Diabetes has emerged as a global health crisis, with its prevalence reaching unprecedented levels and imposing a heavy burden on healthcare systems worldwide. Among the complications of diabetes, diabetic peripheral neuropathy (DPN) is one of the most common and disabling microvascular complications, affecting up to 50% of diabetic patients (Dilshad and Ephrem, 2023), while its estimated prevalence among U.S. adults with diabetes is 28% (Hicks and Selvin, 2019). This progressive neurological disorder is characterized by distal symmetric polyneuropathy and manifests through symptoms such as numbness, burning pain, and motor dysfunction (Sanaye and Kavishwar, 2023). These symptoms not only severely impair patients’ quality of life but also increase the risk of foot ulcers, which often progress to amputations, further worsening disability and even being life-threatening (Volmer-Thole and Lobmann, 2016).

The damage of peripheral nerve by metabolics dysfunction in the late stages leads to DPN to occur (Zhu et al., 2024). Hyperglycemia, which is typical for diabetes mellitus, causes a number of harmful biological processes and excites the excessive production of reactive oxygen species (ROS). Glucose-induced oxidative stress hampers cell’s antioxidant mechanisms and damages lipids, proteins and DNA in neurons (Lee et al., 2020). Constant increase in blood glucose facilitates the accumulation of advanced glycation end products (AGEs), produced by non-enzymatic reactions between sugar and protein. Activation of the pro-inflamatory pathways and neuroinflammation by AGEs ligated to the receptor represents one of the main mechanisms contributing to the nerve damage. In this inflammatory context, tumor necrosis factor-α (TNFα) has been implicated in the development of DPN. A meta-analysis reported elevated circulating TNFα levels in patients with DPN compared with diabetic individuals without neuropathy, supporting an association between TNFα and DPN. Moreover, experimental inactivation/neutralization of TNFα in streptozotocin-induced diabetic mice ameliorated neuropathy-related functional deficits and was accompanied by changes in downstream inflammatory signaling (including NF-κB-related readouts), suggesting that TNFα-associated inflammation may contribute to DPN pathophysiology. Therefore, TNFα represents a plausible mediator linking metabolic stress to inflammation-related peripheral nerve injury, although the causal contribution of TNFα may depend on disease stage and experimental context (Mu et al., 2017; Yamakawa et al., 2011). Mitochondrial dysfunction also has an important role, besides blocking energy production, the latter activates ROS production (Bai and Luo, 2024). They disrupt axonal transport, impair neurovascular endothelial function, and accelerate demyelination (Cardoso and Salles, 2016; Wu et al., 2013).

Atorvastatin, an inhibitor of 3-hydroxy-3-methylglutaryl-coenzyme A (HMG-CoA) reductase, is widely used to lower lipids and cardiovascular risk in hypercholesterolemia (Lee et al., 2020; Meng et al., 2019). Despite there is abundant evidence that atorvastatin is causes neurological adverse reactions, diabetic patients are concerned that prolonged or high-dose use could lead to the exacerbation or progression of peripheral neuropathy (Evangelista et al., 2019). It seems that the atorvastatin neurotoxicity could manifest itself in various aspects. In addition to increasing oxidative stress via ROS overproduction in the neuronal cells, the loss of mitochondrial homeostasis and maintenance of the cellular metabolism are revealed. Inhibition of the cholesterol synthesis in neurons’ membranes and the membranes integrity are involved (Matoba et al., 2019; Xie et al., 2023). These combined factors worsen nerve damage in diabetic patients and can initiate or speed up DPN. A person’s susceptibility to this damage depends on drug interactions, genetic predisposition, and overall health (Meng et al., 2019). Detailed assessment of potential neurotoxicity of atorvastatin is necessary to better understand its benefits when applied to patients with DPN.

Network toxicology is a relatively advanced approach that integrates multi-omics data, protein-protein interaction (PPI) networks, and pathway analysis, providing a very useful framework for addressing such complex situations. By aggregating and analyzing large-scale datasets, it can identify key molecular targets and signal cascades mediating drug effects, which are introduced in references (Lin et al., 2021; Wu et al., 2023). This method has been successfully used to clarify the toxicological mechanisms of environmental pollutants and drugs, such as muscle-related adverse reactions induced by statins. Applying network toxicology to the research of atorvastatin and DPN enables researchers to systematically depict the interactions between drug targets and molecules involved in the pathogenesis of DPN, such as molecules regulating oxidative stress, apoptosis and neurotrophic factor signaling. These molecules support the survival and repair of neurons. In addition, molecular docking can predict the binding affinity of atorvastatin and key targets related to DPN, which can select the most likely biologically relevant interactions that are worthy of experimental verification (Sun et al., 2023). This synergy between computational and experimental methods bridges the gap between in silico analysis and in vivo significance, and accelerate the discovery of actionable mechanisms.

As a supplementary content to network toxicology and molecular docking methods, single-cell RNA sequencing (scRNA-seq) has become a revolutionary tool. It can study complex biological processes and disease mechanisms with resolution never seen before (Awuah et al., 2023). By capturing the complete transcriptome expression profile at the single-cell level, scRNA-seq has solved the limitations of bulk sequencing and it will average the gene expression in cell populations and mask the heterogeneity within tissues. This technology can describe the gene expression patterns within individual cells, thereby revealing the functional diversity specific to cell types, identifying rare or previously undiscovered cell subpopulations, and depicting the dynamic transcriptional changes related to drug exposure or disease progression. With the continuous development of gene chips and bioinformatics technologies, scRNA-seq data are widely used in disease research to crack complex pathogenic mechanisms. In the context of neurological disorders such as DPN, scRNA-seq offers a particular opportunity to determine how atorvastatin regulates gene expression in different cell types involved in peripheral nerve homeostasis, including neurons, Schwann cells, and vascular endothelial cells, etc. It can also provide insights into the mechanisms by which it functions at the cellular level.

This study aims to integrate network toxicology, molecular docking, molecular dynamics simulation, scRNA-seq and other methods to comprehensively explore the possible effects of atorvastatin on DPN (Figure 1). To clarify the potential molecular mechanisms behind the various effects of atorvastatin on peripheral nerves, multiple research strategies have been adopted. This includes conducting systematic network toxicology-based studies on the interaction between atorvastatin and its biological targets, using molecular docking to predict the binding affinity between the drug and the target, verifying the kinetic stability of the conjugate through molecular dynamics simulation methods, and describing the transcriptional responses specific to cell types using single-cell RNA sequencing. The toxicity of atorvastatin on peripheral neuro-related cells was verified in an in vitro cell model by the CCK-8 method, and cell experiments were conducted to confirm the mediating role of related molecules in atorvastatin-induced DPN, deepening our understanding of the role of atorvastatin in DPN and provided more precise and personalized treatment strategies for this intractable complication. Clarifying this complex relationship can create conditions for improving the care of millions of diabetic patients and those with neurological sequelae worldwide.

DPN causes nerve pain and disrupts patients’ daily life. Its characteristic pathological changes include axonal loss and regeneration, myelin abnormalities and demyelination (Jack and Wright, 2012). Without early treatment, it can lead to foot ulcers, then gangrene, and even amputation (Yu, 2021). Dyslipidemia is a major risk factor for DPN in type 2 diabetes (Girach et al., 2006). Studies show that statins do more than lower lipids, they can also improve blood vessel function and reduce inflammation (Davis et al., 2008; Palinski, 2001; Zhou and Liao, 2010). This may help protect nerves and treat DPN. But epidemiological studies link statin use to a higher risk of neuropathy (Corrao et al., 2004). This study integrates network toxicology, single-cell RNA sequencing, molecular docking, and molecular dynamics simulations to systematically identify potential targets of atorvastatin-induced DPN, such as TNF, CTNNB1, and CASP3. Studies show TNF showed high network centrality and relatively favorable predicted interaction with atorvastatin in the docking/MD analyses. Based on these observations together with the in vitro readouts, our results are consistent with a potential involvement of TNF/TNFα-associated responses in atorvastatin-related cellular stress under the tested conditions.

In this study, GO and KEGG enrichment analyses identified TNF, CTNNB1, and CASP3 as key genes associated with the mechanism of atorvastatin-induced DPN. These targets were enriched in critical biological processes such as oxidative stress, neurotransmitter release, and lipid metabolism. Previous studies have indicated that lipophilic statins can induce mitochondrial damage through multiple mechanisms. This triggers oxidative stress, resulting in cellular injury (Kaufmann et al., 2006). Our findings are consistent with this pathway. KEGG analysis indicates that atorvastatin may promote DPN through the AGE-RAGE signaling pathway. Pathway activation recruits mDia1 and, through the PI3K/Akt axis, promotes NF-κB nuclear translocation to regulate downstream transcription. RAGE binding to Aβ peptides on the neuronal surface triggers NF-κB activation, inducing the release of macrophage colony-stimulating factor (M-CSF) and other pro-inflammatory cytokines. M-CSF further interacts with RAGE, amplifying oxidative stress and inflammatory responses (Bhattacharya et al., 2023; Derk et al., 2018). This pathway represents a potential mechanism for atorvastatin-induced DPN. We have centered our discussion on the connections between atorvastatin, its targets, and DPN, focusing on inflammation, oxidative stress, and metabolic disturbances to hypothesize potential underlying pathways. Molecular docking suggested that atorvastatin may form a plausible binding pose with TNFα, CTNNB1, and CASP3, and the predicted binding scores for TNFα were comparatively favorable. These computational results provide supportive, hypothesis-generating evidence for a possible interaction, but additional experimental work is required to determine whether such interactions translate into inflammatory/oxidative-stress changes and peripheral nerve injury in vivo. Notably, prior structural and biochemical studies have shown that small molecules can directly engage soluble TNFα and modulate its trimer stability, for example by accelerating subunit dissociation and thereby reducing receptor-competent TNFα. These reports support the general plausibility that low-molecular-weight ligands may interact with TNFα. Nevertheless, whether atorvastatin binds TNFα under physiological conditions and whether such binding leads to functional modulation remain to be determined, and our docking/MD results should be interpreted as hypothesis-generating (He et al., 2005).

TNF, a key inflammatory mediator, is associated with the pathological progression of neuropathy in type 2 diabetes (Hussain et al., 2013). Among its subtypes, TNFα serves as the central functional molecule within the TNF family and plays significant roles in neuroinflammation, apoptosis, and tumorigenesis (Zelová and Hošek, 2013). Meta-analyses of database studies have revealed that serum levels of TNFα are significantly elevated in patients with DPN compared to healthy controls and type 2 diabetic patients without DPN (Mu et al., 2017). Under hyperglycemic conditions, upregulated expression of inflammatory mediators including TNFα and NF-κB not only impairs mitochondrial function and neurotrophic support, but also damages microvascular integrity via protein kinase C activation and enhanced hexosamine pathway activity, accelerating the progression of peripheral neuropathy (Mendez et al., 1996; Morohoshi et al., 1996; Vlassara et al., 1988; Xue et al., 2021). Mechanistic studies demonstrate that modulating the TNFα pathway can ameliorate DPN. Xiaoke Bitong Capsule (XBC) has been shown to alleviate neuroinflammatory damage in model animals by suppressing TNF signaling (Tian et al., 2024). Photobiomodulation (PBM) therapy reduces pain and nerve injury by decreasing TNFα expression in the central nervous system (Ferreira et al., 2024). Mudan Granules used in clinical management of DPN, reported/predicted to interact with TNFα (Long et al., 2025). These findings support the relevance of TNFα in DPN treatment. The relationship between statins and DPN involves complex mechanisms. Studies indicate that atorvastatin at higher concentrations, may induce neurotoxicity through multiple pathways: they promote macrophage infiltration into adipose tissue, increasing TNFα production, and impair mitochondrial function by disrupting calcium homeostasis, inhibiting the electron transport chain, and inducing mitochondrial membrane potential collapse (Golbidi and Laher, 2014; Murinson et al., 2012; Sirvent et al., 2005). These alterations lead to reduced ROS clearance, increased electron leakage, and subsequent oxidative stress (Esposito et al., 1999; Kaufmann et al., 2006). Oxidative stress and TNFα expression form a vicious cycle, wherein ROS promotes TNFα generation via activation of pathways such as NF-κB, while elevated TNFα further exacerbates oxidative stress (Husain et al., 2015; Ozsoy et al., 2008). Additionally, statin-induced depletion of coenzyme Q10 not only compromises mitochondrial antioxidant capacity but also elevates circulating levels of inflammatory factors including TNFα, collectively contributing to peripheral nerve damage (Banach et al., 2015; Duberley et al., 2014; Hou et al., 2023). These mechanisms together constitute the molecular basis by which statins may induce or exacerbate DPN.

Neuropathic pain is a key clinical feature of DPN, and sensory neuron–specific mechanisms are increasingly recognized. In line with the increasing focus on sensory neuron–related mechanisms, transcriptomic profiling of human dorsal root ganglia (DRGs) from patients with painful DPN has reported an inflammatory gene signature (including macrophage-associated transcripts) together with reduced expression of multiple neuronal genes, suggesting that neuroimmune interactions within the ganglion may contribute to pain hypersensitivity. These human data provide a clinical context in which TNF/NF-κB–related inflammatory signaling may be relevant, although they do not establish a direct causal link with atorvastatin exposure. VGLUT2-positive sensory neurons have been proposed as major carriers of peripheral nociceptive signaling to the spinal cord, and VGLUT2-associated changes have been reported in STZ-induced DPN models. Moreover, aberrant activation of the sTNF/TNFR1 axis in VGLUT2+ neurons has been implicated in neuroinflammatory regulation through NF-κB signaling. In this study, we provide in vitro support for involvement of the TNF/NF-κB pathway by showing increased phosphorylation of NF-κB p65 under HG + ATV conditions and its attenuation by R-7050. However, VGLUT2 expression changes and TNF–VGLUT2 relationships were not directly assessed here and warrant targeted validation in sensory neuron models and/or in vivo DPN settings (Zhang et al., 2024). Oxidative stress and inflammation are tightly linked in DPN pathogenesis. PRDX6 has been proposed as a redox-regulatory molecule with both antioxidant and anti-inflammatory properties and may represent a candidate node connecting ROS homeostasis with TNF/NF-κB signaling. In the present study, we provided experimental support for activation of the TNF/NF-κB pathway by showing increased p65 phosphorylation under HG + ATV conditions and its attenuation by R-7050. However, PRDX6 expression/activity (GPx/iPLA2) and its potential relationship with TNF/NF-κB activation were not examined and warrant targeted investigation in future work (Cao et al., 2022).

CTNNB1 encodes a multifunctional protein that is critically involved in cellular adhesion, signal transduction, and proliferation. It is highly expressed in brain regions such as the cerebral cortex, hippocampus, and cerebellum, playing a pivotal role in early brain development and has been strongly implicated in neurodevelopmental disorders, including intellectual disability, schizophrenia, and autism spectrum disorder (Zhuang et al., 2023).Pathogenic variants in the CTNNB1 gene are also associated with the pathogenesis of primary aldosteronism(PA) (Teo et al., 2015). In PA, adrenal overproduction of aldosterone leads to dyslipidemia, a key risk factor for DPN, suggesting a potential mechanism whereby CTNNB1 may contribute to DPN through the modulation of lipid metabolism (Seidel et al., 2019). At the molecular level, CTNNB1 acts as a central effector and transcriptional co-activator within the canonical Wnt signaling pathway (Liu et al., 2022). Within the hyperglycemic milieu, the Wnt pathway drives DPN progression through its regulation of Schwann cell apoptosis; pathway activation promotes cell immortalization, and its inhibition conversely compromises cell viability (Liu et al., 2020). GSK3β serves as a critical node, bridging the Wnt and PI3K/AKT signaling pathways to indirectly regulate Schwann cell apoptosis and thereby influence DPN pathogenesis (Liu et al., 2016). Based on these findings, we hypothesize that atorvastatin may contribute to DPN development by targeting CTNNB1. Furthermore, crosstalk exists between TNFα and β-catenin signaling; TNFα activates the β-catenin pathway via a signaling cascade initiated through the TNFR1 death domain, thereby promoting adipogenesis (Cawthorn et al., 2007; Kramer et al., 2023). This suggests a plausible mechanism through which TNFα may also accelerate the progression of DPN.

CASP3 serves as a pivotal executioner protease in the apoptotic pathway. In diabetic kidney disease (DKD) models, elevated CASP3 levels are closely associated with enhanced oxidative stress and inflammatory apoptosis (Yu et al., 2025). This enzyme executes its function through proteolytic cleavage of key substrates including poly(ADP-ribose) polymerase (PARP) (Salvesen and Dixit, 1997). Pathologically, CASP3 contributes to neurodegenerative processes, where its pharmacological inhibition effectively reverses HIV-1-induced TDP-43 aggregation and associated neurotoxicity (Yang et al., 2024). In DPN, Schwann cell apoptosis represents a central pathological event, and inhibition of activated CASP3 can partially ameliorate both autophagy suppression and apoptotic death in these cells (Zhang et al., 2021). Furthermore, in vitro studies confirm that atorvastatin triggers mitochondrial cytochrome c release, subsequently activating CASP3, which then cleaves essential proteins involved in RNA splicing, DNA repair, and other critical cellular processes, ultimately inducing apoptotic cell death (McIlwain et al., 2013; Panou et al., 2024). Based on current evidence, we propose a mechanistic hypothesis wherein atorvastatin promotes DPN progression via CASP3 activation: atorvastatin directly targets mitochondria, disrupting membrane integrity and inducing cytochrome c release into the cytosol. This process activates the caspase cascade, upregulating the “executioner” CASP3, which subsequently triggers apoptosis in peripheral neurons or glial cells (such as Schwann cells), thereby accelerating DPN pathogenesis.

From the perspective of neuronal excitability regulation, voltage-gated sodium channels (VGSCs) occupy a pivotal position in action potential propagation and represent crucial therapeutic targets for chronic pain management (Bennett et al., 2019). In DPN, TNFα has been demonstrated to directly activate the Nav1.7 subtype of sodium channels, thereby contributing to the development of neuropathic pain (Huang et al., 2014). Concurrently, disruption of calcium homeostasis plays an essential role in disease pathogenesis (Fernyhough and Calcutt, 2010). Downregulation of the mitochondrial calcium uniporter (MCU) compromises cellular calcium buffering, exacerbates oxidative stress, and promotes calcium influx, collectively facilitating apoptosis through both direct signaling and indirectly via induction of mitochondrial dysfunction (Jiang et al., 2025; Osmanlıoğlu and Nazıroğlu, 2024). Sustained calcium overload further activates the mitochondrial permeability transition pore (mPTP), causing collapse of the mitochondrial membrane potential and cytochrome c release, which initiates the caspase-dependent apoptotic cascade (Li et al., 1997). Taken together, our analyses suggest that atorvastatin exposure is associated with cellular stress/injury-related signals under the tested experimental conditions, and TNF emerged as a prioritized candidate target from the integrated analyses. As a selective, competitive HMG-CoA reductase inhibitor, atorvastatin is widely used to regulate hyperlipidemia. Its lipophilic nature allows broad distribution in the body and direct access to HMG-CoA reductase, making it a preferred clinical choice (Haj Hussein et al., 2022). Although the pathogenic mechanism of DPN remains unclear, extensive clinical data indicate that DPN can be triggered by mild inflammation, intraepidermal nerve fiber degeneration, impaired blood supply, hyperglycemic toxicity, and autoimmune disorders. Among these, low-grade autoinflammation caused by hyperglycemia, mitochondrial dysfunction, and dyslipidemia is a significant contributing factor. Thus, as a lipid-lowering agent, atorvastatin has the potential to mitigate the onset of DPN. However, clinical applications have revealed that some patients developed peripheral neuropathy—such as limb numbness, paresthesia, and burning pain—after taking statins (Mulchandani et al., 2018). Our team employed network toxicology and molecular docking to identify potential interaction targets between atorvastatin and DPN. Key targets were screened and their relevance demonstrated, elucidating the mechanistic basis of atorvastatin’s toxic effects on DPN.

To validate the predictions from network toxicology and molecular docking experiments, and to further elucidate the association between atorvastatin-induced neurotoxicity and the TNFα signaling pathway, we conducted cell-based assays. These experiments were performed under high-glucose conditions to simulate the diabetic microenvironment, using RSC 96 cells as a representative model for neural cells. ELISA and CCK-8 assays demonstrated that in a high-glucose environment, atorvastatin-treated RSC 96 cells exhibited significantly elevated levels of TNFα. Furthermore, under high-glucose conditions, atorvastatin exerted cytotoxic effects on RSC 96 cells in a time- and dose-dependent manner. These findings are consistent with atorvastatin-associated cytotoxicity under high-glucose conditions and suggest a possible association with TNFα elevation. However, TNFα blockade/knockdown studies are required to determine whether TNFα is necessary for the observed effects.

The findings of this study may provide useful information for future investigations into atorvastatin-related cellular responses in the context of DPN. However, several limitations should be noted. First, our network toxicology and molecular docking analyses are mainly based on predicted molecular interactions and cellular-level evidence, and therefore may not fully capture the complexity of in vivo pathophysiology. In addition, bioinformatics-based target identification depends on the completeness and annotation quality of public databases and on the chosen filtering criteria; thus, false positives/negatives cannot be fully excluded, and enrichment results should be interpreted as hypothesis-generating rather than definitive. Likewise, molecular docking and molecular dynamics simulations provide in silico estimates of plausible binding poses and interaction stability under simplified assumptions. Predicted docking scores/energies do not directly demonstrate physical binding in a biological milieu, nor do they establish functional modulation of the target. Therefore, any inferred “modulatory role” from ligand–protein interactions should be considered preliminary and requires orthogonal experimental validation (e.g., biophysical binding assays and/or target perturbation studies. In addition, no in vivo DPN model was included in the present study; therefore, the translational relevance of the observed associations remains to be determined and warrants validation in established diabetic neuropathy animal models (e.g., STZ-induced or db/db models) with functional and pathological endpoints. Neuron-specific validation, including assessment of VGLUT2 expression and potential TNF–VGLUT2 interactions relevant to pain signaling, was not performed in the current study and should be addressed in future work. Effects observed under controlled conditions may be amplified or attenuated in intact organisms, and clinical confounding factors such as patient heterogeneity (e.g., genetic background, duration of diabetes, glycemic control status) and concomitant medications cannot be addressed in the current design. Furthermore, we did not systematically investigate dose–response relationships or temporal dynamics. In vitro experiments often use relatively high drug concentrations to elicit measurable changes within a practical timeframe, which may not reflect chronic exposure conditions in patients. We also did not assess PRDX6 expression/activity (GPx/iPLA2) or integrate oxidative stress readouts (e.g., ROS) with TNF/NF-κB signaling, and therefore the proposed redox–inflammation crosstalk remains to be tested. Importantly, we did not perform TNFα blockade/knockdown experiments in this study; therefore, a causal requirement of TNFα for the observed cytotoxicity cannot be established, and future work using pharmacological inhibition or siRNA will be necessary.

This study combined network toxicology, scRNA-seq analysis, molecular docking/molecular dynamics simulations, and in vitro experiments to explore potential atorvastatin-associated cellular responses relevant to diabetic peripheral neuropathy (DPN). TNFα, CTNNB1, and CASP3 emerged as prioritized candidates from the integrated computational analyses, with enrichment in pathways related to inflammation and oxidative stress. The external scRNA-seq dataset suggested that these genes are expressed across multiple cell types, including a neuronal/neuronal-like cluster and other cells involved in tissue homeostasis. In silico docking/MD analyses indicated a plausible interaction between atorvastatin and these targets, with TNFα showing relatively strong predicted affinity; however, such interactions should be considered hypothesis-generating and require orthogonal validation. In high glucose–challenged RSC96 Schwann cells, atorvastatin was associated with reduced cell viability and increased TNFα-related inflammatory readouts, and our downstream assessment of NF-κB signaling (p65 phosphorylation) provided supportive evidence that TNF/NF-κB activation may be involved under the tested in vitro conditions.

Overall, our findings are consistent with the possibility that atorvastatin may contribute to DPN-relevant cellular stress through TNF/TNFα-associated inflammatory signaling, while additional pathways (including those related to oxidative stress and apoptosis) may also participate and should be examined in future work. These results may help motivate further mechanistic and translational studies aimed at evaluating atorvastatin safety in diabetic neuropathy settings, including in vivo validation with functional endpoints. Given the predominantly computational and in vitro design, conclusions regarding causality and clinical implications should be drawn cautiously.