Skin rash was a frequent adverse event among patients with non-small cell lung cancer (NSCLC) who received sintilimab, necessitating early detection to mitigate treatment discontinuation risk and improve outcomes. As the first untargeted metabolomics study investigating sintilimab-induced cutaneous toxicity in NSCLC, we identified significant plasma metabolic alterations between rash-affected patients and matched controls. Pathway analysis revealed perturbations in oxytocin signaling, GnRH signaling, platelet activation, FcγR-mediated phagocytosis, retrograde endocannabinoid signaling, pantothenate and CoA biosynthesis, FcεRI signaling, and aldosterone synthesis/secretion. Univariate analysis (AUC ≥ 0.800) identified 25 significantly differentially expressed metabolites, while SHAP analysis selected 20 metabolites with substantial predictive value. Cross-comparison between these two metabolite panels identified five overlapping metabolites: N,N,N-trimethyl-L-histidine, laurolactam, 2-naphthalenesulfonic acid, limonenecarboxylic acid, and N-lauroylsarcosine. These co-identified metabolites offer novel insights into the pathogenesis of immune checkpoint inhibitor-related skin toxicity and represent promising candidates for further biomarker development.

To investigate the clinical context of these metabolic disturbances, we analyzed a cohort of lung cancer patients with a median age of 60 years (range: 42–78 years). Neither age (P = 0.069) nor sex distribution (P = 0.257) differed significantly between patients with and without cutaneous rash. Nevertheless, the observed trend toward a younger age in the rash group, though not statistically significant, suggests that age might be a modifying factor deserving further exploration in larger-scale studies. Current clinical studies and real-world evidence indicate that the development of sintilimab-associated rash lacks a clear direct correlation with drug dosage or plasma concentration levels (Shi et al., 2022). The mechanism is primarily attributed to immune hyperactivation resulting from PD-1/PD-L1 pathway blockade, which leads to autoimmune-like reactions against normal tissues such as the skin, categorizing it as an immune-related adverse event (irAE) (Postow et al., 2018). This response is influenced more by individual variations in immune reactivity than by the conventional “dose-dependent toxicity” profile commonly observed with cytotoxic chemotherapeutic agents. Consequently, the emergence of rash is more closely associated with the extent of immune activation rather than attainment of a specific drug exposure threshold. Supporting this, clinical investigations of other PD-1 inhibitors, including nivolumab (Topalian et al., 2012), have similarly reported no significant association between rash incidence or severity and administered dose or systemic drug concentrations.

Notably, patients with rash exhibited distinct biochemical alterations compared to those without rash. Specifically, the rash group had significantly higher levels of total bile acids (TBA), glucose (GLU), and basophil percentage (BAS%) (P < 0.05). Conversely, they showed significantly lower AST/ALT ratio, alkaline phosphatase (ALP), lactate dehydrogenase (LDH), phosphorus (P), neutrophil count (NEU), and high-sensitivity C-reactive protein (hsCRP) (P < 0.05). Our data implicate sintilimab-induced rash in a multifactorial immunoinflammatory cascade. Mechanistically, the significantly elevated BAS% suggests Th2-polarized immunity, aligning with immune checkpoint inhibitor (ICI)-induced atopic-like cutaneous reactions (Sibaud, 2018). The reduced NEU indicates a non-neutrophilic inflammatory pattern, consistent with reported hematologic irAEs in ICI therapy (Yerolatsite et al., 2024). Markedly increased TBA may activate the pruritogenic receptor TGR5 on cutaneous sensory neurons, directly promoting itch and barrier dysfunction (Lieu et al., 2014), while hyperglycemia exacerbates skin inflammation through the AGE-RAGE axis, as demonstrated in inflammatory dermatoses (Ikumi et al., 2019). Collectively, these alterations pointed to concurrent hepatobiliary injury, metabolic dysregulation, and aberrant Th2 immunity as multifactorial drivers of sintilimab-related cutaneous toxicity.

To elucidate the molecular pathways underpinning these observed clinical and biochemical changes, we performed KEGG enrichment analysis of the altered metabolites. This analysis revealed that the differentially expressed metabolites were significantly enriched in pathways critical for immune effector functions, inflammatory signaling, and metabolic regulation pertinent to cutaneous immunity. These included platelet activation, FcγR-mediated phagocytosis, retrograde endocannabinoid signaling, pantothenate and CoA biosynthesis, FcεRI signaling, and aldosterone synthesis and secretion. Platelet activation and FcγR-mediated phagocytosis pathways facilitate the clearance of opsonized targets and initiate inflammatory responses; their dysregulation may contribute to immune-mediated tissue injury (Nieswandt et al., 2011). Notably, the FcεRI signaling pathway directly regulates mast cell degranulation, a key pathogenic event in drug-induced skin rashes involving histamine release and urticaria. Concurrently, perturbations in retrograde endocannabinoid signaling could disrupt cutaneous neuro-immune homeostasis, potentially modulating inflammatory thresholds in the skin (Liu et al., 2021). Furthermore, pantothenate and CoA biosynthesis, essential for acyl carrier protein function, fatty acid metabolism, and cellular energy status, influences inflammatory mediator production; its impairment may compromise epidermal barrier repair (Miallot et al., 2023). Dysregulated aldosterone synthesis and secretion might also exacerbate inflammatory edema by affecting electrolyte balance and vascular permeability within the dermal microenvironment (Herrada et al., 2011). These metabolic and immune pathway disturbances are consistent with recent pharmacological findings linking PD-1 inhibition, metabolic reprogramming, and cancer therapeutic responses (Tian et al., 2025a; Tian et al., 2026). Collectively, the enrichment of these pathways suggests a multifaceted interplay between sintilimab-induced immunomodulation, altered effector cell functions (particularly involving Fc receptors and mast cells), dysregulation of neuro-immune-endocrine crosstalk, and impaired metabolic support, collectively contributing to the development of cutaneous adverse reactions, including rash.

Among the differentially expressed metabolites contributing to these pathway perturbations, we focused on N,N,N-trimethyl-L-histidine, more commonly known as ergothioneine, given its potential link to histaminergic signaling and the FcεRI pathway. This muscle-derived metabolite was significantly elevated in NSCLC patients who developed an immune-related rash following sintilimab treatment, highlighting a potential link between PD-1 inhibitor-induced immunometabolic reprogramming and cutaneous toxicity. Sintilimab-enhanced T-cell activation promotes skin infiltration of Th1/Th17 cells and mast cell degranulation via cytokines including IL-3, IL-31, and GM-CSF, leading to histamine release and classic rash symptoms such as vasodilation, edema, and pruritus (Watanabe and Yamaguchi, 2023; Yamamoto, 2022). The accumulation of ergothioneine, a methylated histidine derivative, reflects systemic reprogramming of histamine metabolism, likely indicating impaired degradation through histamine-N-methyltransferase (HNMT) and a histamine-driven hyperinflammatory state (Hamid and Hamidi, 2024). This methylated amino acid dysregulation may arise from immune-induced hyperactivity of S-adenosylmethionine metabolism and keratinocyte stress responses (Chen et al., 2023). Notably, ergothioneine can activate the NLRP3 inflammasome to promote IL-1β release and directly aggravate skin inflammation (Higashikuni et al., 2023); its structural similarity to histamine also suggests potential competitive binding at H1 receptors, likely modulating pruritus signaling. Furthermore, clonal expansion of activated T cells heightens dependence on amino acid availability, potentially disturbing muscle metabolite homeostasis and facilitating release of endogenous peptides including carnosine derivatives into circulation (Ma et al., 2024; Reschke et al., 2022). Histamine may also enhance local dendritic cell activation, amplify T-cell responses and create a pro-inflammatory feedback loop that sustains skin inflammation (Liu et al., 2023). Thus, ergothioneine may serve not only as a biomarker of histamine pathway dysregulation in sintilimab-induced rash, but also play a functional role in linking immune-mediated amino acid metabolism and cutaneous inflammatory responses.

Another differentially abundant metabolite identified in this study is laurolactam, a 12-membered ring lactam structurally associated with fatty acid metabolism and potentially originating from the cyclization of long-chain fatty acids or amino acids in vivo (Fukuta et al., 2009). Notably, its structural similarity to Azone (laurocapsome), a known skin penetration enhancer that disrupts stratum corneum lipid organization to promote transdermal drug delivery (Fukuta et al., 2009), leads us to hypothesize that accumulated laurolactam may act as an endogenous penetration enhancer under sintilimab treatment, facilitating the entry of drug components or inflammatory mediators into the skin and thereby contributing to rash initiation. Moreover, the skin barrier relies on a lipid matrix comprising ceramides, cholesterol, and free fatty acids (Wertz, 2021). Sintilimab-induced systemic T-cell immune activation and release of pro-inflammatory cytokines (e.g., TNF-α, IFN-γ, IL-17, IL-22) can directly inhibit the expression of key lipid metabolic enzymes (such as serine palmitoyl transferase and β-glucocerebrosidase) in keratinocytes, resulting in disrupted epidermal lipid homeostasis (Martins et al., 2020). As both a marker and potential functional contributor to this metabolic dysregulation, elevated laurolactam may not only signify barrier impairment but also further destabilize lipid architecture, increase trans epidermal water loss, and facilitate the invasion of foreign allergens and microorganisms. These effects could synergistically enhance local immune responses and T-cell infiltration, ultimately driving severe and persistent inflammatory rash (Gooran et al., 2023).

In addition to endogenous metabolites, we also observed significant alteration of 2-naphthalenesulfonic acid, an exogenous aromatic sulfonic acid frequently found in dyes and chemical intermediates, which is primarily derived from the sulfonation of pharmaceuticals or environmental chemicals in vivo (Cao et al., 2023). Its pronounced elevation suggests that sintilimab therapy may perturb host metabolic homeostasis, potentially through dysregulation of drug-metabolizing enzymes and gut microbial imbalance. Growing evidence links immune-related adverse events induced by checkpoint inhibitors to alterations in gut microbiota composition; specific dysbiosis may facilitate the overgrowth of bacteria capable of generating aromatic metabolites such as 2-naphthalenesulfonic acid, promoting its systemic accumulation (Pezo et al., 2019). This compound can translocate to the skin, where it may behave as a hapten by covalently binding cutaneous proteins (e.g., keratins) to form neoantigens. Under conditions of skin barrier impairment—often associated with T-cell-mediated inflammation—these neoantigens can be captured by antigen-presenting cells and presented to sintilimab-primed T cells, potentially eliciting a delayed-type (type IV) hypersensitivity reaction that amplifies rash initiation and progression (Zemelka-Wiacek et al., 2022). Furthermore, 2-naphthalenesulfonic acid possesses documented antioxidant and antimicrobial properties that may modulate the local redox balance and microbial composition in the skin, thereby indirectly influencing inflammatory pathways (Mehra and Chadha, 2020). Thus, the accumulation of this metabolite not only signifies aberrant processing of exogenous compounds during treatment but may also actively contribute to the immunopathological mechanisms underlying cutaneous adverse reactions.

Similarly, limonene carboxylic acid—an oxidized metabolite of limonene, a monoterpene abundant in citrus fruits—can be generated through both endogenous metabolic pathways and atmospheric oxidation processes (Wang and Wang, 2021; Rinaldi de Alvarenga et al., 2022). Although limonene itself possesses low sensitization potential, its oxidation derivatives, including limonene hydroperoxide and limonene carboxylic acid, are recognized as potent contact allergens (Bennike et al., 2019). Notably, this metabolite has also been associated with anti-inflammatory, antioxidant, and immunomodulatory activities, such as suppression of the NF-κB pathway, reduction of pro-inflammatory cytokines (e.g., TNF-α and IL-6), and elevation of anti-inflammatory IL-10 (Kathem et al., 2024). Within the setting of sintilimab treatment, characterized by systemic immune hyperactivation and likely pre-existing skin barrier impairment through mechanisms noted earlier, externally derived limonene carboxylic acid—potentially from dietary citrus, personal care products, or fragrances—may permeate the skin and function as a hapten. By forming neoantigens via covalent binding to cutaneous proteins, it can be internalized by Langerhans cells and presented to PD-1-inhibitor-primed T cells, thereby provoking a strong allergic contact dermatitis-like reaction (Nishijima et al., 1997). Conversely, endogenously produced limonene carboxylic acid may be upregulated as a compensatory anti-inflammatory mediator in an attempt to counterregulate localized inflammation; however, its efficacy is likely overwhelmed in the context of robust T cell-mediated skin injury. Collectively, the emergence of limonene carboxylic acid underscores a critical interaction between exogenous environmental exposure and endogenous immune sensitization, supporting its potential involvement in the pathogenesis and exacerbation of sintilimab-induced skin rash.

N-lauroylsarcosine, identified as another differential metabolite in this study, is an anionic surfactant formed through the condensation of lauric acid and sarcosine (N-methylglycine). Its aberrant accumulation likely stems from sintilimab-induced systemic inflammation, wherein pro-inflammatory cytokines suppress key epidermal lipid metabolic enzymes, disrupt fatty acid homeostasis—including lauric acid metabolism—and perturb one-carbon and methylation pathways (Morellato et al., 2021; Yan et al., 2022). Possessing strong lipid- and protein-dissolving properties, N-lauroylsarcosine destabilizes cell membranes and impairs stratum corneum bilayer integrity. Even at endogenously low concentrations, it acts similarly to exogenous surfactants by compromising skin barrier function through physicochemical mechanisms, resulting in excessive lipid removal, elevated transepidermal water loss, and enhanced penetration of external irritants (Kim et al., 2008). This sustained barrier dysfunction perpetuates “danger signals” that activate innate immune pathways—such as Toll-like receptor cascades—promote inflammatory cytokine secretion, and amplify T-cell-driven immune responses, thereby establishing a self-renewing cycle of inflammation and barrier failure (Wong et al., 2009). Additionally, its inherent antimicrobial activity may disrupt the cutaneous microbiota and alter the local immune microenvironment (Zhao et al., 2024). In contrast to alkaline surfactants like sodium laurate, N-lauroylsarcosine is mildly acidic and helps maintain native skin surface pH, thereby avoiding delayed pruritus, though it can still provoke acute mechanical irritation (Inami et al., 2013). Importantly, this metabolite may also arise through microbial metabolism of drug excipients: microbial esterases could hydrolyze polysorbate 80 in the sintilimab formulation to release lauric acid, which may then conjugate with host-derived sarcosine, highlighting a potential drug–microbiome–host interplay in immune-related dermal toxicity. Thus, N-lauroylsarcosine not only serves as a biomarker of disturbed epidermal lipid metabolism but may also play an active role in the initiation and persistence of sintilimab-associated rash via direct barrier disruption and potentiation of innate immunity.

It should be noted that the endogenous or dietary/environmental origin of these metabolites was inferred based on database annotations and literature rather than confirmed by experiments; future studies with stable isotope tracing or controlled dietary interventions would help verify their precise origins. Additionally, all of these top metabolites were among the differential metabolites used for KEGG enrichment, and each is directly or indirectly involved in the top pathways discussed above (e.g., ergothioneine in FcεRI signaling, laurolactam in CoA biosynthesis). Collectively, the above mechanistic interpretations of the five key metabolites are hypothetical and exploratory, and further in vitro, in vivo, or clinical experiments are warranted to verify their exact roles in sintilimab-induced rash.

While the metabolomic profiles generated in this study offer valuable mechanistic insights into the pathogenesis of sintilimab-associated cutaneous toxicity, several critical limitations need to be acknowledged. Primarily, our findings are preliminary due to the small sample size and absence of an independent external validation cohort in this single-center study, which markedly limits the generalizability of the identified biomarker panel. As such, these results are hypothesis-generating in nature and mandate further verification in independent patient populations. The top differentially expressed metabolites screened via univariate analysis and machine learning algorithms should be prioritized for targeted quantitative validation using orthogonal methodologies (e.g., MRM LC-MS/MS) in large-scale, multicenter cohorts. In addition, although our pathway enrichment analysis was based on dysregulated metabolites, direct statistical correlations between individual key metabolites and the activity scores of top enriched pathways were not assessed in the current study; this correlation analysis is a crucial priority for subsequent large-cohort investigations to clarify the mechanistic links between metabolic dysregulation and cutaneous toxicity. Finally, exclusive reliance on plasma biospecimens narrows the metabolic scope of this study; future research should prioritize multi-compartment metabolic profiling (e.g., skin interstitial fluid, skin tissue biopsies) to establish a comprehensive multi-omics atlas of immune checkpoint inhibitor (ICI)-induced adverse events. Looking forward, the candidate metabolite panel identified herein holds clinical translational potential for two core applications: (i) pre-treatment risk stratification, whereby baseline metabolite levels could be measured to identify high-risk patients requiring enhanced dermatological surveillance; and (ii) early pharmacodynamic monitoring, via serial metabolite testing during the first 1–2 treatment cycles to detect metabolic perturbations prior to the clinical onset of rash. Prospective multicenter validation in independent cohorts is indispensable prior to clinical implementation of this biomarker panel. Furthermore, future integration of pharmacometabolomics with genome-scale functional module analysis will refine the accuracy and translational value of predictive models for sintilimab-induced rash (Tian et al., 2025b).