Activating mutations in the epidermal growth factor receptor define a major molecular subset of non-small cell lung cancer, accounting for roughly 10 to 15 percent of cases in Western populations and 40 to 50 percent in East Asian cohorts. EGFR tyrosine kinase inhibitors transformed the outlook for these patients, producing response rates and survival that far exceed what chemotherapy can achieve. Yet resistance is nearly universal. Tumors that initially melt away under treatment almost inevitably evolve escape routes, including secondary EGFR mutations, bypass signaling through other receptor pathways, epithelial-mesenchymal transition and lineage plasticity. The new study adds a previously underappreciated mechanism to this list: metabolic reprogramming of drug-insensitive cells by the tumor’s own stromal environment.

The research team, led by investigators at Tianjin Medical University Cancer Institute and Hospital, began with a deceptively simple observation. Although EGFR-mutant tumors are usually treated as genetically uniform, they frequently contain subclones of tumor cells that lack the mutation altogether. These EGFR-wild-type cells may pre-exist within the tumor or emerge under the selective pressure of therapy, and amplification of wild-type EGFR alleles has already been documented as a route to acquired resistance against third-generation inhibitors. What remained unclear was how these wild-type cells manage to persist and eventually dominate residual disease.

To answer that question, the researchers isolated mesenchymal stromal cells from the tumors and paired non-tumorous lung tissue of patients with EGFR-mutant and EGFR-wild-type non-small cell lung cancer. The cells displayed the classic stromal phenotype, expressing CD73, CD90, CD105 and CD166 while lacking hematopoietic markers such as CD34, CD45 and HLA-DR. Transcriptomic comparison revealed that mesenchymal cells derived from EGFR-mutant tumors were strikingly different from their counterparts in adjacent healthy lung tissue, with more than 2,400 differentially expressed genes and marked enrichment of inflammatory signaling, complement cascades and extracellular matrix remodeling pathways. Among the most prominent changes were elevated levels of the cytokines interleukin-6 and interleukin-1 alpha.

When the team exposed EGFR-wild-type lung cancer cell lines to conditioned medium from these tumor-derived stromal cells, the results were unambiguous. The tumor cells upregulated S100A9, an inflammatory calcium-binding protein, more dramatically than with any other treatment, and their metabolism shifted decisively toward de novo fatty acid synthesis. Targeted metabolomics showed accumulation of citrate and oxaloacetate, key intermediates of the citrate shuttle that supplies cytosolic acetyl-CoA for lipid production, along with increased pools of saturated, monounsaturated and polyunsaturated free fatty acids. Mechanistically, blocking interleukin-6 with tocilizumab or interleukin-1 alpha signaling with a receptor antagonist reduced these effects, and pharmacologic inhibition of STAT3 with stattic confirmed that the canonical interleukin-6 downstream pathway was directly driving S100A9 transcription through confirmed binding sites in the gene’s promoter.

The signaling cascade continued downstream of S100A9. Knockdown experiments showed that silencing S100A9 reduced the elevated free fatty acid levels and lowered the expression of c-Myc, beta-catenin, fatty acid synthase and the glucose transporter GLUT1. Because S100A9 signals through Toll-like receptor 4 and the receptor for advanced glycation end products, the team tested specific inhibitors of both receptors and found that blocking either one attenuated the downstream transcriptional program. Further experiments established that beta-catenin regulates c-Myc expression, that c-Myc binds directly to the promoters of fatty acid synthase, acetyl-CoA carboxylase and ATP citrate lyase, and that beta-catenin controls GLUT1 through a c-Myc-independent route. Together these transcription factors activated the citrate-acetyl-CoA-malonyl-CoA axis that fuels lipid production.

Perhaps the most striking consequence of this metabolic rewiring was the acquisition of stem-like traits. Tumor cells exposed to the conditioned medium increased their expression of OCT4, SOX2, CD44 and beta-catenin, hallmarks of cancer stemness, and these changes were reversed when fatty acid synthase was silenced. Because fatty acid synthesis is tightly coupled to maintenance of a stem-like state, the inflammatory signal from stromal cells effectively transformed relatively drug-insensitive wild-type tumor cells into a more resilient, less proliferative population primed to survive therapy. In vivo, co-implantation of EGFR-wild-type A549 cells with tumor-derived mesenchymal stromal cells in immunocompromised mice significantly accelerated tumor growth compared with co-implantation of paired tumor-free stromal cells or tumor cells alone, and the resulting tumors showed elevated levels of both metabolic and stemness markers.

To model the clinical situation more faithfully, the researchers added EGFR-mutant PC9 cells to the co-implantation system, creating mixed tumors that contained both cell populations. When mice received daily osimertinib, the mutant cells died as expected, but the wild-type cells, particularly in the presence of tumor-derived stromal cells, persisted and eventually constituted the largest residual population. Combining osimertinib with tocilizumab, an antibody that blocks the interleukin-6 receptor, produced a synergistic effect that suppressed the growth of both cell types and significantly reduced the proportion of mutant cells in residual tumors. This finding suggests that targeting the stromal inflammatory signal, rather than the tumor cell directly, can resensitize resistant disease to standard therapy.

The researchers then turned to human tissue to see whether the mechanism operates in actual patients. Using multiplex immunofluorescence and serial immunohistochemistry on samples from 23 patients with EGFR-mutant non-small cell lung cancer who had received adjuvant tyrosine kinase inhibitor treatment, they quantified the composition and spatial organization of tumor cells and stromal cells. Although all tumors were classified as EGFR-mutant, EGFR-wild-type tumor cells constituted the majority of the tumor cell population, with a median proportion of 66.78 percent compared with 33.22 percent for mutant cells. When patients were stratified by treatment response, resistant tumors showed dramatically higher expression of fatty acid synthase, S100A9 and the stemness marker SOX2 within their wild-type tumor cells than sensitive tumors did, with median positive fractions of roughly 46 percent, 47 percent and 36 percent versus 2 percent, 6 percent and 5 percent respectively.

Spatial analysis added another layer of insight. In approximately 60 percent of the tumor area, tumor-derived mesenchymal stromal cells formed continuous band-like structures wrapping around tumor nests, while in the remainder they were scattered individually. The density of this peritumoral wrapping was significantly higher in drug-resistant samples than in sensitive ones. Nearest-neighbor distance calculations revealed that EGFR-mutant tumor cells were consistently located closer to the stromal cells than wild-type cells were, regardless of treatment response, suggesting an intrinsic spatial relationship that facilitates paracrine education of the stromal population. These patterns indicate that the tumor microenvironment is not a passive backdrop but an architect of resistance, physically and chemically shaping which cells survive therapy.

The study does not resolve every question. The authors note that their in vivo model with interleukin-6 blockade does not exclude contributions from direct cell-cell contact, and that TKI-treated patient specimens were not available for spatial validation of the mechanism after therapy. They also observed that tumor-derived stromal cells shared metabolic features with bone marrow-derived and umbilical cord-derived mesenchymal stromal cells rather than with paired tumor-free stromal cells, raising unresolved questions about the ontogeny and functional diversity of these populations. The extracellular metabolic consequences of fatty acid synthase inhibition, including increased extracellular free fatty acids and lactate and decreased glucose, suggest further complexity in how stromal-tumor metabolic crosstalk reshapes the local environment and influences immune cell function.

Nevertheless, the clinical implications are substantial. Resistance to EGFR tyrosine kinase inhibitors remains the central barrier to durable benefit in this patient population, and existing second-line strategies have focused largely on tumor-cell-intrinsic mechanisms such as secondary mutations and bypass pathways. By demonstrating that stromal cells can metabolically arm otherwise vulnerable wild-type tumor cells, the study opens a therapeutic avenue that targets the tumor microenvironment rather than the tumor cell genome. The finding that tocilizumab, a drug already approved for rheumatoid arthritis and other inflammatory conditions, synergizes with osimertinib in preclinical models suggests that clinical testing of this combination could be feasible in the near term. More broadly, the work underscores the importance of intratumoral heterogeneity and metabolic reprogramming in treatment failure, and it provides a mechanistic rationale for integrating cytokine blockade or fatty acid synthesis inhibition with targeted therapy to suppress residual disease and improve outcomes for patients with EGFR-mutant lung cancer.