatives, organooxygen compounds, carboxylic acids and derivatives, and glycerophospholipids, which is consistent with previous studies. Zhang et al. performed a gene microarray analysis on fibroblasts from normal, hypertrophic scars, and keloid tissues. The results suggested that differentially expressed genes were strongly associated with sterol, fatty acid, and steroid metabolic processes (15). Compared to the normal human epidermis, the number of triglycerides, cholesteryl ester, and wax ester decreased by 60%, 33%, and 80% in the epidermis of keloid lesions, respectively (16). Consistently, our study also found that cholesterol levels decreased in the plasma of keloid patients detected in ESI+ mode. In addition, lipid extraction, fractionation, and fatty acid identification of skin lesions from African keloid patients revealed a deficiency of essential fatty acids, such as linoleic acid, γ-linolenic acid, dihomo-γ-linolenic acid, α-linolenic acid, and eicosapentaenoic acid. However, it showed elevated levels of arachidonic acid in keloids (17). In the untargeted metabolomic analysis, α-linolenic acid, linoleic acid, and γ-linolenic acid were significantly down-regulated metabolites in the keloid plasma. In contrast, arachidonic acid was significantly up-regulated. α-Linolenic acid dose-dependently inhibits the growth of fibroblasts in keloid patients, suggesting a protective effect on keloid formation (18). However, the decline in α-linolenic acid in the plasma and skin lesions of keloid patients may contribute to keloid susceptibility. Therefore, the release of [3H]arachidonic acid from keloid-derived fibroblasts was greater than that from normal skin-derived fibroblasts (19). Arachidonic acid is pro-inflammatory, and its downstream products, such as leukotrienes, prostaglandins, prostacyclins, and thromboxane, are considered to be pro-inflammatory modulators (9).

We identified and confirmed for the first time using the LASSO logistic regression model that up-regulated levels of palmitoylcarnitine and sphingosine, and down-regulated levels of phosphocholine and phenylalanylisoleucine are strongly correlated with the early identification of keloids. High levels of plasma palmitoylcarnitine have been observed in various cancers and chronic diseases such as obesity and type 2 diabetes (20–22). Palmitoylcarnitine has been found to induce Ca²⁺ influx and promote the secretion of the pro-inflammatory factor IL-6 (23). GO analysis of lncRNA expression profiles in keloid tissues revealed a vital role in calcium pathway signaling, and calcium channel blockers have been reported to treat keloids (24, 25). IL-6 has been shown to promote proliferation and collagen synthesis in keloid-derived fibroblasts, thus participating in keloid pathogenesis (26). Therefore, we hypothesized that palmitoylcarnitine plays a potential role in promoting keloid formation. Sphingosine is a classic product of sphingolipid metabolism. Together with ceramide, sphingosine-1-phosphate is fundamental in regulating cell proliferation, survival, and death. Although ceramide and sphingosine generally suppress proliferation and promote apoptosis, sphingosine-1-phosphate is associated with growth stimulation and apoptosis inhibition; hence, there is a “sphingolipid rheostat” effect between ceramide, sphingosine, and sphingosine-1-phosphate, indicating that these metabolites are interconvertible. Notably, the fate of cells is determined by their relative levels rather than their absolute amounts of these metabolites (27). Sphingosine and sphingosine-1-phosphate have been reported to promote cell proliferation at low concentrations through calcium metabolism and activation of the MAPK pathway in Swiss 3T3 cells; however, they are cytotoxic at high concentrations (28). Fingolimod (FTY720), an analog of sphingosine-1-phosphate, has been used to treat keloid scars (29). Therefore, we hypothesized that elevated sphingosine levels in keloid plasma might be involved in keloid development by affecting the relative concentration of sphingosine-1-phosphate. Phosphocholine is the polar headgroup of the dominant membrane phospholipid phosphatidylcholine. In addition, bioactive phosphocholine-containing lipids can be recognized by the innate immune system and play a crucial role in endothelial dysfunction, apoptosis, and endoplasmic reticulum stress (30). Phenylalanylisoleucine is a dipeptide composed of phenylalanine and isoleucine and is an incomplete breakdown product of protein digestion or catabolism. Dysregulation of amino acid metabolism is associated with protein synthesis, glucose and lipid homeostasis, mTOR signalling pathways, and immune responses (31). Although the specific mechanisms of these four metabolite predictors is yet to be elucidated in the pathogenesis of keloids, our findings highlight the possible clinical application of plasma metabolic markers in predicting keloid formation.

There are certain limitations to this study. Therefore, future prospective multicenter studies with large size of samples from different periods of keloids are required to further validate the diagnostic model’s specificity and sensitivity. In addition, complete coverage of the keloid metabolome by detecting and identifying all metabolites remains challenging and requires the application of different detection platforms and the development of metabolomic databases.

In summary, we successfully determined the plasma metabolic profile of keloid patients and developed a risk score model for keloid identification based on the 4-metabolite fingerprint classifier in plasma. We hope that this study will aid the development of non-invasive and early identification or prediction tools for keloids, thus opening up new possibilities for the early personalized treatment of keloid patients.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by Institutional Ethical Review Board of Peking Union Medical College. The patients/participants provided their written informed consent to participate in this study.

DH, KC and JZ conceived and designed the project. YH, XZ, RL, SJ and LL performed the experiments. MJ, CL, HC, LC and ZW analyzed experimental results. YH and XZ wrote the manuscript with approval from all authors. DH, KC and JZ revised the manuscript. All authors contributed to the article and approved the submitted version.

This study is supported by grants from National Natural Science Foundation of China (No. 82273552, 82203947 and 82073445), Natural Science Foundation of Jiangsu Province (No. BK20210049), CAMS Innovation Fund for Medical Sciences (CIFMS-2021-I2M-1-001) and Medical Scientific Research Project of Jiangsu Provincial Health Commission (M2022113).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.