Lipid metabolic reprogramming of tumor-associated macrophages drives resistance to immune checkpoint blockade in lung cancer: a narrative review of mechanisms and therapeutic strategies

Abstract

Background and Objective

Immune checkpoint inhibitors (ICIs), represented by programmed cell death protein 1 (PD-1)/programmed death-ligand 1 (PD-L1), have shown remarkable efficacy in non-small cell lung cancer (NSCLC); however, many patients still develop resistance to immunotherapy. Although small cell lung cancer (SCLC) is also an important histological type of lung cancer, NSCLC accounts for the majority of lung cancer cases. Current research on ICI development, first-line treatment efficacy, and the mechanisms of lipid metabolism in tumor-associated macrophages (TAMs) is predominantly focused on NSCLC. In patients with advanced NSCLC, objective response rates (ORRs) with PD-1/PD-L1 inhibitor monotherapy remain limited. Only in patients with high PD-L1 expression [tumor proportion score (TPS) ≥50%] and without sensitizing epidermal growth factor receptor (EGFR) mutations or anaplastic lymphoma kinase (ALK) rearrangements does the ORR increase to approximately 40–45%. TAMs are a key component of the immunosuppressive tumor microenvironment (TME). Lipid metabolic reprogramming profoundly influences the functional and transcriptional features of TAMs. This review aims to integrate relevant evidence, elucidate how TAM lipid metabolism promotes immunosuppression and resistance to ICIs, and outline potential therapeutic strategies.

Methods

We searched PubMed/MEDLINE, Web of Science, and Scopus for publications up to June 2026 using terms combining lung cancer, TAMs, lipid metabolism, and immune checkpoint blockade/resistance. Mechanistic, translational, and clinically relevant studies were selected by author consensus.

Key Content and Findings

Lipid uptake, de novo lipogenesis, fatty acid oxidation (FAO), cholesterol remodeling, and eicosanoid metabolism are not independent processes in TAMs. Lipid metabolic reprogramming in TAMs ultimately suppresses type I interferon (IFN-I) signaling, upregulates PD-L1 expression, and impairs the function of CD8⁺ T cells with stem-like features, thereby establishing an immunosuppressive TME and leading to resistance to ICIs. In lung cancer, hypoxia, high lactate levels, and tobacco exposure further shape the lipid phenotype of TAMs, such as lipid raft enrichment and lipid-laden macrophage subsets like SPP1⁺ macrophages. Different driver genomic backgrounds differentially impact tumor cell-intrinsic metabolism and the lipid metabolic programs of myeloid cells. In preclinical models, interventions targeting these metabolic axes, including TAM-directed delivery systems, have demonstrated potential therapeutic benefit when combined with anti-PD-1/PD-L1 therapy.

Conclusions

Targeting TAM lipid metabolism to convert immunologically cold tumors into more inflamed, ICI-responsive tumors is a promising strategy to overcome resistance in NSCLC. Identification of predictive biomarkers of therapeutic response and development of cell-selective drug delivery systems come to be major challenges.

Introduction

Lung cancer is one of the most common and deadliest malignant tumors worldwide (1). Among its subtypes, non-small cell lung cancer (NSCLC) accounts for approximately 85% of cases, while small cell lung cancer (SCLC) constitutes the majority of the remainder and is characterized by distinct neuroendocrine biology (2). Immune checkpoint inhibitors (ICIs) targeting programmed cell death protein 1 (PD-1)/programmed death-ligand 1 (PD-L1) have become an important therapeutic strategy for advanced NSCLC and have significantly improved overall survival. However, apart from specific chemo-immunotherapy regimens, SCLC shows a limited response to PD-1/PD-L1 monotherapy blockade (3). In addition, a considerable proportion of NSCLC patients still fails to benefit from ICI treatment. The objective response rate (ORR) with PD-1/PD-L1 monotherapy remains modest: approximately 15–20% in previously treated, largely unselected patients with advanced NSCLC and approximately 40–45% in treatment-naïve patients with a PD-L1 tumor proportion score (TPS) ≥50% and no sensitizing epidermal growth factor receptor (EGFR) mutations or anaplastic lymphoma kinase (ALK) rearrangements (4-6). In clinical practice, resistance is generally classified as primary or acquired. The underlying mechanisms of both types have not been fully elucidated, severely limiting therapeutic efficacy. Therefore, clarifying the mechanisms of resistance and developing corresponding translational therapeutic strategies have become major goals in the field of lung cancer immunotherapy.

ICI resistance is largely attributed to the immunosuppressive state of the tumor microenvironment (TME). Within the NSCLC TME, tumor-associated macrophages (TAMs) constitute a major immune-cell population, and many adopt immunosuppressive M2-like, pro-tumor states. These TAMs can express PD-L1, secrete interleukin-10 (IL-10) and transforming growth factor-β (TGF-β), recruit regulatory T cells, and impede effective infiltration by cytotoxic T cells (7). Collectively, these mechanisms mediate immunosuppression and promote tumor immune evasion, leading to diminished responses or resistance to ICI therapy. Given their high abundance and remarkable plasticity, modulating TAM recruitment, polarization, or function to enhance tumor sensitivity to ICIs has emerged as a potential therapeutic strategy worth exploring. However, merely regulating TAM phenotypes is insufficient; it is therefore necessary to further investigate the upstream signaling pathways that sustain the immunosuppressive effects of TAMs.

Immunometabolism studies have revealed that metabolic reprogramming in TAMs is an important factor regulating their phenotype and function. Beyond glucose and glutamine metabolism, lipid metabolism has a particularly important role in the TME. Solid tumors such as lung cancer typically harbor a microenvironment that is lipid-rich, hypoxic, and acidic. Lipids simultaneously serve as energy substrates, precursors of signaling molecules, and structural components of cell membranes, enabling them to modulate TAM polarization and phenotype (8-10). Pro-tumor TAMs exhibit enhanced lipid uptake, lipid droplet accumulation, and increased dependence on fatty acid oxidation (FAO). These features are closely associated with suppressed anti-tumor immunity and poor response to ICI therapy. Understanding the mechanisms of TAM lipid metabolic reprogramming is therefore important for identifying potential translational therapeutic targets.

This review aims to elucidate the immunosuppressive microenvironment mediated by TAM lipid metabolic reprogramming in lung cancer, define its role in driving ICI resistance, and summarize relevant translational therapeutic strategies. First, we discuss the fundamental metabolic characteristics of TAMs in lung cancer; next, in a stepwise progression, we analyze the functions and mechanisms of lipid uptake, lipid synthesis, FAO, cholesterol metabolism, and eicosanoid metabolism in promoting immunosuppression and therapeutic resistance, clearly distinguishing ferroptosis occurring in TAMs themselves from TAM-driven ferroptosis in CD8⁺ T cells.

Furthermore, we systematically dissect the association between TAM lipid metabolic reprogramming and ICI resistance from multiple lung cancer-specific perspectives, including the histological heterogeneity between lung adenocarcinoma (LUAD) and lung squamous cell carcinoma (LUSC); characteristic microenvironmental factors in lung cancer, such as hypoxia, high lactate levels, and tobacco exposure; the spatiotemporal distribution patterns of lipid-associated TAMs; and tumor-intrinsic genomic alterations involving EGFR, KRAS (with STK11/LKB1 and KEAP1 co-mutations), and ALK. We also integrate an analysis of the interplay between TAM lipid metabolic reprogramming and neutrophil infiltration, T cell exhaustion, and the cancer-associated fibroblast (CAF) barrier. Subsequently, we summarize currently reported approaches to reverse ICI resistance, candidate predictive biomarkers of efficacy categorized by therapeutic modality, and the challenges in translational research. Figure 1 outlines the core framework of this review. We present this article in accordance with the Narrative Review reporting checklist (available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2026-0871/rc).

Figure 1.

Central thread. TAM lipid metabolic reprogramming generates an immunosuppressive microenvironment that drives immune checkpoint blockade (ICB) resistance, which can be reversed by combining lipid-metabolic targeting with anti-PD-1/PD-L1 therapy. COX2i, cyclooxygenase-2 inhibitor; FAO, fatty acid oxidation; IFN, interferon; NK, natural killer; PD-1, programmed death-1; PD-L1, programmed death ligand 1; TAM, tumor-associated macrophage.

Methods

We searched PubMed/MEDLINE, Web of Science, and Scopus for English-language, peer-reviewed articles published through June 2026. Search terms included combinations of lung cancer, tumor-associated macrophages, lipid metabolism, immune checkpoint blockade, and therapeutic resistance. Original studies and relevant systematic and narrative reviews were prioritized. The final reference set was determined by consensus among the authors. Because this was a narrative review, a formal systematic-review protocol was not applied (Table 1).

Table 1. The search strategy summary.

Items	Specification
Date of search	Up to 30 June 2026
Databases and other sources searched	PubMed/MEDLINE, Web of Science Core Collection, Scopus; supplemented by manual reference-list screening
Search terms used	(“lung cancer” OR NSCLC OR “lung adenocarcinoma” OR “lung squamous”) AND (“tumor associated macrophage” OR TAM) AND (“lipid metabolism” OR “fatty acid” OR cholesterol OR CD36 OR FASN OR SCD1 OR CPT1 OR “fatty acid oxidation” OR COX2 OR PGE2) AND (“immune checkpoint” OR PD-1 OR PD-L1 OR immunotherapy OR resistance)
Timeframe	2010– June 2026, with emphasis on 2019–2026
Inclusion and exclusion criteria	Inclusion: English-language, peer-reviewed original or review articles addressing TAM lipid metabolism and/or ICB in lung cancer or relevant tumor models. Exclusion: conference abstracts without full data, and non-peer-reviewed sources
Selection process	Conducted independently by the authors; disagreements resolved by consensus

ICB, immune checkpoint blockade; TAM, tumor-associated macrophage.

TAMs and lipid metabolic reprogramming in lung cancer: an overview

TAMs in NSCLC originate from two distinct lineages: bone marrow-derived monocytes and tissue-resident macrophages (11,12). These distinct origins contribute to the marked heterogeneity of TAM phenotypes and functions. In response to TME signals, including hypoxia, lactate, tumor-derived cytokines, and lipids, TAMs can shift along a continuum from pro-inflammatory toward pro-tumor states (13). The canonical model categorizes macrophages into glycolysis-dependent M1 macrophages and M2 macrophages that rely on FAO and oxidative phosphorylation (14,15). However, single-cell RNA sequencing and spatial transcriptomics have demonstrated that the metabolic states of TAMs are not adequately captured by a binary model; instead, they shift dynamically in response to the microenvironment and display diverse, overlapping metabolic signatures (16-18). S Specific subsets, such as lipid-associated macrophages, are strongly correlated with clinical prognosis and immunotherapy response (19,20).

Lipid metabolic reprogramming in TAMs involves five interconnected processes: (I) uptake of exogenous fatty acids and oxidized lipoproteins via scavenger receptors and transporters such as CD36; (II) de novo lipid synthesis and storage driven by the SREBP1-FASN-SCD1 axis, with lipid-droplet formation; (III) mitochondrial import and FAO of long-chain fatty acids mediated by CPT1A; (IV) cholesterol metabolism, which remodels membrane architecture; and (V) eicosanoid metabolism, which generates lipid signaling mediators. The integrated effects of these processes regulate the immunosuppressive capacity of TAMs (7,10) (Figure 2).

Figure 2.

Integrated mechanism of the four TAM lipid-metabolic axes. Uptake (CD36, transporters, tumor-derived vesicles), de novo synthesis and storage (SREBP1-FASN-SCD1, lipid droplets), oxidation (CPT1A-FAO-OXPHOS, acetyl-CoA-driven epigenetic effects), and cholesterol/lipid-raft and COX2-PGE2 signaling converge on immunosuppressive outcomes (mTORC2 suppression, type-I IFN suppression, PD-L1 up-regulation, and CD8⁺ T-cell suppression/ferroptosis). ABC, ATP-binding cassette; EV, extracellular vesicle; FAO, fatty acid oxidation; FASN, fatty acid synthase; FFA, free fatty acids; GSH, glutathione; IFN, interferon; IL, interleukin; LXR, liver X receptor; mPGES-1, microsomal prostaglandin e synthase-1; mTORC2, mammalian target of rapamycin complex 2; ox-LDL, oxidized low-density lipoprotein; OXPHOS, oxidative phosphorylation; PD-L1, programmed death ligand 1; PGE2, prostaglandin E2; PUFA, polyunsaturated fatty acids; SCD1, stearoyl-CoA desaturase 1; TAM, tumor-associated macrophage.

Lipid metabolism drives immunosuppression and ICI resistance

These lipid metabolic reprogramming pathways do not function independently but instead form an integrated lipid metabolic network. Exogenous lipids are stored while simultaneously being utilized for oxidation (8). Stored lipids can also be mobilized through lipolysis (21). The diverse lipid molecules and signaling mediators [prostaglandin E2 (PGE2), oxysterols, lipid-peroxidation products] generated from these processes ultimately exert regulatory effects on TAMs as well as other cell types within the TME (Figure 2).

Lipid uptake and lipid droplet accumulation

CD36 (also known as fatty acid translocase, FAT/SR-B2) is a major scavenger receptor that mediates uptake by TAM uptake of long-chain fatty acids and oxidized low-density lipoproteins (ox-LDL) (22,23). Its expression is commonly upregulated by nuclear receptors including PPAR-γ (24). Imported fatty acids are esterified into triglycerides by diacylglycerol acyltransferases (DGAT) and stored as lipid droplets (25). Enhanced lipid uptake and lipid droplet deposition have been validated as critical drivers of TAM differentiation, activation, and polarization toward a pro-tumoral and immunosuppressive phenotype (26). Mechanistically, excessive lipid accumulation represses mTORC2 signaling, dampens pro-inflammatory responses, and sustains an immunosuppressive microenvironment (27,28). CD36-expressing TAMs suppress type I interferon (IFN-I) signaling and impair antigen presentation and effector T-cell recruitment. This induces a cold tumor phenotype that serves as a hallmark of primary resistance to ICIs (29). Immunotherapeutic efficacy is tightly linked to IFN-I signaling, and preclinical studies suggest that combining PD-1/PD-L1 blockade with IFN-I improves anti-tumor therapeutic outcomes (30,31). Moreover, tumor-derived extracellular vesicles can encapsulate lipids and are preferentially internalized via CD36 to support pro-tumoral TAM functions. In lung cancer, lipid-associated TAMs correlate with extracellular matrix remodeling, diminished cytotoxic T cell infiltration and poor clinical prognosis (32). Collectively, these data indicate that CD36-dependent lipid uptake constitutes a promising therapeutic target to reverse ICI resistance.

De novo lipogenesis pathway

Beyond uptake, TAMs accumulate lipids via an intrinsic de novo lipid synthesis program governed principally by the upstream master transcription factor sterol regulatory element-binding protein 1 (SREBP1) (33,34). Activation of SREBP1 concurrently upregulates the expression of fatty-acid synthase (FASN) and stearoyl-CoA desaturase 1 (SCD1) (35,36). This synthetic pathway can alter antioxidant reserves, including glutathione, and thereby influence alternative macrophage activation (33,37). Within the oxidatively stressed TME, such consumption predisposes TAMs to polarize toward an M2-like phenotype and sustain a persistent immunosuppressive function (33). FASN generates saturated fatty acid precursors, which are subsequently desaturated under the catalysis by SCD1 to produce monounsaturated fatty acids (38,39). This process alters the composition and unsaturation level of membrane lipids and may promote tumor progression (40,41). This further disrupts membrane receptor clustering, impairs Toll-like receptor signaling transduction, and reshapes the expression profile of inflammatory genes (42,43). In LUAD, transcriptional signatures constructed from lipogenesis-related genes have demonstrated predictive value for immunotherapy response (44,45). Collectively, these findings reveal that TAM lipid pools arise from two main sources: exogenous uptake mediated by CD36 and endogenous de novo synthesis dependent on the SREBP1-FASN-SCD1 axis. This mechanism ensures ample lipid substrate availability. CD36-mediated free fatty acid uptake has also been shown to regulate SREBP1-dependent de novo lipogenesis (46,47). This crosstalk indicates the potential of therapeutic strategies focused on lipid metabolic reprogramming.

FAO

TAMs are highly reliant on FAO for energy production (48,49). Carnitine palmitoyltransferase 1A (CPT1A) transports long-chain fatty acids into mitochondria for β-oxidation, thereby supporting oxidative phosphorylation (50,51). By contrast, anti-tumor immune cells such as infiltrating effector T cells depend heavily on glycolysis for energy metabolism (52,53). Hypoxia and tumor-cell glucose reprogramming restrict glucose availability (54,55). TAMs compete with these immune cells for limited substrates (56). Through their strong capacity for fatty acid uptake and oxidation, TAMs can contribute to impaired infiltration and function of antitumor immune cells (57-59). TAMs maintain intracellular energy homeostasis and immunosuppressive functions via FAO, thereby exerting persistent immunosuppressive effects (60,61). FAO intermediates such as acetyl-CoA remodel gene expression through histone acetylation and stimulate the secretion of pro-tumor cytokines including IL-1β to further facilitate tumor migration (61). Overall, upregulated FAO significantly enhances lipid metabolic activity in TAMs and contributes to a transcriptional signature of immunosuppression (49,62-64). Inhibition of CPT1A-mediated FAO disrupts energy metabolism in TAMs and promotes their pro-inflammatory phenotype (65,66). However, etomoxir has concentration-dependent off-target effects and potential hepatotoxicity (67,68). Therefore, translational research demands more selective molecules and targeted delivery approaches.

Cholesterol-mediated cellular membrane remodeling

Membrane cholesterol homeostasis is another characteristic of TAM reprogramming (69,70). Cholesterol efflux via ATP-binding cassette transporters ABCA1 and ABCG1 can drive TAM reprogramming and tumor progression (69,71,72). Meanwhile, cholesterol and its oxidized derivatives, oxysterols, act through liver X receptor (LXR) signaling to regulate lipid homeostasis and immune gene expression and to modulate TAM polarization and suppression of neighboring T cells (73-76). Cholesterol also organizes lipid rafts, membrane microdomains that act as platforms for immune receptor clustering and modulate downstream signaling cascades (77-79). Direct in vivo evidence demonstrates that hypoxia and lipid raft formation cooperatively inhibit T-cell infiltration in LUAD and are positively associated with poor prognosis. Interventions targeting lipid rafts and mitochondrial respiration via albumin-bound statins remodel the TME and potentiate the therapeutic efficacy of ICIs (80). Statins are inexpensive, clinically approved, and widely used agents. These advantages render the cholesterol-lipid raft axis a potentially tractable target for sensitizing ICIs in lung cancer (81,82). Nevertheless, prospective clinical trials are needed to fully validate this therapeutic strategy.

The COX2-PGE2 pathway

Arachidonic acid is catalyzed by cyclooxygenase-2 (COX2) and microsomal prostaglandin E synthase-1 (mPGES1) to generate PGE2 (83,84). PGE2 upregulates PD-L1 expression in TAMs and myeloid-derived suppressor cells (MDSCs), and simultaneously directly suppresses effector T-cell function (84). In addition, signaling triggered by PGE2 through E-type prostanoid (EP) receptors restricts the expansion of infiltrating stem-like CD8⁺ T cells (85-87). The self-renewal capacity of stem-like CD8⁺ T cells sustains durable therapeutic responses induced by PD-1 blockade. Accordingly, elevated PGE2 expression attenuates antitumor cytotoxicity and long-term immune memory on which ICI efficacy depends (85). Given the established immunosuppressive role of PGE2 and the broad clinical availability of COX2 inhibitors such as celecoxib (88), the combination of COX2/PGE2 pathway inhibitors with ICIs possesses robust mechanistic rationale and promising translational prospects for lung cancer.

Lipid peroxidation and ferroptosis

Lipid peroxidation and ferroptosis play a dual role in anti-tumor immunity, with their impact depending on which cell type undergoes iron-catalyzed membrane damage (89,90). For conceptual clarity, the two distinct pathways should be separated. First, ferroptosis in TAMs themselves can alleviate immunosuppression: selectively inducing ferroptosis in lipid-laden M2-like macrophages reduces their survival and secretory burden and restores the myeloid microenvironment to a more inflammatory state (91,92). Therefore, specifically increasing lipid reactive oxygen species (ROS) in TAMs through pharmacological or genetic strategies (for example, via CPT1A-associated metabolic stress or targeted iron overload in the macrophage system) may serve as a means to counteract the effects of immunosuppressive myeloid cells (19,91,92). Conversely, from the perspective of adaptive immunity, TAMs and the lipid-rich TME can promote ferroptosis in infiltrating CD8⁺ T cells. CD36-mediated uptake of oxidized lipids and polyunsaturated fatty acids induces lipid peroxidation and ferroptosis in CD8⁺ T cells, resulting in loss of effector function and impaired antitumor immunity (23,93-95). In this TAM-driven axis, macrophages function less as cell-autonomous death targets and more as sources of lipid cargo, oxidized lipoproteins, and spatial niches that compromise T cells. Such cell type-specific effects indicate that ferroptosis-targeted interventions must balance cytotoxicity against immunosuppressive TAMs with the protection of effector T cells, thus requiring a high degree of cell selectivity (96-98). In lung cancer, intraoperative and single-cell data have demonstrated that lipid-associated macrophage subsets correlate with T cell exclusion and poor prognosis (32,99,100), and CPT1A-targeted metabolic intervention can combine ferroptosis modulation with immunotherapeutic benefit in lung cancer models (19). However, direct spatial evidence showing the co-localization of lipid-laden TAMs and ferroptotic CD8⁺ T cells within the human NSCLC microenvironment is still lacking. Integrating single-cell RNA sequencing with spatial transcriptomics and lipid imaging is essential to further map these two ferroptosis programs and guide the design of cell-selective combination strategies in NSCLC, rather than relying solely on generalized rules from solid tumors.

TAMs lipid metabolism in lung-cancer-specific context

Lung cancer exhibits heterogeneity in both histology and the microenvironment, with diverse cell types and spatial architectures (101-103). Single-cell RNA sequencing studies have revealed markedly different immune landscapes in LUAD and LUSC, including differences in TAM abundance, phenotype, and metabolic state (104,105). Therefore, TAM lipid metabolic reprogramming should be understood separately according to histological subtype, and interventions aimed at reversing ICI resistance should be investigated in a subtype-specific manner, rather than being approached as a uniform entity.

Hypoxia

Hypoxia is a hallmark feature in lung tumors and can rewire TAM lipid handling. Hypoxia-inducible factor-1α (HIF-1α) regulates phagocytic receptor-related programs, lipid droplet biogenesis, and SPP1-associated macrophage states. Perilipin-2 (PLIN2) can act downstream of the HIF-1α/SPP1 axis to promote triglyceride storage and accelerate tumor progression (80,99,100,106). Furthermore, hypoxia cooperates with cholesterol-rich lipid rafts to exclude T cells in LUAD, thereby constituting a structural membrane-associated mechanism of metabolic immunosuppression that can be remodeled by statins (80,106). Thus, hypoxia is not merely a background stressor but a regulatory factor closely linked to lipid storage, lipid droplet architecture, and spatial T cell exclusion in NSCLC.

Lactate and extracellular acidosis

High lactate and extracellular acidosis further couple glycolytic tumor metabolism to myeloid lipid phenotypes. Lactate import and monocarboxylate transporter activity reshape macrophage polarization, and lactate can serve both as a fuel and as a substrate for lactylation-linked epigenetic regulation that reinforces immunosuppressive gene programs (107,108). In the nutrient-deprived and acidic intrapulmonary TME, lactate-educated TAMs exhibit a preference for FAO and lipid droplet accumulation, while impairing effector T cell function through metabolic competition and inhibitory mediators (56,107-110). These observations suggest that lactate is not merely a by-product of glycolytic tumor metabolism but may also contribute to the macrophage lipid reprogramming that sustains ICI resistance.

Tobacco exposure

Tobacco exposure is a major etiological feature of many NSCLC cases, particularly LUSC and smoking-associated LUAD. Smoking reshapes macrophage-tumor interactions through oxidative stress, lipid carbonylation, and altered soluble factors that influence myeloid activation. Smoking-induced metabolic crosstalk, including hyperglycemia and IGF2-associated PD-L1 induction in macrophage-cancer cell circuits, illustrates how smoking-related systemic and local metabolic stress can reinforce immune evasion and resistance to immune checkpoint blockade (17). Oxidized lipoproteins and lipid peroxidation products, enriched under smoke-related oxidative stress, promote CD36-dependent uptake pathways in both myeloid cells and T cells, which converge with the uptake and ferroptosis axes described above (22,23,29). Thus, integrating smoking history with histological data can facilitate more accurate identification of TAM lipid pathways associated with resistance to immune checkpoint blockade in individual patients.

Cross-tumor multi-omics studies have associated lipid-laden TAMs, including SPP1⁺ macrophages, with tumor progression (99). In ovarian cancer, PLIN2 promotes intracellular lipid accumulation in macrophages via the HIF-1α/SPP1 signaling axis and accelerates tumor progression (100). Moreover, in NSCLC, SPP1⁺ macrophages spatially co-localize and interact with FAP⁺ CAFs (111). Collectively, these cell populations contribute to an immunosuppressive microenvironment that impairs T-cell function through physical barriers and intercellular communication (112). Lipid-associated macrophages are also governed by spatially directed metabolic microenvironments, where local hypoxia, nutrient gradients, and matrix architecture synergistically determine TAM metabolic states and T cell exclusion (113-118).

It has also been reported that in NSCLC, lipid rafts within the hypoxic microenvironment suppress T cell infiltration and are associated with poor clinical outcomes (80,106). In LUAD, a molecular clustering framework based on lipid metabolism has been employed to screen candidate drugs, such as the WEE1 inhibitor MK1775 (119). Furthermore, subpopulations with an immunologically ‘cold’ phenotype are enriched across multiple driver-defined LUAD subgroups, including EGFR-mutant disease (120,121). These observations suggest that the intrinsic genetic background of tumors is closely linked to TAM metabolic status, and together they shape the drug-resistant TME (Figure 3).

Figure 3.

Lung-specific lipid-laden TAMs and the spatial metabolic niche. Single-cell-defined lipid-laden TAM subsets (e.g., SPP1⁺ macrophages) in LUAD/LUSC and their spatial co-localization with FAP⁺ fibroblasts within hypoxia-, lactate-, and tobacco-conditioned niches that favor lipid-droplet accumulation, lipid-raft remodeling, and CD8⁺ T cell exclusion. Driver contexts are annotated differentially: EGFR-mutant immunosuppressive microenvironment; KRAS with STK11/LKB1 ± KEAP1 co-mutation as a metabolically driven resistant, myeloid-suppressive state; and ALK rearrangement as a clinically critical genotype with still-limited TAM lipid mechanistic resolution. ALK, anaplastic lymphoma kinase; EGFR, epidermal growth factor receptor; HIF1α, hypoxia-inducible factor 1-alpha; LUAD, lung adenocarcinoma; LUSC, lung squamous cell carcinoma; TAM, tumor-associated macrophage.

Tumor-intrinsic genomic heterogeneity coupling lipid metabolism in ICIs resistance

Tumor-intrinsic driver alterations not only promote cell proliferation but also rewire metabolism to facilitate immune evasion (122,123). Therefore, the role of TAMs must be interpreted in the context of tumor genotype. The following section highlights the differences in TAM lipid and myeloid features across common oncogenic driver contexts in NSCLC.

EGFR mutation

EGFR-mutant NSCLC is generally associated with an immunologically “cold” TME, characterized by limited CD8⁺ infiltration, reduced antigenicity after prolonged tyrosine kinase inhibitor (TKI) therapy, and low overall response rates to PD-1/PD-L1 monotherapy blockade (120,121,124,125). Following TKI resistance, the dynamic remodeling of myeloid cells and stromal architecture further limits the efficacy of ICIs. Although direct large-scale analyses of the TAM lipidome in EGFR-mutant tumors are still incomplete, the cold myeloid cell-enriched microenvironment, lipid raft/hypoxia programs, and the reliance of immunosuppressive TAMs on the CD36/FAO pathway collectively suggest a potential role for TAM lipid programs in resistance but do not yet establish an EGFR-specific metabolic axis (126). Genotype-stratified spatial multi-omics studies are needed before EGFR-specific lipid nodes can be considered drug targets, rather than relying on extrapolations from wild-type models alone (120,121).

KRAS mutation

KRAS-mutant LUAD, particularly in the presence of co-occurring STK11/LKB1 or KEAP1 alterations, constitutes the most compelling genotype-metabolism-immunity paradigm. A substantial body of literature has demonstrated that STK11/LKB1 mutations are among the genomic alterations associated with resistance to PD-1/PD-L1 inhibitors in KRAS-mutant LUAD (127-129). These tumors typically exhibit low PD-L1 levels, reduced effector T cell infiltration, and an immunosuppressive TME with elevated proportions of neutrophils and monocytic suppressor cells (127,130). LKB1 is an upstream kinase of AMP-activated protein kinase (AMPK); its loss disrupts AMPK-dependent metabolic control, alters lipid metabolism and ferroptosis susceptibility, and reshapes the myeloid compartment (130,131). Co-occurring KEAP1 mutations activate the NRF2 antioxidant pathway and glutamine dependency, enhance resistance to oxidative stress and ferroptosis, and suppress the STING/interferon pathway—pathways that, in TAMs, run parallel to CD36-mediated IFN-I suppression (132-137). ATR inhibition combined with ICIs can partially reverse the myeloid-suppressive phenotype in STK11/KEAP1-mutant NSCLC, while metabolic conditioning (e.g., metformin or caloric restriction) has shown benefit in LKB1-mutant models (138,139).

ALK mutation

ALK-rearranged NSCLC represents another oncogene-addicted subset in which first-line treatment centers on ALK TKIs rather than single-agent ICIs. Compared with KRAS/STK11-mutant disease, high-resolution evidence linking ALK fusions to specific TAM lipid enzymes is still lacking. Current data support a relatively excluded T cell microenvironment under oncogene-addicted conditions and caution against the routine use of single-agent ICIs; however, there remains no clear evidence as to whether ALK-driven metabolic by-products, apoptosis-derived lipids, or treatment-perturbed myeloid states specifically activate the CD36-FAO or COX2-PGE2 pathways (125,140). Consequently, ALK rearrangement should currently be viewed primarily as a clinically relevant stratification variable, rather than as definitive evidence of a defined TAM lipid metabolic axis. In ALK-positive NSCLC, metabolic-ICI combination strategies may be more appropriately investigated in specific clinical contexts, such as after the emergence of resistance to ALK-targeted therapy. More broadly, integrating the genomic status of EGFR, KRAS, STK11, KEAP1, and ALK with TAM lipid biomarkers can help distinguish established genotype-metabolism associations from exploratory ones and identify patient subgroups that are most likely to benefit from metabolic-ICI combination strategies (Figure 3).

These molecular subgroups should not be regarded as sharing a uniform genotype-metabolism-immunity program. EGFR-mutant NSCLC predominantly manifests an immune-excluded, T cell-poor microenvironment, and the evidence linking EGFR alterations to specific TAM lipid programs remains preliminary. In contrast, KRAS-mutant tumors, particularly those with concurrent STK11/LKB1 or KEAP1 alterations, provide the strongest evidence for a metabolic-myeloid axis involving aberrant fatty acid synthesis, ferroptosis-related vulnerabilities, neutrophil-mediated immunosuppression, and impaired IFN-I signaling. In ALK-rearranged NSCLC, therapeutic decisions primarily concern the sequencing of ALK-targeted therapies, and the role of TAM lipid metabolism remains insufficiently elucidated. Therefore, biomarker development should integrate driver gene status with direct measurements of myeloid lipid activity, rather than extrapolating metabolic mechanisms or therapeutic targets across distinct oncogenic subgroups (127,128,130-139).

Integration with other resistance mechanisms

TAM lipid reprogramming does not operate in isolation but intersects with multiple classical mechanisms of ICI resistance, forming a multi-layered network (Figure 4).

Figure 4.

Crosstalk network linking TAM lipid metabolism with glycolysis/hypoxia-lactate stress, neutrophil/MDSC infiltration, T-cell exhaustion (including PGE2 and CD36-linked ferroptosis routes), and CAF/matrix barriers, together with paracrine lipid mediator networks. The schematic illustrates a multi-node resistance network and the rationale for genotype-aware, multi-target combinations. CAF, cancer-associated fibroblast; DC, dendritic cell; EV, extracellular vesicle; FAO, fatty acid oxidation; G, glucose; MDSC, myeloid-derived suppressor cell; NK, natural killer; PGE2, prostaglandin E2; TAM, tumor-associated macrophage.

Metabolic competition under hypoxic and acidic conditions

In the nutrient-deprived pulmonary TME, lipid metabolism is closely intertwined with glycolysis and hypoxic adaptation (107,109,110). These metabolic programs support the persistence and suppressive function of myeloid cells and intensify competition for glucose and oxygen, thereby diminishing the function and survival of effector T cells (108).

Neutrophil and MDSC infiltration

Tumor genotypes such as STK11/LKB1 loss demonstrate that the TME recruits neutrophils and monocytic MDSCs, which share lipid uptake and FAO dependency with TAMs (130,141-144). CD36 and FAO jointly support the suppressive function across this myeloid lineage; therefore, a lipid-targeting agent may simultaneously act on TAMs, MDSCs, and their associated subsets—offering broad therapeutic potential while also necessitating attention to infection risk and the safety of tissue-resident macrophages (144).

T cell exhaustion and effector dysfunction

Beyond acute metabolic starvation, chronic antigen stimulation and lipid ROS drive CD8⁺ T cells toward an exhausted phenotype marked by the expression of inhibitory receptors and the loss of stem-like precursor cells (9,145). The PGE2-EP signaling axis restricts the PD-1 blockade-dependent expansion of stem-like CD8⁺ cells (85-87), while CD36-driven T cell ferroptosis provides a metabolic pathway leading to T cell functional collapse (23,93-95). Thus, TAM lipid programs contribute to both the initiation and the eventual maintenance of T cell dysfunction.

CAF barrier and matrix remodeling

In NSCLC, lipid-laden SPP1⁺ macrophages interact with FAP⁺ CAFs to form a fibrotic and chemorepulsive border that segregates T cells from tumor nests (111,112). Paracrine lipid messengers and extracellular vesicles produced by the TAM-CAF unit can further amplify the suppression of NK cells and dendritic cells (146-152). This stromal mechanism explains why neutralization of a single immune checkpoint on T cells is often of limited efficacy without disruption of the lipid-matrix microenvironment.

Taken together, the operational resistance network in NSCLC can be summarized as: tumor-intrinsic drivers (EGFR/KRAS-STK11/KEAP1/ALK) × TAM lipid axis (uptake-synthesis-FAO-cholesterol/PGE2-ferroptosis) × microenvironment (neutrophils/MDSCs, exhausted T cells, CAF barrier) interacting under lung-specific conditions of hypoxia, lactate, and tobacco exposure. Therefore, combination therapies should be designed as multi-nodal, genotype-based interventional strategies rather than single-agent blockade approaches (Figure 4).

Targeting lipid metabolism to reverse ICI resistance

Current strategies targeting lipid metabolism to reverse immunotherapy resistance mainly fall into three categories: single-target lipid-metabolism inhibitors combined with ICIs, dual-target therapeutic strategies, and targeted drug delivery systems. These approaches remodel the immune microenvironment and modulate immune responses via distinct lipid metabolic pathways, thereby potentiating the efficacy of immunotherapy (Figure 5).

Figure 5.

Strategies to reverse ICB resistance by targeting lipid metabolism: single-target combinations with anti-PD-1/PD-L1, dual-target/combination approaches addressing uptake-synthesis compensation, and TAM-specific delivery, culminating in a cold-to-hot tumor conversion. FFA, free fatty acids; ICB, immune checkpoint blockade; IFN, interferon; ox-LDL, oxidized low-density lipoprotein; PD-L1, programmed death ligand 1; SFA, saturated fatty acid; TAM, tumor-associated macrophage.

Single targets combined with anti-PD-1/PD-L1

Targeting individual lipid-metabolic nodes (CD36, FASN, SCD1, CPT1, COX2, cholesterol efflux) together with PD-1/PD-L1 blockade yields additive benefit in preclinical models by reversing M2 polarization, restoring IFN-I signaling, lowering PD-L1, and relieving CD8⁺ T-cell suppression—collectively converting a ‘cold’ tumor toward a ‘hot’, ICI-responsive phenotype (144). In lung-cancer models, the WEE1 inhibitor MK1775 improves anti-PD-1 efficacy by modulating lipid crosstalk among tumor cells, TAMs, and CD8⁺ T cells (119); albumin-bound statins targeting lipid rafts and mitochondrial respiration enhance PD-1 blockade in NSCLC (106); and the DPP4 inhibitor anagliptin reduces TAM abundance and M2 polarization to augment anti-PD-L1 activity, including in resistant settings (153). Extending the genomic-metabolic logic above, ATR inhibition combined with PD-(L)1 blockade reprograms the suppressive myeloid compartment of STK11/KEAP1-altered tumors (137). A consolidated target-agent-evidence summary is provided in Table 2.

Table 2. Lipid-metabolic targets, representative agents, mechanism and evidence in combination with immune checkpoint blockade, evidence stage, and references.

Target (lipid axis)	Representative agent/intervention	Mechanism and evidence in combination with ICB	Stage	Reference
CD36 (uptake)	CD36 inhibitor (e.g., SSO)/anti-CD36 antibody	Blocks fatty-acid and ox-LDL uptake, reduces lipid-droplet accumulation, restores type-I IFN signaling and reverses M2 polarization; additive benefit with anti-PD-1/PD-L1 (cold-to-hot conversion)	Preclinical	(11,12,21,28)
FASN (de novo synthesis)	FASN inhibitor (e.g., TVB-2640/denifanstat, orlistat)	Suppresses de novo fatty-acid synthesis, relieves antioxidant depletion and M2 lock-in; preclinical synergy and modulation of tumor-TAM-CD8 lipid crosstalk	Preclinical; FASNi in early-phase trials	(15,21,23)
SCD1 (desaturation)	SCD1 inhibitor/siRNA	Inhibits fatty-acid desaturation and membrane remodeling; combined CD36 inhibition + SCD1 silencing blocks uptake-synthesis compensation and outperforms single targets	Preclinical	(28)
CPT1A/FAO (oxidation)	CPT1 inhibitor (etomoxir, perhexiline)	Blocks mitochondrial fatty-acid oxidation, undercutting the energetic basis of the M2 phenotype (note: high-dose etomoxir has off-target/hepatic toxicity)	Preclinical	(16,21)
COX2/PGE2 (eicosanoids)	COX2 inhibitor (celecoxib); EP-receptor antagonists	Lowers PGE2, down-regulates PD-L1 on TAMs/MDSCs, restores effector expansion of stem-like CD8+ T cells; mechanistically tied to ICB resistance	Preclinical + clinical exploration	(18,19)
Cholesterol efflux/lipid raft	Statins (albumin-bound); LXR modulators; ABCA1/ABCG1	Reduce membrane cholesterol and lipid rafts, reprogram TAMs and improve T-cell infiltration; enhance PD-1 blockade in NSCLC; statins are repurposable approved drugs	Preclinical; statins approved (other use)	(14,24)
Lipid-crosstalk node (WEE1)	MK1775/adavosertib	Modulates lipid crosstalk among tumor cells, TAMs and CD8+ T cells; improves anti-PD-1 efficacy in LUAD models	Preclinical	(23)
TAM differentiation (DPP4)	Anagliptin (DPP4 inhibitor)	Inhibits monocyte-to-macrophage differentiation and M2 polarization; enhances anti-PD-L1 efficacy in NSCLC, including resistant settings	Preclinical	(27)

ICB, immune checkpoint blockade; IFN, interferon; LUAD, lung adenocarcinoma; MDSC, myeloid-derived suppressor cell; NSCLC, non-small cell lung cancer; ox-LDL, oxidized low-density lipoprotein; PD-1, programmed cell death 1; PD-L1, programmed death-ligand 1; SSO, sulfo-N-succinimidyl oleate.

Combination and dual-target strategies

Preclinical evidence suggests that compensatory crosstalk between exogenous fatty-acid uptake and endogenous lipogenesis may limit the efficacy of single-node inhibition (154,155). Combined blockade of uptake and synthesis—for example, a CD36 inhibitor plus SCD1 silencing—outperformed single targets in refractory tumor models, providing proof of concept for dual-target design that is readily transferable to lung cancer (154). Beyond dual metabolic targeting, lipid-directed agents may also be paired with conventional modalities (chemotherapy, radiotherapy, and anti-angiogenic therapy) that themselves alter the metabolic and immune milieu, or with metabolic ‘conditioning’ approaches (e.g., metformin or dietary intervention) that shift the TME toward a hot state (139). Rational combinations should be guided by the compensatory-network biology of the specific tumor genotype and, wherever possible, designed to spare effector T cells from lipid-peroxidation-driven dysfunction.

Macrophage-specific delivery

Because most lipid-metabolic targets are not TAM-specific, systemic blockade may perturb normal metabolic tissues and other immune subsets, thereby narrowing the therapeutic window. This concern is exemplified by high-dose etomoxir, whose effects include CPT1-independent disruption of CoA homeostasis, mitochondrial complex-I inhibition, oxidative stress, and potential hepatotoxicity (68,156). Macrophage-directed delivery is therefore a central engineering priority. Nanoparticles, liposomes, and ligands targeting scavenger receptors or mannose receptors can enrich metabolic modulators within TAMs, improving the therapeutic index and potentially enabling agents that would be intolerable systemically (92,157-160). Such delivery platforms can also co-package a metabolic modulator with an immune agonist to reprogram TAMs and stimulate adaptive immunity simultaneously (161).

Clinical status and translational barriers

Direct clinical evidence for combining TAM lipid-targeted interventions with ICIs remains limited, and most available data derive from preclinical models or observational studies of widely used agents such as statins, COX2 inhibitors, and metformin (81,162,163). Prospective clinical validation of TAM lipid biomarkers in NSCLC is therefore needed to support biomarker-based patient selection and target validation (9,10). The major obstacles currently facing translational therapy include: compensatory and escape mechanisms arising from metabolic plasticity; confounding by systemic metabolic factors—such as obesity, dyslipidemia, diet, and concomitant medications like statins and metformin—which may either bias observational data or present opportunities for intervention; and the potential for on-target toxicities in normal macrophages and other tissues (164). Addressing these issues requires well-designed correlative studies embedded within combination therapy trials.

Candidate efficacy biomarkers for the TAM lipid-ICI axis in NSCLC

The biological insights described above require fit-for-purpose, stratified biomarkers—baseline indicators for predicting ICI efficacy, pharmacodynamic markers for monitoring metabolic reprogramming, and markers for enriching lipid-targeted combination therapy. Tissue-based and spatial biomarkers include lipid-laden SPP1⁺ TAMs (macrophages), PLIN2⁺ lipid droplet macrophages, CD36 expression on myeloid cells and T cells, and the hypoxia-lipid raft signature reflecting T cell exclusion (29,32,80,99,100,106,111,117,118). These detection methods should be integrated with multiplex immunohistochemistry or spatial transcriptomics to help identify which “cold” NSCLCs are metabolically locked in a myeloid state rather than merely exhibiting low PD-L1 expression. Circulating and systemic molecular markers include plasma lipidomic profiles, circulating lipoprotein signatures, tumor-derived exosomal lipid cargo, and transcriptional lipid synthesis or FAO modules previously explored in association with immunotherapy in LUAD (44,45). Genomic companion markers should be scored simultaneously: EGFR mutations and ALK rearrangements indicate that single-agent ICI therapy warrants caution and require dedicated metabolic annotation (118,124,140). Co-mutations of KRAS with STK11/LKB1 and/or KEAP1 identify a myeloid-suppressive, ferroptosis-modulated state that may be amenable to metabolic-ICI combination therapy (120,121,127-139). Functional immune indicators—such as IFN-I/STING pathway activity, the ratio of stem-like to exhausted CD8⁺ cells, and PGE2 pathway activation—link lipid circuits to the mechanism of action of ICIs (29-31,85-87,137). Prospectively, a composite score integrating genotype, myeloid lipid phenotype, and IFN-I/T cell status may be more informative than any single indicator (9,145). Validation requires standardized pre-analytical definitions of the lipid-enriched TAM state, multi-cohort calibration within driver gene subgroups, and the embedding of biomarker endpoints in early-phase combination therapy trials.

Limitations

Its limitations arise from the narrative design: (I) The literature search prioritizes comprehensiveness over exhaustiveness, which may introduce selection bias. (II) A substantial body of mechanistic and therapeutic evidence derives from preclinical models or non-lung cancer tumors and is extrapolated to lung cancer, with markedly uneven evidence across EGFR/ALK and KRAS/STK11 contexts. (III) The field is evolving rapidly, and conclusions should therefore be reassessed as direct lung cancer and clinical data accumulate.

Conclusions

This review integrates lipid metabolic mechanisms, lung cancer-specific factors (including hypoxia, lactate, and tobacco exposure), genomic backgrounds, multi-omics evidence, interactions with other resistance pathways, biomarker candidates, and translational strategies into a unified framework centered on ICI resistance and provides an actionable target-drug-evidence summary.

TAM lipid metabolic reprogramming is a core driver of the immunosuppressive microenvironment and ICI resistance in lung cancer. Uptake, de novo synthesis, oxidation, and the cholesterol/eicosanoid axis form an interconnected network that locks TAMs in a pro-tumor phenotype—suppressing IFN-I signaling, upregulating PD-L1 expression, promoting exhaustion and ferroptosis-related dysfunction in effector T cells, and ultimately blunting the efficacy of anti-PD-1/PD-L1 therapy (7,9,10). In lung cancer, this metabolic-immune circuit is further reinforced by lipid-laden TAM subsets (such as SPP1⁺ macrophages), the spatial microenvironment shaped by hypoxia, lactate accumulation, and tobacco-related oxidative stress, CAF-associated barriers, cooperative suppression by neutrophils/MDSCs, and tumor-intrinsic genomic backgrounds encompassing EGFR, KRAS/STK11/KEAP1, and ALK, thereby coupling metabolic status to a “cold” and drug-resistant state (117,126,129). Targeting these critical nodes through rationally designed dual-target combinations, repurposed drugs, and macrophage-directed delivery systems, together with patient selection based on candidate predictive biomarkers, holds promise for reversing resistance and enabling a “cold”-to-”hot” conversion. Achieving this goal will require generating direct lung cancer and human spatial data that can distinguish TAM ferroptosis from T cell ferroptosis, developing cell-selective agents, integrating metabolic and genomic biomarkers, and testing rational combinations in biomarker-guided trials with embedded correlative studies. With these advances, TAM lipid metabolic reprogramming may evolve from a mechanistic concept into a precise and clinically applicable strategy for overcoming immunotherapy resistance in NSCLC.

Supplementary

The article’s supplementary files as

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.