Integrative Proteogenomic and Single-Cell Analysis Reveals RTK Switching and Metabolic Reprogramming as Synthetic Lethal Vulnerabilities in FGFR Inhibitor Resistance

Integrative Proteogenomic and Single-Cell Analysis Reveals RTK Switching and Metabolic Reprogramming as Synthetic Lethal Vulnerabilities in FGFR Inhibitor Resistance | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Integrative Proteogenomic and Single-Cell Analysis Reveals RTK Switching and Metabolic Reprogramming as Synthetic Lethal Vulnerabilities in FGFR Inhibitor Resistance Linghui Tan, Tianlun Hou, Pingting Ying, Xian Wang, Hongchuan Jin, and 1 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8869688/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 4 You are reading this latest preprint version Abstract Background Although fibroblast growth factor receptor (FGFR) inhibitors (FGFRi) have demonstrated clinical promise, the inevitable emergence of acquired resistance remains a distinct bottleneck, severely compromising their long-term clinical efficacy. The pan-cancer molecular landscape and heterogeneous mechanisms driving this resistance, ranging from genetic alterations to dynamic network rewiring, remain poorly understood. Methods We integrated large-scale pharmacogenomic profiling (GDSC2 and PRISM) with single-cell RNA sequencing to dissect the proteogenomic landscape of FGFRi resistance across 312 cell lines from 8 cancer types, complemented by machine learning modeling and systematic synthetic lethality screening to uncover actionable therapeutic targets. Results Our dual-database analysis unveiled a multi-dimensional atlas of FGFRi resistance. We identified cancer-specific genomic drivers, such as ELF4 amplification in glioblastoma, alongside key transcriptomic markers including UCP2 and FSCN1 , highlighting a shift towards metabolic reprogramming and epithelial-mesenchymal transition (EMT). Single-cell resolution analysis unveiled that resistance is predominantly associated with the enrichment of subpopulations harboring aberrant cell-cycle dysregulation (MP2), suggesting a model of clonal selection rather than purely transcriptional plasticity-driven adaptation. Furthermore, a Random Forest model based on 52 mRNA features was constructed, demonstrating robust predictive capability for FGFRi sensitivity (AUC > 0.7). Most notably, our synthetic lethal screening revealed a convergent reliance on compensatory RTK signaling (specifically EGFR pathway enrichment) and downstream MAPK/PI3K cascades in resistant phenotypes, providing robust evidence for an "RTK switching" mechanism. Conclusions This study establishes a high-resolution proteogenomic atlas of FGFRi resistance, identifying a convergent evolution towards metabolic reprogramming and EGFR-mediated bypass signaling. Our findings characterize resistance as a dynamic network rewiring and propose rational combination therapies (e.g., FGFRi combined with EGFR or metabolic inhibitors) to overcome resistance. FGFR inhibitor Drug resistance Pan-cancer analysis Multi-omics integration Synthetic lethality Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Background The Fibroblast Growth Factor Receptor (FGFR) family (FGFR1-4), a subset of the receptor tyrosine kinase (RTK) superfamily, governs critical physiological processes ranging from embryonic development to tissue repair[ 1 , 2 ]. Upon binding to Fibroblast Growth Factor (FGF) ligands, FGFR initiates cascading reactions of downstream signaling pathways, including PI3K-AKT and RAS-MAPK, thereby regulating cell proliferation, differentiation, migration, and survival[ 1 , 3 ]. Aberrant activation of FGFR signaling—driven by gene fusion, mutation, or amplification—has been established as a pivotal oncogenic driver across a diverse spectrum of malignancies[ 2 , 4 ]. Consequently, targeting this axis with small-molecule tyrosine kinase inhibitors (TKIs), such as the selective inhibitor AZD4547, has emerged as a cornerstone of precision oncology[ 5 – 9 ]. Despite the transformative potential of these therapies, the clinical reality remains sobering. While initial responses in patients with FGFR alterations—particularly those with urothelial carcinoma or cholangiocarcinoma—are encouraging, they are often transient[ 10 , 11 ]. The inevitable emergence of acquired resistance severely limits the durability of response, typically resulting in disease progression within months. Current paradigms of FGFR inhibitor (FGFRi) resistance are predominantly categorized into four mechanisms: (1) Secondary activating mutations in the FGFR kinase domain, leading to impaired inhibitor binding or sustained enhancement of kinase activity[ 12 – 14 ]. (2) Compensatory activation of alternative RTKs (e.g., EGFR or MET), bypassing signal transduction mediated by the FGFR pathway[ 15 , 16 ]. (3) Reactivation of downstream pathways such as PI3K-AKT and RAS-MAPK, offsetting the growth inhibitory effect after FGFR inhibition[ 12 , 17 ]. (4) Mutations in tumor suppressor genes such as TP53 , regulating cell cycle checkpoints or DNA damage repair capacity and reducing drug sensitivity[ 18 – 20 ]. Among these, 'RTK switching' has emerged as a prominent yet complex escape strategy. Notably, Wu et al. demonstrated that compensatory EGFR activation drives resistance specifically in FGFR2-fusion positive cholangiocarcinoma, providing a rationale for dual blockade[ 15 ]. However, our understanding of this phenomenon remains fragmented and confined to isolated cancer types. Critical systemic questions remain unanswered: Is this EGFR-mediated bypass a universal vulnerability across the pan-cancer landscape, or is it a context-dependent event? More importantly, does this signaling rewiring operate in isolation, or is it merely one facet of a multimodal adaptive strategy that also involves metabolic reprogramming? Crucially, existing studies predominantly focused on distinct molecular layers in isolation, thereby failing to capture the concurrent evolution of signaling networks and metabolic states. To bridge this gap, we designed a comprehensive multi-omics framework leveraging large-scale pharmacogenomic data from 312 cell lines across eight major cancer types: Bladder Urothelial Carcinoma (BLCA), Breast Invasive Carcinoma (BRCA), Colon Adenocarcinoma (COAD), Esophageal Squamous Cell Carcinoma (ESCC), Glioblastoma (GBM), Lung Adenocarcinoma (LUAD), Melanoma (MEL), and Pancreatic Adenocarcinoma (PAAD). To ensure robustness, we implemented a rigorous cross-validation framework utilizing two independent datasets, GDSC2 and PRISM. Beyond traditional bulk-level profiling, we integrated single-cell RNA sequencing (scRNA-seq) to dissect the clonal evolution of resistance at cellular resolution. Finally, we translated these biological insights into actionable strategies by constructing a machine learning-based predictive model and conducting systematic synthetic lethality screening. In this study, we present a multi-dimensional atlas of FGFRi resistance. We uncover that resistance is driven not only by genetic alterations but also by a convergent evolution involving metabolic reprogramming (e.g., UCP2 upregulation) and RTK switching (specifically towards EGFR signaling). By integrating these mechanistic insights with synthetic lethal vulnerability mapping, we propose a rational framework for tiered precision combination therapies, offering new avenues to overcome the clinical bottleneck of FGFRi resistance. Methods Download of cell line multi-omics data and AZD4547 sensitivity data Multi-dimensional omics data of pan-cancer single cell lines, including gene mutation data, copy number variation (CNV) data, DNA methylation data, transcriptome sequencing data, and proteome quantification data, were systematically downloaded from the DepMap database ( https://depmap.org/portal/ ). Meanwhile, drug response data from the Genomics of Drug Sensitivity in Cancer (GDSC2) database ( https://www.cancerrxgene.org/ ) of the Sanger Institute and the PRISM Repurposing Public 24Q2 database ( https://depmap.org/portal/prism/ ) of the Broad Institute were obtained. FGFR-abnormal cell line screening criteria Based on gene variation annotation information, tumor cell lines with FGFR abnormalities were screened. The inclusion criteria included: point mutations, covering functional impact mutations such as frameshift_variant, splice_acceptor_variant, splice_donor_variant, start_lost, stop_gained, stop_lost, protein_altering_variant, and missense mutation; copy number variations, gene amplification with copy number ≥ 4; gene fusions, cell lines with FGFR family gene fusion events. In the PRISM database, drug sensitivity was represented by the log-fold change (LFC), calculated as: LFC = log₂(number of viable cells in the experimental group / number of viable cells in the control group). An LFC < 0 indicates that the drug has an inhibitory effect on the cell line, and a smaller LFC value indicates a stronger inhibitory effect. To maximize the biological contrast between responsive and non-responsive phenotypes, the grouping criteria were strictly defined: LFC > 0 as the FGFR inhibitor (FGFRi) resistant group, and LFC < 0 as the FGFRi sensitive group. To ensure statistical test power and robust feature extraction, only cancer types with ≥ 3 cell lines in both the sensitive and resistant groups were retained for subsequent analysis. Crucially, to mitigate the potential bias from this binary classification, all candidate markers derived from this grouping were further cross-validated using continuous drug response metrics (AUC) from the independent GDSC2 database to ensure the robustness of the findings. After the above screening process, proteogenomic data of 312 eligible tumor cell lines were finally obtained, covering 8 cancer types (BLCA, BRCA, COAD, ESCC, GBM, LUAD, MEL, and PAAD). Detailed information of the screened cell lines was provided in Supplementary Tables S1 and S2. Drug response analysis in the GDSC2 database For each cancer cohort, Student’s t-test was used to compare the difference in the Area Under the Curve (AUC) of the AZD4547 drug response curve between FGFR-abnormal mutant and wild-type (WT) cell lines (a lower AUC value indicates higher drug sensitivity). P-values were corrected for multiple testing using the Benjamini-Hochberg method. Screening criteria: genes with adjusted P < 0.05 and Fold Change (FC) ≥ 1.2 were defined as FGFRi resistance-related mutant genes; genes with adjusted P < 0.05 and FC ≤ 0.8 were defined as FGFRi sensitivity-related mutant genes. For the correlation between CNV, mRNA, and protein levels and drug response, Spearman rank correlation analysis was used to detect their correlation with the AZD4547 AUC value, and P -values were corrected using the Benjamini-Hochberg method. Judgment criteria: molecules with adjusted P < 0.05 and Spearman correlation coefficient (r) ≥ 0.1 were defined as FGFRi resistance-related molecules; molecules with adjusted P < 0.05 and r ≤ -0.1 were defined as FGFRi sensitivity-related molecules. Drug response analysis in the PRISM database Fisher’s exact test was used to compare the distribution difference of gene variations (mutant/wild-type) between the FGFRi resistant and sensitive groups in each cancer cohort, and P -values were corrected using the Benjamini-Hochberg method; genes with adjusted P < 0.05 were considered drug response-related mutant genes, among which genes with significantly enriched mutants in the resistant group were FGFRi resistance-related mutant genes, and the rest were FGFRi sensitivity-related mutant genes. Unpaired Student’s t-test was used to compare the differences in CNV levels, mRNA expression levels, and protein abundances between the resistant and sensitive groups, and P-values were corrected using the Benjamini-Hochberg method. Screening criteria: molecules with adjusted P < 0.05 and FC ≥ 1.2 were defined as FGFRi resistance-related molecules; molecules with adjusted P < 0.05 and FC ≤ 0.8 were defined as FGFRi sensitivity-related molecules. Cross-database screening of high-confidence molecules To ensure the robustness of our findings, we employed a strict cross-validation strategy using GDSC2 and PRISM as independent discovery cohorts. Statistical analyses, including differential expression profiling and drug response correlation, were performed independently within each dataset. Overlapping cell lines between the two datasets were retained to assess technical robustness against experimental variations. High-confidence FGFRi resistance/sensitivity-related molecules were defined based on the following criteria: 1. Consistency in Significance: The feature (mutation, CNV, mRNA, or protein) must be statistically significant ( P < 0.05) in both datasets. 2. Concordance in Direction: The feature must exhibit the same direction of association (e.g., consistently upregulated or positively correlated with drug sensitivity) in both cohorts. Only molecular features satisfying these intersection criteria were retained for subsequent functional enrichment and mechanistic analyses. GO and Reactome pathway enrichment analysis Gene Ontology (GO) and Reactome pathway enrichment analyses were performed on high-confidence FGFRi resistance/sensitivity-related mRNAs and proteins using the R package “BioEnricher”. An adjusted P value < 0.05 was used as the statistical significance criterion to screen differentially enriched biological pathways. Gene Set Enrichment Analysis (GSEA) GSEA was performed using the R package clusterProfiler, with the background gene set selected from 50 hallmark gene sets in the MSIGDB database. A gene ranking list was constructed for each cancer cohort: in the GDSC2 database, genome-wide mRNAs and proteins were included and ranked by their Spearman correlation coefficient with the AZD4547 AUC value; in the PRISM database, genome-wide mRNAs and proteins were included and ranked by the FC value between the resistant and sensitive groups. The analysis parameters were set to 1000 permutations, with a False Discovery Rate (FDR) 1 were defined as FGFRi resistance-related candidate pathways, and pathways with NES < -1 were defined as FGFRi sensitivity-related candidate pathways. Criteria for defining significant pathways: each pathway was required to obtain four types of enrichment results (GDSC2-mRNA, GDSC2-protein, PRISM-mRNA, PRISM-protein), among which at least 2 types met the significant enrichment criteria, and the functional directions (resistance/sensitivity) of all significant results were consistent to be identified as a significantly enriched pathway related to FGFRi response in that cancer type. The intersection of molecules in significantly enriched pathways with high-confidence FGFRi resistance/sensitivity-related mRNAs and proteins was taken to determine the core regulatory molecules of the FGFRi response phenotype. Regulatory effect of gene mutations on mRNA/Protein expression Among high-confidence FGFRi resistance/sensitivity-related mutant genes, the effect of their mutation status on the expression of core regulatory mRNAs and proteins was explored. By integrating SNV, mRNA, and protein data of cell lines from the GDSC2 and PRISM databases, Student’s t-test was used to compare the differences in the expression levels of core regulatory mRNAs and protein abundances between mutant and wild-type cell lines, and P -values were corrected using the Benjamini-Hochberg method; mutation-mRNA/protein combinations with adjusted P < 0.05 were defined as significant regulatory relationships. Correlation analysis of CNV, methylation, transcription factors (TFs) with mRNA/Protein A list of TFs was extracted from high-confidence FGFRi sensitivity/resistance-related mRNAs. CNV, methylation, and transcription factor expression data from the GDSC2 and PRISM databases were integrated, and Spearman rank correlation analysis was used to detect the correlation between each CNV, methylation site, TFs expression level and the expression of core regulatory mRNAs/proteins, respectively. After P-value correction by the Benjamini-Hochberg method, significant CNV-mRNA/protein, methylation-mRNA/protein, and transcription factor-mRNA/protein regulatory associations were determined with adjusted P < 0.05 as the criterion. Download of bulk multi-omics data from public cancer patients Multi-omics datasets and matched clinicopathological metadata from The Cancer Genome Atlas (TCGA) were downloaded from the official TCGA data portal ( https://gdc.cancer.gov ). In contrast, data corresponding to the Clinical Proteomic Tumor Analysis Consortium (CPTAC) cohort were derived from a prior publication by Li et al[ 21 ]. Download and processing of single-cell expression data of pan-cancer cell Lines Single-cell expression profiles and clinical annotation data of cancer cell lines were downloaded from the GEO database (GSE157220), which contains single-cell expression data of 205 cancer cell lines (280 cells per cell line). Single-cell expression data were merged with drug sensitivity data from the PRISM database, and quality control and preprocessing were performed using the Seurat software package[ 22 ]. The filtering criteria were as follows: cells with the number of detected genes (nFeature_RNA) > 200 were retained; cells with nFeature_RNA > 5000 (high-complexity abnormal cells) were removed; cells with mitochondrial gene ratio 400 were retained; genes expressed in at least 5 cells were retained. After the above filtering, 9997 cells were finally retained for subsequent analysis. The SCTransform function of the Seurat package was used to normalize the single-cell expression data, principal component analysis (PCA) was performed using the RunPCA function, and the top 20 principal components were selected for subsequent cell clustering and dimensionality reduction analysis. Non-linear dimensionality reduction and visualization were performed using the RunTSNE function. Pan-cancer gene metaprograms analysis Gene metaprograms identification was performed using the R package geneNMF. First, the Seurat object was split by sample ID, and multi-k value (k = 4 ~ 9) non-negative matrix factorization (NMF) analysis was performed on each sample using the multiNMF function, and 800 highly variable genes were selected to identify gene programs of different dimensions; the similarity between gene programs was evaluated based on cosine similarity/Jaccard index, and highly similar gene programs were aggregated into robust metaprograms (MPs) using the getMetaPrograms function, with a weight explanation threshold (weight.explained = 0.6) and a maximum number of genes per metaprogram (max.genes = 300) set to ensure the biological validity of metaprograms; for the number of metaprograms with k = 4 ~ 9, quantitative indicators such as sample coverage, silhouette coefficient, average similarity, number of core genes, and number of original gene program integrations were calculated, and combined with heatmap visualization of the similarity clustering characteristics between metaprograms to determine the optimal number of metaprograms; GO functional enrichment analysis was performed on the core gene sets of each metaprogram, and significantly enriched biological pathways were screened with adjusted P value < 0.05 as the threshold; metaprogram scores of individual cells were calculated using two algorithms, AddModuleScore and AUCell, respectively. Construction of FGFRi Sensitivity Prediction Model Based on Machine Learning First, we used the stratified sampling method of the createDataPartition function in the R package caret to divide the PRISM dataset into a training set (70%), a validation set (20%), and a test set (10%) at a ratio of 7:2:1, ensuring that the ratio of sensitive/resistant samples in each subset was consistent. Lasso (Least Absolute Shrinkage and Selection Operator) regression (the penalty parameter α set to 1) in the R package glmnet was used for feature selection in the training set, and the optimal λ value (lambda.min) that minimized the prediction error was determined through 10-fold cross-validation, with genes with non-zero coefficients in the model retained as candidate features (lambda.min = 0.05366335). Based on the candidate features, 7 mainstream machine learning algorithms were used for FGFRi sensitive/resistant binary classification prediction using the mlr3verse ecosystem (mlr3, mlr3fselect, mlr3tuning packages) of R language, including: Random Forest (RF; classif.ranger), Logistic Regression (LogReg; classif.log_reg), Decision Tree (DT; classif.rpart), Extreme Gradient Boosting (XGBoost; classif.xgboost), k-Nearest Neighbor (kNN; classif.kknn), Naive Bayes (Naive Bayes; classif.naive_bayes), and Multilayer Perceptron (MLP; classif.nnet). Hyperparameter tuning was performed for each algorithm using the tnr ("random_search") random search method with 10 ~ 500 iterations (term_evals), and rsmp ("holdout") leave-one-out resampling and classification accuracy (classif.acc) were used as optimization indicators. The final model was obtained by retraining on the complete training set based on this combination. The model performance was quantified using multi-dimensional indicators on the validation set and test set: the main evaluation indicator was the Area Under the Receiver Operating Characteristic Curve (AUC; classif.auc), and the secondary evaluation indicators included accuracy (ACC), precision, recall, sensitivity, specificity, Negative Predictive Value (NPV), and Positive Predictive Value (PPV). Meanwhile, the true positive/false positive/true negative/false negative rates were analyzed through the confusion matrix. The core hyperparameters of the optimal random forest model determined via hyperparameter optimization are set as follows: number of decision trees (num.trees) = 15, number of randomly selected features per tree (mtry) = 3, minimum number of samples per node (min.node.size) = 3, and maximum tree depth (max.depth) = 6. For the model validation, overlapping cell lines from GDSC2 dataset were excluded to ensure independence. To elucidate the prediction mechanism of the optimally performing RF model and to elucidate the contributions of key features to drug efficacy predictions, we employed Shapley Additive Explanations (SHAP) for an interpretable analysis of the model. Grounded in the Shapley Value principle from game theory, SHAP breaks down the model's predictions into the cumulative contribution of each input feature, allowing for precise interpretation of individual predictions and a quantitative assessment of global feature importance. We computed the SHAP values for all input features using the trained optimal random forest model. We utilized the mean absolute SHAP value of each feature as a quantitative measure of feature importance, ranked the features accordingly, and identified the core elements that significantly influence drug efficacy predictions. Screening of Synthetic Lethal Vulnerabilities To identify robust synthetic lethal (SL) partners for the FGF/FGFR family, we developed a systematic computational framework integrating pharmacogenomic profiling, functional genomic screens, and biological relevance filtering. The screening pipeline was conducted across 8 cancer types based on the following steps: 1. Genotype Stratification For each query gene (designated as Gene A), cell lines within a specific cancer type were stratified into "Gene A-inactive" and "Gene A-active" cohorts. A cell line was classified as "Gene A-inactive" if it met at least one of the following characteristics: Deleterious Mutation: Harboring a somatic mutation classified as damaging, including frameshift variants, splice acceptor/donor variants, start lost, stop gained, and stop lost. Missense variants were included only if predicted as "deleterious" by both SIFT and PolyPhen-2 algorithms. Copy Number Loss: Gene copy number falling within the bottom 25th percentile of all cell lines in the corresponding cancer type. Low Expression: mRNA expression levels (TPM) falling within the bottom 25th percentile of all cell lines in the corresponding cancer type. Cell lines failing to meet any of these criteria were classified as "Gene A-active". 2. Pharmacogenomic Interaction Profiling We utilized drug sensitivity data from the PRISM database to identify compounds exhibiting selective lethality in the Gene A-inactive cells. For each compound, we compared the log-fold change (LFC) values between Gene A-inactive and Gene A-active groups using an unpaired Student’s t-test. Drugs showing significantly lower viability in the inactive group were considered significant hits (Benjamini-Hochberg adjusted P < 0.05). The annotated molecular targets of these compounds were extracted as potential synthetic lethal partners (Candidate Gene B). 3. CRISPR-Cas9 Functional Validation To corroborate pharmacological findings with genetic evidence, we analyzed genome-wide CRISPR-Cas9 gene dependency scores from the DepMap portal. We compared the dependency scores of the Candidate Gene B list between Gene A-inactive and Gene A-active groups. Genes exhibiting a significantly lower dependency score (indicating higher essentiality) in the Gene A-inactive group were prioritized. 4. Biological Functional Similarity Filtering To ensure biological plausibility and reduce false positives, we evaluated the functional relatedness between the query gene (Gene A) and candidate partners (Gene B). Functional similarity scores were calculated based on Gene Ontology (GO) terms using the R package GOSemSim. Only gene pairs with score > 0.3 were regarded as potential SL pairs. Finally, to define a high-confidence set of SL pairs, we applied an intersection approach: only genes that demonstrated consistent lethality in both pharmacogenomic and CRISPR screens, and shared significant functional similarity (score > 0.3) with Gene A, were retained as final SL targets. Statistical analysis Statistical analysis was performed with R software (version 4.2.6). Spearman’ s correlation coefficients were utilized to assess the associations between two continuous variables. For categorical data, fisher’s exact test was applied, and for continuous variables, the wilcoxon rank-sum test or the t-test was used for comparison. The P -value correction was performed using the Benjamini-Hochberg method, with statistical significance defined as an adjusted P -value < 0.05. Ethical consideration This study utilized publicly available, de-identified datasets from the GEO (GSE157220), DepMap, GDSC2, PRISM, TCGA, and CPTAC repositories. All original data collection and sharing protocols were conducted in accordance with the ethical standards and informed consent requirements of the respective source institutions. Since this research involved only the secondary analysis of anonymized public data, further institutional review board (IRB) approval was not required. Results Study cohort and proteogenomic profiling To systematically dissect the molecular mechanisms underlying FGFRi resistance, we constructed a comprehensive multi-omics framework integrating pharmacogenomic data from the GDSC2 and PRISM databases (Fig. 1 A). Our analysis encompassed somatic mutations, DNA methylation, CNVs, transcriptomics, and proteomics profiles across 312 cell lines spanning the eight cancer lineages (Fig. 1 B, C). To bridge the gap between in vitro findings and clinical reality, we leveraged patient cohorts from TCGA and CPTAC to validate the prognostic significance of candidate biomarkers identified from these bulk profiles. Beyond bulk analysis, we incorporated scRNA-seq data to capture resistance heterogeneity at cellular resolution. Furthermore, to translate these multi-omics insights into therapeutic strategies, we developed a machine learning-based predictive model and performed systematic synthetic lethality screening. The overall study design and the multi-dimensional data availability are illustrated in Fig. 1 A-C. The Genomic Landscape of FGFRi Resistance Reveals Context-Specific Alterations To characterize the genomic features associated with FGFRi sensitivity, we systematically profiled somatic mutations, DNA methylation patterns, and CNVs across diverse cancer lineages. Profiling of the GDSC2 MEL cohort revealed that cell lines harboring mutations in ARHGAP33 , PDZD2 , and UGT3A2 exhibited significantly reduced sensitivity (higher AUC values) to AZD4547 (Fig. 2 A). Subsequent validation in the PRISM dataset corroborated this association, demonstrating a significant enrichment of these mutant genotypes specifically within the resistant subgroup (Fig. 2 B). Given the known roles of ARHGAP33 in Rho signaling and UGT3A2 in drug metabolism, these alterations may suggest potential mechanisms involving cytoskeletal dynamics and metabolic clearance, warranting further validation. Across other cancer lineages, resistant phenotypes were characterized by distinct genomic signatures, such as TENM4 mutations in LUAD and APC2 mutations in ESCC, which were similarly associated with diminished drug response (Fig. 2 A, B). Notably, as APC2 is a component of the Wnt destruction complex, its mutational enrichment implicates aberrant Wnt signaling as a potential feature of the resistant phenotype. Beyond point mutations, epigenetic and structural analyses revealed multi-layered heterogeneity. In BRCA, resistance was associated with higher methylation levels of homeobox genes like EN1 and HOXC6 (Fig. 2 C, D), suggesting the epigenetic silencing of these developmental regulators. Conversely, the resistant landscape of GBM was marked by recurrent copy number amplifications. We identified 403 resistance-correlated CNVs in GBM (Table S3 ), including the amplification of ELF4 and ARHGEF6 (Fig. 2 E). Previous studies have linked ELF4 to stemness maintenance in GBM, and ARHGEF6 to tumor progression in various cancers[ 23 – 26 ], suggesting that the amplification-driven upregulation of these oncogenic drivers may support the aggressive resistant state. Transcriptomic and proteomic signatures of FGFRi response To identify high-confidence molecular signatures associated with FGFRi response, we performed an integrated analysis of pharmacogenomic data from the GDSC2 and PRISM cohorts. We specifically screened for candidate genes that demonstrated consistent drug response associations between the two datasets within each specific cancer lineage, thereby yielding a robust panel of biomarkers (Fig. 3 A, Figure S1 A, B; Table S4). To further assess their clinical relevance, we mapped these candidate genes to patient data from the TCGA and CPTAC cohorts (Fig. 3 B). This comprehensive landscape revealed that many identified markers exhibited tumor-specific differential expression and significant prognostic value. Specifically, a subset of resistance-associated transcripts—highlighted by UCP2 , COL11A1 , PRAME , and UNC5A —demonstrated a striking multi-dimensional consistency. These genes were not only correlated with AZD4547 resistance but were also identified as prognostic risk factors in patient cohorts. Notably, markers such as COL11A1 and PRAME were also significantly upregulated in tumor tissues compared to normal controls. Conversely, we identified a panel of sensitivity-associated markers, such as ATOH8 , ATP8B1 , LIMCH1 , and NTN4 . These genes formed a distinct cluster characterized by negative drug response correlations and favorable patient prognosis. Complementing the transcriptomic landscape, we further leveraged proteomic profiling to capture functional alterations. Notably, the identified proteomic signatures showed minimal overlap with the transcriptomic markers (Figs. 3 C, D, Figure S1 C), indicating the prevalence of post-transcriptional regulation in driving resistance. This analysis highlighted distinct protein-level markers, such as TM9SF4 and ZMPSTE24, which were consistently upregulated in resistant phenotypes across both datasets (Figs. 3 C, D; Table S5). Clinical validation in the CPTAC dataset confirmed these candidates were significantly overexpressed in tumor tissues (Fig. 3 E). Crucially, our proteomic screen also highlighted ADAM17 as a robust resistance marker. Given its established role in shedding EGFR ligands, the upregulation of ADAM17 provides proteomic evidence corroborating the "RTK switching" mechanism toward EGFR signaling, which we further validate in downstream analyses. Lineage-specific and convergent functional landscapes of FGFRi resistance To transition from discrete molecular markers to systems-level resistance mechanisms, we performed GO and Reactome functional enrichment analyses on the mRNA and protein signatures identified from our integrative screen. This initial profiling revealed that the molecular landscape of resistance is characterized by lineage-dependent adaptive programs rather than a uniform pan-cancer mechanism (Figure S1 D-G). For instance, transcriptomic signatures in COAD were predominantly associated with metabolic pathways, such as 'Respiratory electron transport' (Figure S1 E), whereas ESCC and BLCA exhibited proteomic enrichment in structural modules like 'Assembly of collagen fibrils' (Figure S1 G). To ensure the reliability of these lineage-specific findings and mitigate potential platform- or omics-level discordance, we performed GSEA across four independent discovery layers: GDSC2-mRNA/protein and PRISM-mRNA/protein. To prioritize high-confidence biological modules, we implemented a stringent consensus criterion, requiring significant enrichment in at least two data layers with consistent functional polarity. This integrative approach identified six convergent functional modules associated with FGFRi resistance: metabolic reprogramming, cell-cycle dysregulation, estrogen response, immunity & inflammation, signaling transduction, and invasion & migration (Figs. 4 A; Table S6). Further analysis revealed that these resistance programs are highly context-dependent, manifesting as distinct molecular signatures across specific cancer lineages. In COAD, the resistant phenotype displayed a multidimensional functional profile encompassing metabolic reprogramming (e.g., adipogenesis), signaling transduction (e.g., mTORC1 signaling), and immunity & inflammation (e.g., interferon-gamma response) (Fig. 4 A, Figure S2 A). Within these pathways, ATP1B3 and UCP2 were prioritized as high-confidence regulatory components representing the mTORC1 signaling and adipogenesis programs, respectively. These markers exhibited concordant associations with resistance, characterized by positive drug-response correlations in GDSC2 and significantly elevated expression in the PRISM resistant cohort (Fig. 4 B). Furthermore, UCP2 was validated as a significant prognostic risk factor in both CPTAC and TCGA datasets (Fig. 4 C), aligning with its documented role in modulating mitochondrial bioenergetics and reinforcing the clinical relevance of mitochondrial metabolic adaptation[ 27 , 28 ]. Collectively, these findings suggest that resistant cells in COAD may exploit metabolic plasticity and signaling rewiring to elevate the threshold for drug-induced cell death. Conversely, the resistance landscape in BLCA was marked by an invasion & migration signature, primarily involving Epithelial-Mesenchymal Transition (EMT) and apical junctions (Fig. 4 A, Figure S2 B). The actin-bundling protein FSCN1 was identified as a core component of this phenotype; its expression was positively correlated with FGFRi resistance and was identified as a significant prognostic risk factor in bladder cancer patients (Fig. 4 D, E). The prominence of FSCN1 suggests that resistant cells may leverage cytoskeletal remodeling to enhance cellular plasticity and invasive potential, thereby facilitating therapeutic evasion. Additionally, the resistant phenotype in BRCA was significantly linked to estrogen response pathways and elevated MLPH expression (Fig. 4 A, Figure S2 C, D). Despite these lineage-specific variations, a convergent reliance on cell-cycle progression was identified as a shared hallmark of resistance in PAAD and GBM. These resistant phenotypes were significantly enriched for MYC targets, E2F targets, and G2M checkpoint programs (Fig. 4 A, Figures S2 E, F). In PAAD, MCM7 expression was positively correlated with FGFRi resistance and identified as a significant prognostic risk factor (Fig. 4 F, G). Similarly, the resistant state in GBM was characterized by the upregulation of genes involved in cell-cycle progression and nucleotide metabolism, most notably MCM4 and HPRT1 (Fig. 4 H, Figures S2 G, H). Both markers exhibited striking associations with unfavorable outcomes in CPTAC and TCGA-GBM cohorts (Figs. 4 I, J). Given the fundamental role of the Minichromosome Maintenance (MCM) family in DNA replication initiation[ 29 ], these findings point toward a link between FGFRi resistance and a hyper-proliferative state in PAAD and GBM, likely supported by the coordinated activation of cell-cycle and metabolic programs. Distinct multi-omics regulatory landscapes of FGFRi resistance To elucidate how genomic aberrations propagate across molecular hierarchies to shape the FGFRi resistance phenotype, we systematically analyzed the correlations between high-confidence somatic alterations (mutations, CNVs) and epigenetic features (DNA methylation) with downstream transcriptome and proteome profiles. This analysis revealed highly heterogeneous regulatory patterns, where distinct modes of signaling rewiring explain the lineage-specific phenotypes observed earlier. In GBM, we uncovered an extensive regulatory network integrating both CNV effects and transcription factor (TF) associations. The analysis detected 410 CNV-mRNA and 1,677 CNV-protein regulatory pairs (Table S7), underscoring the pervasiveness of CNV impact in GBM. Beyond this genomic baseline, the network was further orchestrated by specialized transcriptional programs. Notably, the TFs CTCF, HDAC1, and RBMX emerged as central regulatory hubs. Among them, RBMX exhibited strong positive correlations with both prognostic risk factors MCM4 and HPRT1, whereas CTCF and HDAC1 were primarily linked to the regulation of MCM4 (Fig. 5 A). This hierarchical flow suggests that CNV-TF coordination may sustain the hyper-proliferative state and nucleotide metabolic reprogramming associated with the FGFRi-resistant phenotype in GBM. In contrast, other cancer cohorts exhibited more focused regulatory profiles. In BLCA, we observed a discrete multi-dimensional landscape involving 14 CNV-mRNA, 9 CNV-protein, and 8 TF-mRNA associations. A notable genomic-transcriptomic link was the significant negative correlation between CHEK2P2 copy number and the expression of the resistance marker NT5E (Fig. 5 B). Furthermore, we identified the transcription factor EHMT2 as a potential regulator of the actin-bundling protein FSCN1 (Fig. 5 B). This EHMT2-FSCN1 axis may provide a mechanistic basis for the EMT and pro-invasive phenotype previously linked to FGFRi resistance in bladder cancer patients. Meanwhile, BRCA resistance was uniquely characterized by epigenetic modulation. Specifically, the expression of the resistance-associated marker MLPH exhibited significant positive correlations with the methylation levels of HAAO and EN1 (Fig. 5 C). In PAAD, a convergent regulatory mode was identified, featuring 10 TF-mRNA associations. Notably, the transcriptional co-activator TRRAP emerged as a pivotal regulator, significantly orchestrating the mRNA expression of ABL1 and MCM7 (Fig. 5 D). Finally, these core regulatory molecules were synthesized into a unified multi-omics interaction network (Fig. 5 E), providing a holistic view of the coordinated rewiring of genomic, epigenetic, and transcriptional landscapes in FGFRi resistance. Single-cell Profiling Resolves Clonal Selection and Conserved Metaprograms Underlying FGFRi Resistance To dissect the cellular-level heterogeneity and evolutionary dynamics governing FGFRi resistance, we transitioned from bulk-level profiling to scRNA-seq analysis across a pan-cancer atlas. Following a stringent quality control pipeline, a high-resolution landscape of 9,997 individual cells was established, encompassing diverse lineages and distinct drug-response phenotypes (Figs. 6 A-C). To identify conserved biological states across these heterogeneous cell lines, we utilized NMF to deconvolve the transcriptomic data into robust gene metaprograms (MPs). This analysis derived four recurrent MPs representing distinct oncogenic processes: tumor immunity (MP1), cell cycle (MP2), DNA damage repair (MP3), and telomerase activity (MP4) (Figs. 6 D, E). Notably, the activity landscapes of these MPs exhibited lineage-dependent characteristics. Specifically, MP2 was activated in MEL and GBM cells, while MP3 showed prominent activation in MEL and BRCA lineages (Fig. 6 F), defining their baseline transcriptomic features. Crucially, the statistical enrichment of these MPs in resistant phenotypes was highly context-dependent. MP2 was consistently enriched in the resistant subcohorts of PAAD and MEL across two independent scoring algorithms, whereas MP3 activity was over-represented in resistant populations of MEL and GBM (Figs. 6 G, H). The enrichment of MP2—characterized by core replication drivers such as CDK1 and BUB1—within resistant PAAD and MEL subcohorts suggests that FGFRi resistance is predominantly underpinned by the clonal selection of pre-existing subpopulations harboring intrinsic cell-cycle dysregulation. Furthermore, the upregulation of MP3 in resistant MEL and GBM indicates that enhanced genomic maintenance serves as a critical survival adaptation under therapeutic stress. Together, these findings demonstrate that FGFRi resistance arises from the clonal selection of pre-existing cell states rather than uniform transcriptomic adaptation, and diverse lineages employ distinct strategies—from hyper-proliferation to fortified DNA repair—to collectively elevate the cellular survival threshold against FGFR inhibition. Development of a Biologically Grounded Machine Learning model for Predicting FGFRi Response To translate our discovery into clinical utility, we developed a robust transcriptome-based predictive model to stratify FGFRi sensitivity. The PRISM discovery cohort was randomly partitioned into a training set (70%), a validation set (20%), and an independent test set (10%) to ensure a rigorous and unbiased evaluation of model performance (Fig. 7 A). Utilizing LASSO regression, we distilled the high-dimensional transcriptional landscape into a parsimonious 52-gene signature (Figs. 7 B, C). We then benchmarked seven mainstream machine learning paradigms to select the optimal algorithm. The Random Forest ensemble model was identified as the superior classifier, achieving top-performing discriminative power with the highest AUC in both the internal validation set (AUC = 0.782) and the independent test set (AUC = 0.714) (Fig. 7 D, E). Beyond its remarkable AUC, the model consistently excelled across a comprehensive suite of evaluation metrics—including sensitivity, precision, recall, and accuracy—within both the validation and test cohorts (Figs. 7 F, G). The generalizability of this framework was further corroborated via external validation in the GDSC2 database, where cell lines predicted as resistant by the model exhibited significantly higher drug response (AUC) values than those predicted as sensitive ( P = 0.021) (Fig. 7 H). This multi-dimensional superiority underscores the predictive robustness and translational potential of the 52-gene signature in identifying FGFRi-responsive phenotypes across diverse malignancies. To deconstruct the algorithmic interpretability of the Random Forest model and ensure its biological fidelity, we employed SHAP to quantify the contribution of individual features to the predictive output (Figs. 7 I, J). The prioritization of top-ranking features, including USP18 , HIP1R , and RASGRF2 , underscores the model’s alignment with established oncogenic signaling dependencies. Specifically, USP18 has been documented to fuel tumor growth and proliferation by activating the AKT signaling pathway[ 30 , 31 ], while while HIP1R modulates cellular survival and invasive potential primarily via the PI3K/AKT axis[ 32 , 33 ], and RASGRF2 orchestrates both Src/PI3K and NF-κB cascades to promote metastatic phenotypes[ 34 ]. Notably, the convergence of these top-ranked molecules on the PI3K/AKT axis—a pathway we subsequently characterize as a core synthetic lethal vulnerability (see Fig. 8 )—suggests that our framework accurately captures the fundamental signaling rewiring that facilitates therapeutic evasion. Furthermore, the inclusion of IFNA13 , a mediator of interferon signaling, and GALT, a key enzyme involved in galactose metabolism, within the top feature set reflects the model's capacity to integrate inflammatory response and metabolic reprogramming. These features represent two convergent adaptive strategies that we previously identified through GSEA and proteogenomic profiling as hallmarks of the FGFRi-resistant phenotype. This mechanistic consistency emphasizes that the 52-gene signature is not merely a collection of statistical correlates but a biologically grounded representation of the systemic network rewiring underlying pan-cancer FGFRi resistance. Identification of Synthetic Lethal Vulnerabilities Linked to the FGF/FGFR Axis To uncover potential therapeutic targets that synergize with FGFR inhibition, we implemented a systematic computational framework to screen for synthetic lethal (SL) partners of the FGF/FGFR axis (comprising 4 FGFRs and 22 FGF ligands) across eight cancer types. By aggregating the significant pairs identified independently across the cancer types, we constructed a comprehensive pan-cancer SL landscape for the FGF/FGFR family (Table S8). For instance, in ESCC, we identified 25 SL pairs, while 39 pairs were identified in PAAD. To elucidate the biological implications of these SL interactions, we performed an integrated pathway enrichment analysis combining both GO and KEGG databases. This multi-dimensional annotation revealed a convergent functional landscape focused on five core biological modules: signaling transduction, metabolic reprogramming, cell cycle and proliferation, invasion and migration, and immunity (Figs. 8 A, B). First, Signal Transduction Reactivation. Both databases consistently highlighted a heavy reliance on canonical survival cascades, specifically "Positive regulation of MAPK/ERK cascade" (GO) and "PI3K-Akt signaling pathway" (KEGG). These findings resonate with our earlier multi-omics GSEA results (Fig. 4 ), where the KRAS and mTORC1 signaling pathways were significantly enriched in resistant phenotypes. This convergence suggests that when the FGFR axis is compromised, tumor cells strategically rewire their signaling networks to exploit these parallel hubs. Second, Cell Cycle Dysregulation. Terms related to "Positive regulation of cell cycle" and "G1/S transition" were prominent, which aligns seamlessly with our single-cell analysis that pinpointed the aberrant activation of the cell-cycle metaprogram (MP2) in resistant subpopulations (Fig. 6 ). This indicates that targeting SL partners involved in cell cycle checkpoints (e.g., CDKs) could specifically eliminate the hyper-proliferative resistant clones. Third, RTK Switching. Most notably, we observed a significant enrichment of the 'EGFR tyrosine kinase inhibitor resistance' pathway (KEGG). This finding provides robust computational validation for the "RTK switching" hypothesis introduced earlier, confirming that FGFR inhibition triggers a compensatory upregulation of alternative RTKs, such as EGFR, to sustain downstream signaling. Consequently, dual blockade of FGFR and these identified SL targets represents a rational synthetic lethal strategy to overcome resistance. Consequently, dual blockade of FGFR and these identified SL targets (e.g., EGFR or MEK/ERK pathway components) could represent a promising synthetic lethal strategy to overcome resistance. Discussion Despite the initial clinical efficacy of FGFRi such as AZD4547, the inevitable emergence of acquired resistance remains a major bottleneck in precision oncology[ 2 ]. Current research has largely been confined to isolated resistance mechanisms within specific cancer types, failing to capture the systemic network rewiring that drives therapeutic failure. In this study, we bridged this gap by constructing a comprehensive pan-cancer proteogenomic atlas of FGFRi resistance. By integrating data from 312 cell lines across the GDSC2 and PRISM databases, we deciphered the multi-layered heterogeneity of resistance and identified actionable synthetic lethal vulnerabilities that pave the way for next-generation combination therapies. A key finding of our study is the context-dependent reliance of resistant cells on metabolic reprogramming. In colorectal cancer (COAD), we observed a striking activation of the adipogenesis pathway and upregulation of UCP2 . UCP2 is known to reduce mitochondrial oxidative stress and inhibit apoptosis, thereby raising the threshold for drug-induced cell death[ 27 , 35 ]. Consistent with this, our synthetic lethality screening further highlighted "Choline metabolism" and "Lipid and atherosclerosis" pathways as critical vulnerabilities. This suggests that under FGFRi blockade, tumor cells may undergo a metabolic switch—shifting from glycolysis to fatty acid oxidation or altering membrane lipid composition—to sustain survival signals. This finding aligns with the "warburg effect" plasticity and suggests that combining FGFRi with metabolic inhibitors (e.g., inhibitors of fatty acid synthesis) could be a viable strategy for specific tumor subtypes[ 36 , 37 ]. Since UCP2-mediated metabolic adaptation is often tumor-specific and less prominent in normal tissues [ 38 , 39 ], targeting this vulnerability offers a potential therapeutic window with reduced toxicity. Beyond bulk omics, our single-cell analysis provided high-resolution insights into the evolutionary dynamics of resistance. We identified a specific cell-cycle metaprogram (MP2) that is aberrantly activated in resistant subpopulations across melanoma and glioblastoma. This observation is consistent with a model of clonal selection rather than purely acquired plasticity: FGFR inhibition likely eliminates sensitive clones while sparing pre-existing subpopulations with intrinsic cell-cycle dysregulation. The enrichment of MCM family genes (e.g., MCM7 , MCM4 ) in resistant phenotypes further corroborates this. However, this interpretation requires validation through longitudinal tracking or functional experiments. Perhaps the most translationally significant discovery of our study is the support of the "RTK switching" hypothesis. Our GSEA analysis initially revealed that resistant cells maintain high activity of downstream KRAS and mTORC1 signaling despite FGFR blockade, hinting at the activation of alternative upstream drivers. This hypothesis was supported by our synthetic lethality screening, which identified the "EGFR tyrosine kinase inhibitor resistance" pathway as a top-ranked vulnerability. This indicates that upon FGFR inhibition, tumor cells dynamically rewire their signaling networks to exploit EGFR as a compensatory bypass track to reactivate the MAPK/PI3K cascade. Furthermore, our functional enrichment revealed potential engagement of autophagy and PD-L1 checkpoint pathways, suggesting that this network rewiring extends beyond kinases to include survival-mediated autophagy and immune evasion. These results provide a compelling rationale for dual-blockade strategies (e.g., FGFRi + EGFRi or FGFRi + anti-PD-1) to dismantle these compensatory survival loops. Translating these molecular findings into clinical utility, we developed a machine learning model based on 52 transcriptomic features. Unlike static biomarkers (e.g., genomic mutations), our model integrates dynamic mRNA signatures to accurately predict the probability of FGFRi response. This tool holds significant potential for clinical utility, serving as a robust stratification tool to distinguish patients likely to benefit from AZD4547 monotherapy from those requiring immediate upfront combination strategies. We propose a tiered precision medicine framework: Specifically, patients predicted to be 'sensitive' could proceed with monotherapy, thereby sparing them the added toxicity of combination regimens. Conversely, those predicted to be 'resistant' would be flagged as high-priority candidates for the combination therapies we identified (e.g., FGFRi plus EGFR or metabolic inhibitors). For this resistant subgroup, subsequent profiling of specific markers (e.g., UCP2 levels or EGFR activation) would guide the precise selection of the secondary agent. Mechanistically, the predictive power of our model is biologically grounded. The top features such as USP18 and HIP1R are established regulators of the PI3K/AKT axis[ 31 , 33 ]. This aligns seamlessly with our synthetic lethality findings, where the PI3K/AKT pathway emerged as a convergent survival bypass downstream of EGFR, further validating the biological interpretability of our computational approach. Our study has limitations inherent to cell line-based analyses. First, in vitro models lack the complex tumor microenvironment (TME), which may influence the translatability of immune-related findings, such as PD-L1 associations. Second, the technical limitations of scRNA-seq in capturing transient metabolic states. The cell dissociation may lead to the loss of certain mRNA transcripts or alter the expression of genes sensitive to cellular stress, which could potentially impact the detection of subtle metabolic reprogramming signatures. To address this, future studies incorporating spatial metabolomics or integrated functional experiments are warranted to validate the metabolic reprogramming patterns identified in this study. Third, the unequal sample sizes across cancer types may introduce bias. Fortunately, our cross-platform validation across bulk and clinical cohorts ensures the robustness of the core findings. Forth, this study focused on a single FGFR inhibitor (AZD4547), and the generalizability to other FGFR inhibitors requires validation. Finally, computational synthetic lethal candidates require experimental validation in patient-derived organoids (PDOs) or xenograft (PDX) models to confirm their clinical efficacy. Conclusions In summary, this study presents a first-of-its-kind pan-cancer atlas of FGFRi resistance. By integrating multi-omics profiling with synthetic lethality screening, we unveiled a paradigm where resistance is driven by a dynamic interplay of RTK switching (EGFR), metabolic rewiring (UCP2), and clonal selection of cell-cycle variants. These insights not only clarify the heterogeneous nature of resistance but also offer a concrete, actionable framework for the development of precise, mechanism-based combination therapies to improve patient outcomes. Abbreviations ACC Accuracy AUC Area Under the Curve BLCA Bladder Urothelial Carcinoma BRCA Breast Invasive Carcinoma CNV Copy Number Variation COAD Colon Adenocarcinoma DT Decision Tree EMT Epithelial-Mesenchymal Transition ESCC Esophageal Squamous Cell Carcinoma FC Fold Change FDR False Discovery Rate FGF Fibroblast Growth Factor FGFR Fibroblast Growth Factor Receptor FGFRi Fibroblast Growth Factor Receptor Inhibitor GBM Glioblastoma kNN k-Nearest Neighbor LASSO Least Absolute Shrinkage and Selection Operator LFC Log-Fold Change LogReg Logistic Regression LUAD Lung Adenocarcinoma MEL Melanoma MLP Multilayer Perceptron MP Metaprogram NES Normalized Enrichment Score NMF Non-negative Matrix Factorization NPV Negative Predictive Value PAAD Pancreatic Adenocarcinoma PCA Principal Component Analysis PDO Patient-Derived Organoid PDX Patient-Derived Xenograft PPV Positive Predictive Value RF Random Forest ROC Receiver Operating Characteristic RTK Receptor Tyrosine Kinase SHAP Shapley Additive Explanations SL Synthetic Lethal SNV Single Nucleotide Variant TF Transcription Factor TKI Tyrosine Kinase Inhibitor TME Tumor Microenvironment TPM Transcripts Per Million t-SNE t-Distributed Stochastic Neighbor Embedding WT Wild-Type XGBoost Extreme Gradient Boosting Declarations Ethics approval and consent to participate Not applicable. Consent for publication Not applicable. Competing interests The authors declare that they have no competing interests. Funding Not applicable. Authors' contributions LT conceived the study and performed the bioinformatic analysis. TH and PY assisted with data collection and visualization. XW, HJ and JL supervised the study, provided resources, and revised the manuscript. All authors read and approved the final manuscript. Acknowledgements We thank the authors of the GDSC, PRISM, TCGA, CPTAC, and GEO datasets for making their data publicly available. We also acknowledge the BioEnricher and Seurat development teams for their open-source tools. We would like to thank our friend Jinyang Liu for providing valuable guidance and support during the data analysis process. Availability of data and materials The datasets analyzed in the study are available in the following public repositories: DepMap: https://depmap.org/portal/ GDSC2: https://www.cancerrxgene.org/ PRISM: https://depmap.org/portal/prism/ TCGA: https://gdc.cancer.gov/ CPTAC: The data was derived from the publication by Li et al. [ 21 ], available at https://pubmed.ncbi.nlm.nih.gov/37582357/ or via the CPTAC data portal ( https://pdc.cancer.gov/ ). GEO: Accession number GSE157220 ( https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE157220 ). The code used for the machine learning model and analysis is available from the corresponding author upon reasonable request. References Xie Y, Su N, Yang J, Tan Q, Huang S, Jin M, et al. FGF/FGFR signaling in health and disease. Signal Transduct Target Ther. 2020;5:181. https://doi.org/10.1038/s41392-020-00222-7 . Krook MA, Reeser JW, Ernst G, Barker H, Wilberding M, Li G, et al. Fibroblast growth factor receptors in cancer: genetic alterations, diagnostics, therapeutic targets and mechanisms of resistance. Br J Cancer. 2021;124:880–92. https://doi.org/10.1038/s41416-020-01157-0 . Ternier G, Shahzad K, Edirisinghe O, Okoto P, Alraawi Z, Sonnaila S, et al. Fibroblast Growth Factors: Roles and Emerging Therapeutic Applications. Curr Drug Targets. 2025;26:551–70. https://doi.org/10.2174/0113894501351461250301072444 . Hung KL, Luebeck J, Dehkordi SR, Colón CI, Li R, Wong IT-L, et al. Targeted profiling of human extrachromosomal DNA by CRISPR-CATCH. Nat Genet. 2022;54:1746–54. https://doi.org/10.1038/s41588-022-01190-0 . Paik PK, Shen R, Berger MF, Ferry D, Soria J-C, Mathewson A, et al. A Phase Ib Open-Label Multicenter Study of AZD4547 in Patients with Advanced Squamous Cell Lung Cancers. Clin Cancer Res. 2017;23:5366–73. https://doi.org/10.1158/1078-0432.CCR-17-0645 . Aggarwal C, Redman MW, Lara PN, Borghaei H, Hoffman P, Bradley JD, et al. SWOG S1400D (NCT02965378), a Phase II Study of the Fibroblast Growth Factor Receptor Inhibitor AZD4547 in Previously Treated Patients With Fibroblast Growth Factor Pathway-Activated Stage IV Squamous Cell Lung Cancer (Lung-MAP Substudy). J Thorac Oncol. 2019;14:1847–52. https://doi.org/10.1016/j.jtho.2019.05.041 . Gonzalez-Ericsson PI, Unni N, Jhaveri K, Stringer-Reasor E, Liu Q, Wang Y, et al. Phase Ib Trial of Fulvestrant, Palbociclib, and Erdafitinib, a pan-FGFR Tyrosine Kinase Inhibitor, in HR+/HER2- Metastatic Breast Cancer. Clin Cancer Res. 2025;31:3652–61. https://doi.org/10.1158/1078-0432.CCR-24-3803 . Franza A, Pirovano M, Giannatempo P, Cosmai L. Erdafitinib in locally advanced/metastatic urothelial carcinoma with certain FGFR genetic alterations. Future Oncol. 2022;18:2455–64. https://doi.org/10.2217/fon-2021-1151 . Peng M, Deng J, Li X. Clinical advances and challenges in targeting FGF/FGFR signaling in lung cancer. Mol Cancer. 2024;23:256. https://doi.org/10.1186/s12943-024-02167-9 . Powles T, Carroll D, Chowdhury S, Gravis G, Joly F, Carles J, et al. An adaptive, biomarker-directed platform study of durvalumab in combination with targeted therapies in advanced urothelial cancer. Nat Med. 2021;27:793–801. https://doi.org/10.1038/s41591-021-01317-6 . Saborowski A, Lehmann U, Vogel A. FGFR inhibitors in cholangiocarcinoma: what’s now and what’s next? Ther Adv Med Oncol. 2020;12:1758835920953293. https://doi.org/10.1177/1758835920953293 . Katoh M, Loriot Y, Brandi G, Tavolari S, Wainberg ZA, Katoh M. FGFR-targeted therapeutics: clinical activity, mechanisms of resistance and new directions. Nat Rev Clin Oncol. 2024;21:312–29. https://doi.org/10.1038/s41571-024-00869-z . Goyal L, Saha SK, Liu LY, Siravegna G, Leshchiner I, Ahronian LG, et al. Polyclonal Secondary FGFR2 Mutations Drive Acquired Resistance to FGFR Inhibition in Patients with FGFR2 Fusion-Positive Cholangiocarcinoma. Cancer Discov. 2017;7:252–63. https://doi.org/10.1158/2159-8290.CD-16-1000 . Ryan MR, Sohl CD, Luo B, Anderson KS. The FGFR1 V561M Gatekeeper Mutation Drives AZD4547 Resistance through STAT3 Activation and EMT. Mol Cancer Res. 2019;17:532–43. https://doi.org/10.1158/1541-7786.MCR-18-0429 . Wu Q, Zhen Y, Shi L, Vu P, Greninger P, Adil R, et al. EGFR Inhibition Potentiates FGFR Inhibitor Therapy and Overcomes Resistance in FGFR2 Fusion-Positive Cholangiocarcinoma. Cancer Discov. 2022;12:1378–95. https://doi.org/10.1158/2159-8290.CD-21-1168 . Quintanal-Villalonga A, Molina-Pinelo S, Cirauqui C, Ojeda-Márquez L, Marrugal Á, Suarez R, et al. FGFR1 Cooperates with EGFR in Lung Cancer Oncogenesis, and Their Combined Inhibition Shows Improved Efficacy. J Thorac Oncol. 2019;14:641–55. https://doi.org/10.1016/j.jtho.2018.12.021 . DiPeri TP, Zhao M, Evans KW, Varadarajan K, Moss T, Scott S, et al. Convergent MAPK pathway alterations mediate acquired resistance to FGFR inhibitors in FGFR2 fusion-positive cholangiocarcinoma. J Hepatol. 2024;80:322–34. https://doi.org/10.1016/j.jhep.2023.10.041 . Silverman IM, Hollebecque A, Friboulet L, Owens S, Newton RC, Zhen H, et al. Clinicogenomic Analysis of FGFR2-Rearranged Cholangiocarcinoma Identifies Correlates of Response and Mechanisms of Resistance to Pemigatinib. Cancer Discov. 2021;11:326–39. https://doi.org/10.1158/2159-8290.CD-20-0766 . Jogo T, Nakamura Y, Shitara K, Bando H, Yasui H, Esaki T, et al. Circulating Tumor DNA Analysis Detects FGFR2 Amplification and Concurrent Genomic Alterations Associated with FGFR Inhibitor Efficacy in Advanced Gastric Cancer. Clin Cancer Res. 2021;27:5619–27. https://doi.org/10.1158/1078-0432.CCR-21-1414 . Guercio BJ, Sarfaty M, Teo MY, Ratna N, Duzgol C, Funt SA, et al. Clinical and Genomic Landscape of FGFR3-Altered Urothelial Carcinoma and Treatment Outcomes with Erdafitinib: A Real-World Experience. Clin Cancer Res. 2023;29:4586–95. https://doi.org/10.1158/1078-0432.CCR-23-1283 . Li Y, Porta-Pardo E, Tokheim C, Bailey MH, Yaron TM, Stathias V, et al. Pan-cancer proteogenomics connects oncogenic drivers to functional states. Cell. 2023;186:3921–e394425. https://doi.org/10.1016/j.cell.2023.07.014 . Hao Y, Hao S, Andersen-Nissen E, Mauck WM, Zheng S, Butler A, et al. Integrated analysis of multimodal single-cell data. Cell. 2021;184:3573–e358729. https://doi.org/10.1016/j.cell.2021.04.048 . Kosti A, Chiou J, Guardia GDA, Lei X, Balinda H, Landry T, et al. ELF4 is a critical component of a miRNA-transcription factor network and is a bridge regulator of glioblastoma receptor signaling and lipid dynamics. Neuro Oncol Engl. 2023;25:459–70. https://doi.org/10.1093/neuonc/noac179 . Zhuang Z, Zhang C, Tan Y, Zhang J, Zhong C. ELF4 was a prognostic biomarker and related to immune infiltrates in glioma. J Cancer. 2024;15:5101–17. https://doi.org/10.7150/jca.96886 . Zheng T, Zhou H, Zheng Z, Guo Y, Liu J, Zhang J, et al. The pathological significance and potential mechanism of ARHGEF6 in lung adenocarcinoma. Comput Biol Med. 2023;158:106894. https://doi.org/10.1016/j.compbiomed.2023.106894 . Li K, Wang H, Yang C, Li C, Xue B, Zhou J. Clinical implication and potential function of ARHGEF6 in acute myeloid leukemia: An in vitro study. PLoS ONE. 2023;18:e0283934. https://doi.org/10.1371/journal.pone.0283934 . Yu J, Shi L, Lin W, Lu B, Zhao Y. UCP2 promotes proliferation and chemoresistance through regulating the NF-κB/β-catenin axis and mitochondrial ROS in gallbladder cancer. Biochem Pharmacol. 2020;172:113745. https://doi.org/10.1016/j.bcp.2019.113745 . Wang K, Zhang L, Deng B, Zhao K, Chen C, Wang W. Mitochondrial uncoupling protein 2: a central player in pancreatic disease pathophysiology. Mol Med. 2024;30:259. https://doi.org/10.1186/s10020-024-01027-y . Costa A, Diffley JFX. The Initiation of Eukaryotic DNA Replication. Annu Rev Biochem. 2022;91:107–31. https://doi.org/10.1146/annurev-biochem-072321-110228 . Tan Y, Zhou G, Wang X, Chen W, Gao H. USP18 promotes breast cancer growth by upregulating EGFR and activating the AKT/Skp2 pathway. Int J Oncol. 2018;53:371–83. https://doi.org/10.3892/ijo.2018.4387 . Diao W, Guo Q, Zhu C, Song Y, Feng H, Cao Y, et al. USP18 promotes cell proliferation and suppressed apoptosis in cervical cancer cells via activating AKT signaling pathway. BMC Cancer. 2020;20:741. https://doi.org/10.1186/s12885-020-07241-1 . Zhu S, Xu H, Chen R, Shen Q, Yang D, Peng H, et al. DNA methylation and miR-92a-3p-mediated repression of HIP1R promotes pancreatic cancer progression by activating the PI3K/AKT pathway. J Cell Mol Med. 2023;27:788–802. https://doi.org/10.1111/jcmm.17612 . Zhu J, Wang X, Guan H, Xiao Q, Wu Z, Shi J, et al. HIP1R acts as a tumor suppressor in gastric cancer by promoting cancer cell apoptosis and inhibiting migration and invasion through modulating Akt. J Clin Lab Anal. 2020;34:e23425. https://doi.org/10.1002/jcla.23425 . Lu P, Chen J, Yan L, Yang L, Zhang L, Dai J, et al. RasGRF2 promotes migration and invasion of colorectal cancer cells by modulating expression of MMP9 through Src/Akt/NF-κB pathway. Cancer Biol Ther. 2019;20:435–43. https://doi.org/10.1080/15384047.2018.1529117 . Nesci S, Rubattu S. UCP2, a Member of the Mitochondrial Uncoupling Proteins: An Overview from Physiological to Pathological Roles. Biomedicines. 2024;12:1307. https://doi.org/10.3390/biomedicines12061307 . Martínez-Reyes I, Chandel NS. Cancer metabolism: looking forward. Nat Rev Cancer. 2021;21:669–80. https://doi.org/10.1038/s41568-021-00378-6 . Faubert B, Solmonson A, DeBerardinis RJ. Metabolic reprogramming and cancer progression. Science. 2020;368:eaaw5473. https://doi.org/10.1126/science.aaw5473 . Y S, J C, Q D, Z T. Molecular characteristics of early- and late-onset ovarian cancer: insights from multidimensional evidence. J ovarian Res [Internet] J Ovarian Res; 2025 [cited 2026 Jan 26];18. https://doi.org/10.1186/s13048-025-01664-9 Im FP-V, Jp O-C, D M-L L-A Jr. Epicatechin Decreases UCP2 Gene Expression in MDA-MB-231 Breast Cancer Cells by the Presence of a Regulatory Element in the Promoter. International journal of molecular sciences [Internet]. Int J Mol Sci. 2025. https://doi.org/10.3390/ijms26094102 . [cited 2026 Jan 26];26. Supplementary Files SupplementaryFigures.docx SupplementaryTables.xlsx Cite Share Download PDF Status: Under Review Version 1 posted Reviewers agreed at journal 23 Feb, 2026 Reviewers invited by journal 23 Feb, 2026 Editor assigned by journal 18 Feb, 2026 First submitted to journal 17 Feb, 2026 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8869688","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":595895111,"identity":"9557cc4c-8168-4e86-9835-b42bc36ec79a","order_by":0,"name":"Linghui Tan","email":"","orcid":"","institution":"Zhejiang University School of Medicine Sir Run Run Shaw Hospital","correspondingAuthor":false,"prefix":"","firstName":"Linghui","middleName":"","lastName":"Tan","suffix":""},{"id":595895112,"identity":"b2b43c28-066a-4e94-b82b-9556f453f364","order_by":1,"name":"Tianlun Hou","email":"","orcid":"","institution":"Zhejiang University School of Medicine Sir Run Run Shaw Hospital","correspondingAuthor":false,"prefix":"","firstName":"Tianlun","middleName":"","lastName":"Hou","suffix":""},{"id":595895113,"identity":"d966a9a4-87bf-4806-aecd-5c802c2c9897","order_by":2,"name":"Pingting Ying","email":"","orcid":"","institution":"Zhejiang University School of Medicine Sir Run Run Shaw Hospital","correspondingAuthor":false,"prefix":"","firstName":"Pingting","middleName":"","lastName":"Ying","suffix":""},{"id":595895114,"identity":"954990ab-f8d8-41ef-bc7d-295c2a39e222","order_by":3,"name":"Xian Wang","email":"","orcid":"","institution":"Zhejiang University School of Medicine Sir Run Run Shaw Hospital","correspondingAuthor":false,"prefix":"","firstName":"Xian","middleName":"","lastName":"Wang","suffix":""},{"id":595895115,"identity":"46d2fae1-9f55-429e-8bba-a83d43995b30","order_by":4,"name":"Hongchuan Jin","email":"","orcid":"","institution":"Zhejiang University School of Medicine Sir Run Run Shaw Hospital","correspondingAuthor":false,"prefix":"","firstName":"Hongchuan","middleName":"","lastName":"Jin","suffix":""},{"id":595895116,"identity":"58969863-0ecd-4f44-b970-89942e8f5d23","order_by":5,"name":"Jingfeng Luo","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAuklEQVRIiWNgGAWjYBACPiA2qKhgYAZxJIjSwgbScuYMqVoYzrZBOERqkUg+UHBwXh27wQHmg7d5GOzyiNCSlmBwcBsbs8EBtmRrHobkYiK05BgYf9zGA9TCYybNw3AgsYEYLQYH50gAtfB/I0VLgwHIFjYitfA8SzA4cCyBWfIwm7HlHINkwlr42ZOPGRyoqUvmO9788MabCjvCWkAWGQCJZEhkGhChHgiYHwAJO+LUjoJRMApGwYgEAIMLM0rCj9x8AAAAAElFTkSuQmCC","orcid":"https://orcid.org/0000-0001-9266-6501","institution":"Zhejiang University School of Medicine Sir Run Run Shaw Hospital","correspondingAuthor":true,"prefix":"","firstName":"Jingfeng","middleName":"","lastName":"Luo","suffix":""}],"badges":[],"createdAt":"2026-02-13 09:01:13","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8869688/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8869688/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":103539703,"identity":"bfdf7c38-b977-4c85-b891-e2017efc479e","added_by":"auto","created_at":"2026-02-26 19:39:44","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":49424,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eStudy design and multi-omics landscape of FGFRi resistance.\u003c/strong\u003e (A) A systematic framework integrating pharmacogenomic profiling (GDSC2 and PRISM) and scRNA-seq was established to dissect FGFRi resistance. The study follows three parallel streams: (Left) Development of a machine learning-based predictive model for drug sensitivity; (Middle) Construction of a multi-dimensional resistance atlas, highlighting RTK switching (EGFR) and metabolic reprogramming (UCP2) as key drivers, validated in TCGA and CPTAC clinical cohorts; (Right) Identification of synthetic lethal vulnerabilities to propose actionable combination therapies. The dark blue box at the bottom illustrates the proposed precision medicine framework: 1. Stratification via the 52-gene model; 2. Molecular profiling to identify resistance mechanisms (RTK/Metabolic); and 3. Selection of rational combination therapies. (B, C) Availability of multi-omics profiles (mutations, CNVs, methylation, mRNA transcripts, and proteomic data) across cancer cohorts in the GDSC2 (B) and PRISM (C) databases.\u003c/p\u003e","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-8869688/v1/dd361c892dfce33d1006763a.png"},{"id":103539709,"identity":"fb3afd67-79e7-44e8-b383-7f4edf59db50","added_by":"auto","created_at":"2026-02-26 19:39:46","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":88989,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eThe Genomic Landscape of FGFRi Resistance Reveals Context-Specific Alterations.\u003c/strong\u003e (A) Box plots illustrating the differences in AZD4547 AUC values between wild-type and mutant cell lines of target genes in specific cancer types from the GDSC2 database. (B) Stacked bar charts showing the distribution frequency discrepancies of wild-type and mutant genotypes of specific genes in FGFRi-sensitive (S) versus resistant (R) cell line groups across cancer types in the PRISM database. (C) Correlation scatter plots demonstrating the association between gene methylation levels and AZD4547 AUC values in designated cancer types from the GDSC2 database. (D) Box plots comparing the methylation levels of target genes between FGFRi-sensitive and resistant cell line cohorts of specific cancer types in the PRISM database. (E) Analyses of CNV levels: Student’ s t-test was applied to quantify CNV differences between sensitive and resistant groups for each cancer type in the PRISM database; Spearman’ s rank correlation analysis was conducted to evaluate the association between CNV levels and AZD4547 AUC values in the GDSC2 database.\u003c/p\u003e","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-8869688/v1/47dddf92ffe4536f648e4604.png"},{"id":103539712,"identity":"32a8552f-8fce-461f-811e-da0f7351c919","added_by":"auto","created_at":"2026-02-26 19:39:48","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":134153,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eIdentification of FGFRi response-related mRNAs and proteins. (A) \u003c/strong\u003eScatter plot displaying representative FGFRi response-related mRNAs consistently identified from the GDSC2 and PRISM databases.\u003cstrong\u003e (B) \u003c/strong\u003eHeatmap illustrating the correlation between specific mRNA expression levels and AZD4547 AUC values (GDSC2 database), as well as mRNA expression fold changes between FGFRi-resistant and sensitive groups across cancer types (PRISM database). Statistical significance is indicated: *\u003cem\u003eP\u003c/em\u003e \u0026lt; 0.05; **\u003cem\u003eP\u003c/em\u003e \u0026lt; 0.01; ***\u003cem\u003eP\u003c/em\u003e \u0026lt; 0.001. The color scale represents mRNA expression fold change (resistant vs. sensitive) and the correlation coefficient between mRNA expression and AZD4547 AUC values. (C-D) Volcano plots depicting FGFRi response-related proteins identified from the GDSC2 (C) and PRISM (D) databases, respectively. (E) Heatmap showcasing the correlation between specific protein abundances and AZD4547 AUC values (GDSC2 database), along with protein abundance fold changes between FGFRi-resistant and sensitive groups across cancer types (PRISM database). Statistical significance is denoted: *\u003cem\u003eP\u003c/em\u003e \u0026lt; 0.05; **\u003cem\u003eP\u003c/em\u003e\u0026lt; 0.01; ***\u003cem\u003eP\u003c/em\u003e \u0026lt; 0.001. The color scale indicates protein abundance fold change (resistant vs. sensitive) and the correlation coefficient between protein abundance and AZD4547 AUC values.\u003c/p\u003e","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-8869688/v1/7c99ea85905540e6e3b9626c.png"},{"id":104398407,"identity":"cd29baf9-bc0d-4c61-9f74-4ce3927112ec","added_by":"auto","created_at":"2026-03-11 12:02:14","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":112735,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eIdentification of FGFRi response-related pathways. \u003c/strong\u003e(A) Heatmap of GSEA results from four datasets (GDSC2-mRNA, GDSC2-protein, PRISM-mRNA, PRISM-protein). *FDR \u0026lt; 0.05; **FDR \u0026lt; 0.01; ***FDR \u0026lt; 0.001. Color gradient indicates Normalized Enrichment Score (NES).\u003cstrong\u003e \u003c/strong\u003e(B) Left: Correlation scatter plots of \u003cem\u003eATP1B3\u003c/em\u003e and \u003cem\u003eUCP2\u003c/em\u003emRNA levels vs. AZD4547 AUC in the COAD cohort (GDSC2). Right: Box plots of \u003cem\u003eATP1B3\u003c/em\u003eand \u003cem\u003eUCP2\u003c/em\u003e mRNA levels in FGFRi-sensitive vs. resistant groups (COAD, PRISM).\u003cstrong\u003e \u003c/strong\u003e(C) Survival analyses of \u003cem\u003eUCP2\u003c/em\u003e high- vs. low-expression subgroups in the CPTAC-COAD (left) and TCGA-COAD (right) cohorts.\u003cstrong\u003e \u003c/strong\u003e(D) Left: Correlation scatter plot of \u003cem\u003eFSCN1\u003c/em\u003e mRNA vs. AZD4547 AUC in the BLCA cohort (GDSC2). Right: Box plot of \u003cem\u003eFSCN1\u003c/em\u003e mRNA in FGFRi-sensitive vs. resistant groups (BLCA, PRISM).\u003cstrong\u003e \u003c/strong\u003e(E) Survival analysis of \u003cem\u003eFSCN1\u003c/em\u003ehigh- vs. low-expression subgroups in the TCGA-BLCA cohort.\u003cstrong\u003e \u003c/strong\u003e(F) Left: Correlation scatter plot of \u003cem\u003eMCM7\u003c/em\u003e mRNA vs. AZD4547 AUC in the PAAD cohort (GDSC2). Right: Box plot of \u003cem\u003eMCM7\u003c/em\u003e mRNA in FGFRi-sensitive vs. resistant groups (PAAD, PRISM).\u003cstrong\u003e \u003c/strong\u003e(G) Survival analyses of \u003cem\u003eMCM7\u003c/em\u003e high- vs. low-expression subgroups in the CPTAC-PAAD (left) and TCGA-PAAD (right) cohorts.\u003cstrong\u003e \u003c/strong\u003e(H) Left: Correlation scatter plots of \u003cem\u003eHPRT1\u003c/em\u003e and \u003cem\u003eMCM4\u003c/em\u003emRNA levels vs. AZD4547 AUC in the GBM cohort (GDSC2). Right: Box plots of \u003cem\u003eHPRT1\u003c/em\u003eand \u003cem\u003eMCM4\u003c/em\u003e mRNA in FGFRi-sensitive vs. resistant groups (GBM, PRISM).\u003cstrong\u003e \u003c/strong\u003e(I) Survival analysis of \u003cem\u003eMCM4\u003c/em\u003e high- vs. low-expression subgroups in the TCGA-GBM cohort.\u003cstrong\u003e \u003c/strong\u003e(J) Survival analyses of \u003cem\u003eHPRT1\u003c/em\u003e high- vs. low-expression subgroups in the CPTAC-GBM (left) and TCGA-GBM (right) cohorts.\u003c/p\u003e","description":"","filename":"Onlinefloatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-8869688/v1/c983ba434faf898f2b6c06b1.png"},{"id":103539716,"identity":"6d13d163-5d09-4863-b630-bb28ba9d8987","added_by":"auto","created_at":"2026-02-26 19:39:52","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":142410,"visible":true,"origin":"","legend":"\u003cp\u003eDistinct multi-omics regulatory landscapes governing FGFRi resistance. (A) Heatmap exhibiting the transcriptional regulatory network in GBM (Resistant phenotype), showing Spearman correlation coefficients between TF (columns) and target mRNA expression (rows). (B) Heatmap exhibiting the multi-omics associations in BLCA, including the impact of CNVs on protein and mRNA levels, as well as TF-mRNA regulatory pairs. (C) Epigenetic regulation in BRCA, highlighting the correlation between DNA methylation and gene expression. (D) Transcriptional regulatory patterns in PAAD, showing associations between TFs and downstream mRNA targets. (E) Integrated multi-omics interaction network visualizing core regulatory pathways. For all heatmaps, the color gradient reflects the correlation coefficient (Pink: positive correlation; Green: negative correlation). Statistical significance is denoted by asterisks: *\u003cem\u003eP\u003c/em\u003e \u0026lt; 0.05; **\u003cem\u003eP\u003c/em\u003e \u0026lt; 0.01; ***\u003cem\u003eP\u003c/em\u003e\u0026lt; 0.001.\u003c/p\u003e","description":"","filename":"Onlinefloatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-8869688/v1/52f385abea61f1da0043dab7.png"},{"id":104397772,"identity":"54ac7261-11f2-454d-94a1-6a1c29361b6d","added_by":"auto","created_at":"2026-03-11 11:56:13","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":117572,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eIdentification of gene MPs linked to FGFRi response.\u003c/strong\u003e (A) t-SNE visualization of cell lines from distinct cancer types (colored by cancer type). (B) Pie chart depicting the proportion of cell counts across various cancer types. (C) Bar chart showing the distribution of cell counts in FGFRi-resistant and sensitive groups across cancer types. (D) NMF results of pan-cancer cell line genes and heatmap analysis of four gene MPs. The number of core genes for each metaprogram: MP1: \u003cem\u003en\u003c/em\u003e= 3; MP2: \u003cem\u003en\u003c/em\u003e = 32; MP3: \u003cem\u003en\u003c/em\u003e = 68; MP4: \u003cem\u003en\u003c/em\u003e = 52). (E) GO enrichment analysis of genes in each metaprogram. (F) Bubble chart of metaprogram scores computed via the AUCell algorithm; (G-H) Heatmaps of average metaprogram scores in FGFRi-sensitive and resistant groups across cancer types, calculated using the AUCell algorithm (G) and AddModuleScore algorithm (H), respectively.\u003c/p\u003e","description":"","filename":"Onlinefloatimage6.png","url":"https://assets-eu.researchsquare.com/files/rs-8869688/v1/2ba22db7e581acd83927ed52.png"},{"id":104397634,"identity":"ee15a4d0-15a3-4635-8ae4-fa1df7fa0a32","added_by":"auto","created_at":"2026-03-11 11:53:31","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":86205,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eDevelopment of a machine learning model for predicting FGFRi response.\u003c/strong\u003e (A) Workflow of the prediction model construction. (B-C) In the PRISM training cohort, the optimal λ value was identified when the partial likelihood deviance reached its minimum (B), and the corresponding Lasso coefficients of the most relevant genes were derived (C). (D-E) ROC curves demonstrating the predictive performance of each model in the validation set (D) and test set (E). (F-G) Heatmaps presenting the performance of seven models across multiple metrics in the validation set (F) and test set (G). (H) Box plot comparing the AZD4547 AUC values of cell lines predicted as FGFRi-sensitive or resistant by the RF model in the GDSC2 database. (I) The top 10 most significant features for predicting response to FGFRi, ranked from most to least important. (J) Distribution of the impact of each feature on the model output. The colors represent feature values, with red indicating higher values and blue indicating lower values.\u003c/p\u003e","description":"","filename":"Onlinefloatimage7.png","url":"https://assets-eu.researchsquare.com/files/rs-8869688/v1/55f49f2d52daa9e76ba61a5a.png"},{"id":103539717,"identity":"08d81c18-8ecd-42ab-b8b6-b97b1ca2634e","added_by":"auto","created_at":"2026-02-26 19:39:53","extension":"png","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":66010,"visible":true,"origin":"","legend":"\u003cp\u003eFunctional characterization of synthetic lethal vulnerabilities associated with the FGF/FGFR axis. (A) Gene Ontology (GO) enrichment analysis of identified synthetic lethal candidates. The bar plot illustrates significantly enriched biological processes, systematically categorized into five core functional modules: Signaling Transduction, Metabolic Reprogramming, Cell Cycle \u0026amp; Proliferation, Invasion, and Immunity. The x-axis represents the count of genes involved in each term, and the color gradient indicates statistical significance (–log10 adjusted P-value). (B) Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis. The analysis highlights the reactivation of canonical survival cascades (PI3K-Akt, MAPK) and specifically identifies the \"EGFR tyrosine kinase inhibitor resistance\" pathway, supporting the RTK switching mechanism.\u003c/p\u003e","description":"","filename":"Onlinefloatimage8.png","url":"https://assets-eu.researchsquare.com/files/rs-8869688/v1/1b9e9fc17202b48e46ad1213.png"},{"id":107489531,"identity":"6041678c-0f32-4dd7-944b-146be87bd577","added_by":"auto","created_at":"2026-04-22 02:48:00","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1670371,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8869688/v1/8c039be1-ae17-459c-b686-d4662453d91f.pdf"},{"id":104397989,"identity":"88eaf294-c75e-49e8-b96d-75d6ff272435","added_by":"auto","created_at":"2026-03-11 11:59:09","extension":"docx","order_by":13,"title":"","display":"","copyAsset":false,"role":"supplement","size":975822,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryFigures.docx","url":"https://assets-eu.researchsquare.com/files/rs-8869688/v1/8422d4d460cbd7847d1304e5.docx"},{"id":104397780,"identity":"a0da0028-4351-475d-9ad8-71499484d234","added_by":"auto","created_at":"2026-03-11 11:56:23","extension":"xlsx","order_by":14,"title":"","display":"","copyAsset":false,"role":"supplement","size":663623,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryTables.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-8869688/v1/bc8aea16ee77acb7cec463d4.xlsx"}],"financialInterests":"","formattedTitle":"Integrative Proteogenomic and Single-Cell Analysis Reveals RTK Switching and Metabolic Reprogramming as Synthetic Lethal Vulnerabilities in FGFR Inhibitor Resistance","fulltext":[{"header":"Background","content":"\u003cp\u003eThe Fibroblast Growth Factor Receptor (FGFR) family (FGFR1-4), a subset of the receptor tyrosine kinase (RTK) superfamily, governs critical physiological processes ranging from embryonic development to tissue repair[\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]. Upon binding to Fibroblast Growth Factor (FGF) ligands, FGFR initiates cascading reactions of downstream signaling pathways, including PI3K-AKT and RAS-MAPK, thereby regulating cell proliferation, differentiation, migration, and survival[\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e, \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]. Aberrant activation of FGFR signaling\u0026mdash;driven by gene fusion, mutation, or amplification\u0026mdash;has been established as a pivotal oncogenic driver across a diverse spectrum of malignancies[\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e, \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. Consequently, targeting this axis with small-molecule tyrosine kinase inhibitors (TKIs), such as the selective inhibitor AZD4547, has emerged as a cornerstone of precision oncology[\u003cspan additionalcitationids=\"CR6 CR7 CR8\" citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]. Despite the transformative potential of these therapies, the clinical reality remains sobering. While initial responses in patients with FGFR alterations\u0026mdash;particularly those with urothelial carcinoma or cholangiocarcinoma\u0026mdash;are encouraging, they are often transient[\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]. The inevitable emergence of acquired resistance severely limits the durability of response, typically resulting in disease progression within months.\u003c/p\u003e \u003cp\u003eCurrent paradigms of FGFR inhibitor (FGFRi) resistance are predominantly categorized into four mechanisms: (1) Secondary activating mutations in the FGFR kinase domain, leading to impaired inhibitor binding or sustained enhancement of kinase activity[\u003cspan additionalcitationids=\"CR13\" citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e]. (2) Compensatory activation of alternative RTKs (e.g., EGFR or MET), bypassing signal transduction mediated by the FGFR pathway[\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e, \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e]. (3) Reactivation of downstream pathways such as PI3K-AKT and RAS-MAPK, offsetting the growth inhibitory effect after FGFR inhibition[\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e]. (4) Mutations in tumor suppressor genes such as \u003cem\u003eTP53\u003c/em\u003e, regulating cell cycle checkpoints or DNA damage repair capacity and reducing drug sensitivity[\u003cspan additionalcitationids=\"CR19\" citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eAmong these, 'RTK switching' has emerged as a prominent yet complex escape strategy. Notably, Wu et al. demonstrated that compensatory EGFR activation drives resistance specifically in FGFR2-fusion positive cholangiocarcinoma, providing a rationale for dual blockade[\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e]. However, our understanding of this phenomenon remains fragmented and confined to isolated cancer types. Critical systemic questions remain unanswered: Is this EGFR-mediated bypass a universal vulnerability across the pan-cancer landscape, or is it a context-dependent event? More importantly, does this signaling rewiring operate in isolation, or is it merely one facet of a multimodal adaptive strategy that also involves metabolic reprogramming? Crucially, existing studies predominantly focused on distinct molecular layers in isolation, thereby failing to capture the concurrent evolution of signaling networks and metabolic states.\u003c/p\u003e \u003cp\u003eTo bridge this gap, we designed a comprehensive multi-omics framework leveraging large-scale pharmacogenomic data from 312 cell lines across eight major cancer types: Bladder Urothelial Carcinoma (BLCA), Breast Invasive Carcinoma (BRCA), Colon Adenocarcinoma (COAD), Esophageal Squamous Cell Carcinoma (ESCC), Glioblastoma (GBM), Lung Adenocarcinoma (LUAD), Melanoma (MEL), and Pancreatic Adenocarcinoma (PAAD). To ensure robustness, we implemented a rigorous cross-validation framework utilizing two independent datasets, GDSC2 and PRISM. Beyond traditional bulk-level profiling, we integrated single-cell RNA sequencing (scRNA-seq) to dissect the clonal evolution of resistance at cellular resolution. Finally, we translated these biological insights into actionable strategies by constructing a machine learning-based predictive model and conducting systematic synthetic lethality screening.\u003c/p\u003e \u003cp\u003eIn this study, we present a multi-dimensional atlas of FGFRi resistance. We uncover that resistance is driven not only by genetic alterations but also by a convergent evolution involving metabolic reprogramming (e.g., \u003cem\u003eUCP2\u003c/em\u003e upregulation) and RTK switching (specifically towards EGFR signaling). By integrating these mechanistic insights with synthetic lethal vulnerability mapping, we propose a rational framework for tiered precision combination therapies, offering new avenues to overcome the clinical bottleneck of FGFRi resistance.\u003c/p\u003e"},{"header":"Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eDownload of cell line multi-omics data and AZD4547 sensitivity data\u003c/h2\u003e \u003cp\u003eMulti-dimensional omics data of pan-cancer single cell lines, including gene mutation data, copy number variation (CNV) data, DNA methylation data, transcriptome sequencing data, and proteome quantification data, were systematically downloaded from the DepMap database (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://depmap.org/portal/\u003c/span\u003e\u003cspan address=\"https://depmap.org/portal/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e). Meanwhile, drug response data from the Genomics of Drug Sensitivity in Cancer (GDSC2) database (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.cancerrxgene.org/\u003c/span\u003e\u003cspan address=\"https://www.cancerrxgene.org/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) of the Sanger Institute and the PRISM Repurposing Public 24Q2 database (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://depmap.org/portal/prism/\u003c/span\u003e\u003cspan address=\"https://depmap.org/portal/prism/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) of the Broad Institute were obtained.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eFGFR-abnormal cell line screening criteria\u003c/h3\u003e\n\u003cp\u003eBased on gene variation annotation information, tumor cell lines with FGFR abnormalities were screened. The inclusion criteria included: point mutations, covering functional impact mutations such as frameshift_variant, splice_acceptor_variant, splice_donor_variant, start_lost, stop_gained, stop_lost, protein_altering_variant, and missense mutation; copy number variations, gene amplification with copy number\u0026thinsp;\u0026ge;\u0026thinsp;4; gene fusions, cell lines with FGFR family gene fusion events.\u003c/p\u003e \u003cp\u003eIn the PRISM database, drug sensitivity was represented by the log-fold change (LFC), calculated as: LFC\u0026thinsp;=\u0026thinsp;log₂(number of viable cells in the experimental group / number of viable cells in the control group). An LFC\u0026thinsp;\u0026lt;\u0026thinsp;0 indicates that the drug has an inhibitory effect on the cell line, and a smaller LFC value indicates a stronger inhibitory effect. To maximize the biological contrast between responsive and non-responsive phenotypes, the grouping criteria were strictly defined: LFC\u0026thinsp;\u0026gt;\u0026thinsp;0 as the FGFR inhibitor (FGFRi) resistant group, and LFC\u0026thinsp;\u0026lt;\u0026thinsp;0 as the FGFRi sensitive group. To ensure statistical test power and robust feature extraction, only cancer types with \u0026ge;\u0026thinsp;3 cell lines in both the sensitive and resistant groups were retained for subsequent analysis. Crucially, to mitigate the potential bias from this binary classification, all candidate markers derived from this grouping were further cross-validated using continuous drug response metrics (AUC) from the independent GDSC2 database to ensure the robustness of the findings.\u003c/p\u003e \u003cp\u003eAfter the above screening process, proteogenomic data of 312 eligible tumor cell lines were finally obtained, covering 8 cancer types (BLCA, BRCA, COAD, ESCC, GBM, LUAD, MEL, and PAAD). Detailed information of the screened cell lines was provided in Supplementary Tables S1 and S2.\u003c/p\u003e\n\u003ch3\u003eDrug response analysis in the GDSC2 database\u003c/h3\u003e\n\u003cp\u003eFor each cancer cohort, Student\u0026rsquo;s t-test was used to compare the difference in the Area Under the Curve (AUC) of the AZD4547 drug response curve between FGFR-abnormal mutant and wild-type (WT) cell lines (a lower AUC value indicates higher drug sensitivity). P-values were corrected for multiple testing using the Benjamini-Hochberg method. Screening criteria: genes with adjusted \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05 and Fold Change (FC)\u0026thinsp;\u0026ge;\u0026thinsp;1.2 were defined as FGFRi resistance-related mutant genes; genes with adjusted \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05 and FC\u0026thinsp;\u0026le;\u0026thinsp;0.8 were defined as FGFRi sensitivity-related mutant genes.\u003c/p\u003e \u003cp\u003eFor the correlation between CNV, mRNA, and protein levels and drug response, Spearman rank correlation analysis was used to detect their correlation with the AZD4547 AUC value, and \u003cem\u003eP\u003c/em\u003e-values were corrected using the Benjamini-Hochberg method. Judgment criteria: molecules with adjusted \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05 and Spearman correlation coefficient (r)\u0026thinsp;\u0026ge;\u0026thinsp;0.1 were defined as FGFRi resistance-related molecules; molecules with adjusted \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05 and r \u0026le; -0.1 were defined as FGFRi sensitivity-related molecules.\u003c/p\u003e\n\u003ch3\u003eDrug response analysis in the PRISM database\u003c/h3\u003e\n\u003cp\u003eFisher\u0026rsquo;s exact test was used to compare the distribution difference of gene variations (mutant/wild-type) between the FGFRi resistant and sensitive groups in each cancer cohort, and \u003cem\u003eP\u003c/em\u003e-values were corrected using the Benjamini-Hochberg method; genes with adjusted \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05 were considered drug response-related mutant genes, among which genes with significantly enriched mutants in the resistant group were FGFRi resistance-related mutant genes, and the rest were FGFRi sensitivity-related mutant genes.\u003c/p\u003e \u003cp\u003eUnpaired Student\u0026rsquo;s t-test was used to compare the differences in CNV levels, mRNA expression levels, and protein abundances between the resistant and sensitive groups, and P-values were corrected using the Benjamini-Hochberg method. Screening criteria: molecules with adjusted \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05 and FC\u0026thinsp;\u0026ge;\u0026thinsp;1.2 were defined as FGFRi resistance-related molecules; molecules with adjusted \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05 and FC\u0026thinsp;\u0026le;\u0026thinsp;0.8 were defined as FGFRi sensitivity-related molecules.\u003c/p\u003e\n\u003ch3\u003eCross-database screening of high-confidence molecules\u003c/h3\u003e\n\u003cp\u003eTo ensure the robustness of our findings, we employed a strict cross-validation strategy using GDSC2 and PRISM as independent discovery cohorts. Statistical analyses, including differential expression profiling and drug response correlation, were performed independently within each dataset. Overlapping cell lines between the two datasets were retained to assess technical robustness against experimental variations.\u003c/p\u003e \u003cp\u003eHigh-confidence FGFRi resistance/sensitivity-related molecules were defined based on the following criteria:\u003c/p\u003e \u003cp\u003e \u003col\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003e1. Consistency in Significance: The feature (mutation, CNV, mRNA, or protein) must be statistically significant (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05) in both datasets.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003e2. Concordance in Direction: The feature must exhibit the same direction of association (e.g., consistently upregulated or positively correlated with drug sensitivity) in both cohorts. Only molecular features satisfying these intersection criteria were retained for subsequent functional enrichment and mechanistic analyses.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003c/ol\u003e \u003c/p\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eGO and Reactome pathway enrichment analysis\u003c/h2\u003e \u003cp\u003eGene Ontology (GO) and Reactome pathway enrichment analyses were performed on high-confidence FGFRi resistance/sensitivity-related mRNAs and proteins using the R package \u0026ldquo;BioEnricher\u0026rdquo;. An adjusted \u003cem\u003eP\u003c/em\u003e value\u0026thinsp;\u0026lt;\u0026thinsp;0.05 was used as the statistical significance criterion to screen differentially enriched biological pathways.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eGene Set Enrichment Analysis (GSEA)\u003c/h3\u003e\n\u003cp\u003eGSEA was performed using the R package clusterProfiler, with the background gene set selected from 50 hallmark gene sets in the MSIGDB database. A gene ranking list was constructed for each cancer cohort: in the GDSC2 database, genome-wide mRNAs and proteins were included and ranked by their Spearman correlation coefficient with the AZD4547 AUC value; in the PRISM database, genome-wide mRNAs and proteins were included and ranked by the FC value between the resistant and sensitive groups. The analysis parameters were set to 1000 permutations, with a False Discovery Rate (FDR)\u0026thinsp;\u0026lt;\u0026thinsp;0.05 as the significant enrichment threshold; pathways with a Normalized Enrichment Score (NES)\u0026thinsp;\u0026gt;\u0026thinsp;1 were defined as FGFRi resistance-related candidate pathways, and pathways with NES \u0026lt; -1 were defined as FGFRi sensitivity-related candidate pathways. Criteria for defining significant pathways: each pathway was required to obtain four types of enrichment results (GDSC2-mRNA, GDSC2-protein, PRISM-mRNA, PRISM-protein), among which at least 2 types met the significant enrichment criteria, and the functional directions (resistance/sensitivity) of all significant results were consistent to be identified as a significantly enriched pathway related to FGFRi response in that cancer type. The intersection of molecules in significantly enriched pathways with high-confidence FGFRi resistance/sensitivity-related mRNAs and proteins was taken to determine the core regulatory molecules of the FGFRi response phenotype.\u003c/p\u003e\n\u003ch3\u003eRegulatory effect of gene mutations on mRNA/Protein expression\u003c/h3\u003e\n\u003cp\u003eAmong high-confidence FGFRi resistance/sensitivity-related mutant genes, the effect of their mutation status on the expression of core regulatory mRNAs and proteins was explored. By integrating SNV, mRNA, and protein data of cell lines from the GDSC2 and PRISM databases, Student\u0026rsquo;s t-test was used to compare the differences in the expression levels of core regulatory mRNAs and protein abundances between mutant and wild-type cell lines, and \u003cem\u003eP\u003c/em\u003e-values were corrected using the Benjamini-Hochberg method; mutation-mRNA/protein combinations with adjusted \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05 were defined as significant regulatory relationships.\u003c/p\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eCorrelation analysis of CNV, methylation, transcription factors (TFs) with mRNA/Protein\u003c/h2\u003e \u003cp\u003eA list of TFs was extracted from high-confidence FGFRi sensitivity/resistance-related mRNAs. CNV, methylation, and transcription factor expression data from the GDSC2 and PRISM databases were integrated, and Spearman rank correlation analysis was used to detect the correlation between each CNV, methylation site, TFs expression level and the expression of core regulatory mRNAs/proteins, respectively. After P-value correction by the Benjamini-Hochberg method, significant CNV-mRNA/protein, methylation-mRNA/protein, and transcription factor-mRNA/protein regulatory associations were determined with adjusted \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05 as the criterion.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003eDownload of bulk multi-omics data from public cancer patients\u003c/h2\u003e \u003cp\u003eMulti-omics datasets and matched clinicopathological metadata from The Cancer Genome Atlas (TCGA) were downloaded from the official TCGA data portal (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://gdc.cancer.gov\u003c/span\u003e\u003cspan address=\"https://gdc.cancer.gov\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e). In contrast, data corresponding to the Clinical Proteomic Tumor Analysis Consortium (CPTAC) cohort were derived from a prior publication by Li et al[\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e].\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003eDownload and processing of single-cell expression data of pan-cancer cell Lines\u003c/h2\u003e \u003cp\u003eSingle-cell expression profiles and clinical annotation data of cancer cell lines were downloaded from the GEO database (GSE157220), which contains single-cell expression data of 205 cancer cell lines (280 cells per cell line). Single-cell expression data were merged with drug sensitivity data from the PRISM database, and quality control and preprocessing were performed using the Seurat software package[\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e]. The filtering criteria were as follows: cells with the number of detected genes (nFeature_RNA)\u0026thinsp;\u0026gt;\u0026thinsp;200 were retained; cells with nFeature_RNA\u0026thinsp;\u0026gt;\u0026thinsp;5000 (high-complexity abnormal cells) were removed; cells with mitochondrial gene ratio\u0026thinsp;\u0026lt;\u0026thinsp;15% were screened; cells with gene expression count (nCount_RNA)\u0026thinsp;\u0026gt;\u0026thinsp;400 were retained; genes expressed in at least 5 cells were retained. After the above filtering, 9997 cells were finally retained for subsequent analysis. The SCTransform function of the Seurat package was used to normalize the single-cell expression data, principal component analysis (PCA) was performed using the RunPCA function, and the top 20 principal components were selected for subsequent cell clustering and dimensionality reduction analysis. Non-linear dimensionality reduction and visualization were performed using the RunTSNE function.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003ePan-cancer gene metaprograms analysis\u003c/h2\u003e \u003cp\u003eGene metaprograms identification was performed using the R package geneNMF. First, the Seurat object was split by sample ID, and multi-k value (k\u0026thinsp;=\u0026thinsp;4\u0026thinsp;~\u0026thinsp;9) non-negative matrix factorization (NMF) analysis was performed on each sample using the multiNMF function, and 800 highly variable genes were selected to identify gene programs of different dimensions; the similarity between gene programs was evaluated based on cosine similarity/Jaccard index, and highly similar gene programs were aggregated into robust metaprograms (MPs) using the getMetaPrograms function, with a weight explanation threshold (weight.explained\u0026thinsp;=\u0026thinsp;0.6) and a maximum number of genes per metaprogram (max.genes\u0026thinsp;=\u0026thinsp;300) set to ensure the biological validity of metaprograms; for the number of metaprograms with k\u0026thinsp;=\u0026thinsp;4\u0026thinsp;~\u0026thinsp;9, quantitative indicators such as sample coverage, silhouette coefficient, average similarity, number of core genes, and number of original gene program integrations were calculated, and combined with heatmap visualization of the similarity clustering characteristics between metaprograms to determine the optimal number of metaprograms; GO functional enrichment analysis was performed on the core gene sets of each metaprogram, and significantly enriched biological pathways were screened with adjusted \u003cem\u003eP\u003c/em\u003e value\u0026thinsp;\u0026lt;\u0026thinsp;0.05 as the threshold; metaprogram scores of individual cells were calculated using two algorithms, AddModuleScore and AUCell, respectively.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec15\" class=\"Section2\"\u003e \u003ch2\u003eConstruction of FGFRi Sensitivity Prediction Model Based on Machine Learning\u003c/h2\u003e \u003cp\u003eFirst, we used the stratified sampling method of the createDataPartition function in the R package caret to divide the PRISM dataset into a training set (70%), a validation set (20%), and a test set (10%) at a ratio of 7:2:1, ensuring that the ratio of sensitive/resistant samples in each subset was consistent. Lasso (Least Absolute Shrinkage and Selection Operator) regression (the penalty parameter α set to 1) in the R package glmnet was used for feature selection in the training set, and the optimal λ value (lambda.min) that minimized the prediction error was determined through 10-fold cross-validation, with genes with non-zero coefficients in the model retained as candidate features (lambda.min\u0026thinsp;=\u0026thinsp;0.05366335). Based on the candidate features, 7 mainstream machine learning algorithms were used for FGFRi sensitive/resistant binary classification prediction using the mlr3verse ecosystem (mlr3, mlr3fselect, mlr3tuning packages) of R language, including: Random Forest (RF; classif.ranger), Logistic Regression (LogReg; classif.log_reg), Decision Tree (DT; classif.rpart), Extreme Gradient Boosting (XGBoost; classif.xgboost), k-Nearest Neighbor (kNN; classif.kknn), Naive Bayes (Naive Bayes; classif.naive_bayes), and Multilayer Perceptron (MLP; classif.nnet). Hyperparameter tuning was performed for each algorithm using the tnr (\"random_search\") random search method with 10\u0026thinsp;~\u0026thinsp;500 iterations (term_evals), and rsmp (\"holdout\") leave-one-out resampling and classification accuracy (classif.acc) were used as optimization indicators. The final model was obtained by retraining on the complete training set based on this combination. The model performance was quantified using multi-dimensional indicators on the validation set and test set: the main evaluation indicator was the Area Under the Receiver Operating Characteristic Curve (AUC; classif.auc), and the secondary evaluation indicators included accuracy (ACC), precision, recall, sensitivity, specificity, Negative Predictive Value (NPV), and Positive Predictive Value (PPV). Meanwhile, the true positive/false positive/true negative/false negative rates were analyzed through the confusion matrix. The core hyperparameters of the optimal random forest model determined via hyperparameter optimization are set as follows: number of decision trees (num.trees)\u0026thinsp;=\u0026thinsp;15, number of randomly selected features per tree (mtry)\u0026thinsp;=\u0026thinsp;3, minimum number of samples per node (min.node.size)\u0026thinsp;=\u0026thinsp;3, and maximum tree depth (max.depth)\u0026thinsp;=\u0026thinsp;6. For the model validation, overlapping cell lines from GDSC2 dataset were excluded to ensure independence.\u003c/p\u003e \u003cp\u003eTo elucidate the prediction mechanism of the optimally performing RF model and to elucidate the contributions of key features to drug efficacy predictions, we employed Shapley Additive Explanations (SHAP) for an interpretable analysis of the model. Grounded in the Shapley Value principle from game theory, SHAP breaks down the model's predictions into the cumulative contribution of each input feature, allowing for precise interpretation of individual predictions and a quantitative assessment of global feature importance. We computed the SHAP values for all input features using the trained optimal random forest model. We utilized the mean absolute SHAP value of each feature as a quantitative measure of feature importance, ranked the features accordingly, and identified the core elements that significantly influence drug efficacy predictions.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003eScreening of Synthetic Lethal Vulnerabilities\u003c/h2\u003e \u003cp\u003eTo identify robust synthetic lethal (SL) partners for the FGF/FGFR family, we developed a systematic computational framework integrating pharmacogenomic profiling, functional genomic screens, and biological relevance filtering. The screening pipeline was conducted across 8 cancer types based on the following steps:\u003c/p\u003e \u003cp\u003e1. Genotype Stratification\u003c/p\u003e \u003cp\u003eFor each query gene (designated as Gene A), cell lines within a specific cancer type were stratified into \"Gene A-inactive\" and \"Gene A-active\" cohorts. A cell line was classified as \"Gene A-inactive\" if it met at least one of the following characteristics:\u003c/p\u003e \u003cp\u003e \u003col\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eDeleterious Mutation: Harboring a somatic mutation classified as damaging, including frameshift variants, splice acceptor/donor variants, start lost, stop gained, and stop lost. Missense variants were included only if predicted as \"deleterious\" by both SIFT and PolyPhen-2 algorithms.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eCopy Number Loss: Gene copy number falling within the bottom 25th percentile of all cell lines in the corresponding cancer type.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eLow Expression: mRNA expression levels (TPM) falling within the bottom 25th percentile of all cell lines in the corresponding cancer type.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003c/ol\u003e \u003c/p\u003e \u003cp\u003eCell lines failing to meet any of these criteria were classified as \"Gene A-active\".\u003c/p\u003e \u003cp\u003e2. Pharmacogenomic Interaction Profiling\u003c/p\u003e \u003cp\u003eWe utilized drug sensitivity data from the PRISM database to identify compounds exhibiting selective lethality in the Gene A-inactive cells. For each compound, we compared the log-fold change (LFC) values between Gene A-inactive and Gene A-active groups using an unpaired Student\u0026rsquo;s t-test. Drugs showing significantly lower viability in the inactive group were considered significant hits (Benjamini-Hochberg adjusted \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05). The annotated molecular targets of these compounds were extracted as potential synthetic lethal partners (Candidate Gene B).\u003c/p\u003e \u003cp\u003e3. CRISPR-Cas9 Functional Validation\u003c/p\u003e \u003cp\u003eTo corroborate pharmacological findings with genetic evidence, we analyzed genome-wide CRISPR-Cas9 gene dependency scores from the DepMap portal. We compared the dependency scores of the Candidate Gene B list between Gene A-inactive and Gene A-active groups. Genes exhibiting a significantly lower dependency score (indicating higher essentiality) in the Gene A-inactive group were prioritized.\u003c/p\u003e \u003cp\u003e4. Biological Functional Similarity Filtering\u003c/p\u003e \u003cp\u003eTo ensure biological plausibility and reduce false positives, we evaluated the functional relatedness between the query gene (Gene A) and candidate partners (Gene B). Functional similarity scores were calculated based on Gene Ontology (GO) terms using the R package GOSemSim. Only gene pairs with score\u0026thinsp;\u0026gt;\u0026thinsp;0.3 were regarded as potential SL pairs.\u003c/p\u003e \u003cp\u003eFinally, to define a high-confidence set of SL pairs, we applied an intersection approach: only genes that demonstrated consistent lethality in both pharmacogenomic and CRISPR screens, and shared significant functional similarity (score\u0026thinsp;\u0026gt;\u0026thinsp;0.3) with Gene A, were retained as final SL targets.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec17\" class=\"Section2\"\u003e \u003ch2\u003eStatistical analysis\u003c/h2\u003e \u003cp\u003eStatistical analysis was performed with R software (version 4.2.6). Spearman\u0026rsquo; s correlation coefficients were utilized to assess the associations between two continuous variables. For categorical data, fisher\u0026rsquo;s exact test was applied, and for continuous variables, the wilcoxon rank-sum test or the t-test was used for comparison. The \u003cem\u003eP\u003c/em\u003e-value correction was performed using the Benjamini-Hochberg method, with statistical significance defined as an adjusted \u003cem\u003eP\u003c/em\u003e-value\u0026thinsp;\u0026lt;\u0026thinsp;0.05.\u003c/p\u003e \u003cp\u003eEthical consideration\u003c/p\u003e \u003cp\u003eThis study utilized publicly available, de-identified datasets from the GEO (GSE157220), DepMap, GDSC2, PRISM, TCGA, and CPTAC repositories. All original data collection and sharing protocols were conducted in accordance with the ethical standards and informed consent requirements of the respective source institutions. Since this research involved only the secondary analysis of anonymized public data, further institutional review board (IRB) approval was not required.\u003c/p\u003e \u003c/div\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec19\" class=\"Section2\"\u003e \u003ch2\u003eStudy cohort and proteogenomic profiling\u003c/h2\u003e \u003cp\u003eTo systematically dissect the molecular mechanisms underlying FGFRi resistance, we constructed a comprehensive multi-omics framework integrating pharmacogenomic data from the GDSC2 and PRISM databases (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eA). Our analysis encompassed somatic mutations, DNA methylation, CNVs, transcriptomics, and proteomics profiles across 312 cell lines spanning the eight cancer lineages (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eB, C). To bridge the gap between in vitro findings and clinical reality, we leveraged patient cohorts from TCGA and CPTAC to validate the prognostic significance of candidate biomarkers identified from these bulk profiles. Beyond bulk analysis, we incorporated scRNA-seq data to capture resistance heterogeneity at cellular resolution. Furthermore, to translate these multi-omics insights into therapeutic strategies, we developed a machine learning-based predictive model and performed systematic synthetic lethality screening. The overall study design and the multi-dimensional data availability are illustrated in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eA-C.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec20\" class=\"Section2\"\u003e \u003ch2\u003eThe Genomic Landscape of FGFRi Resistance Reveals Context-Specific Alterations\u003c/h2\u003e \u003cp\u003eTo characterize the genomic features associated with FGFRi sensitivity, we systematically profiled somatic mutations, DNA methylation patterns, and CNVs across diverse cancer lineages. Profiling of the GDSC2 MEL cohort revealed that cell lines harboring mutations in \u003cem\u003eARHGAP33\u003c/em\u003e, \u003cem\u003ePDZD2\u003c/em\u003e, and \u003cem\u003eUGT3A2\u003c/em\u003e exhibited significantly reduced sensitivity (higher AUC values) to AZD4547 (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eA). Subsequent validation in the PRISM dataset corroborated this association, demonstrating a significant enrichment of these mutant genotypes specifically within the resistant subgroup (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eB). Given the known roles of ARHGAP33 in Rho signaling and UGT3A2 in drug metabolism, these alterations may suggest potential mechanisms involving cytoskeletal dynamics and metabolic clearance, warranting further validation.\u003c/p\u003e \u003cp\u003eAcross other cancer lineages, resistant phenotypes were characterized by distinct genomic signatures, such as \u003cem\u003eTENM4\u003c/em\u003e mutations in LUAD and \u003cem\u003eAPC2\u003c/em\u003e mutations in ESCC, which were similarly associated with diminished drug response (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eA, B). Notably, as APC2 is a component of the Wnt destruction complex, its mutational enrichment implicates aberrant Wnt signaling as a potential feature of the resistant phenotype.\u003c/p\u003e \u003cp\u003eBeyond point mutations, epigenetic and structural analyses revealed multi-layered heterogeneity. In BRCA, resistance was associated with higher methylation levels of homeobox genes like \u003cem\u003eEN1\u003c/em\u003e and \u003cem\u003eHOXC6\u003c/em\u003e (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eC, D), suggesting the epigenetic silencing of these developmental regulators. Conversely, the resistant landscape of GBM was marked by recurrent copy number amplifications. We identified 403 resistance-correlated CNVs in GBM (Table \u003cspan refid=\"MOESM3\" class=\"InternalRef\"\u003eS3\u003c/span\u003e), including the amplification of \u003cem\u003eELF4\u003c/em\u003e and \u003cem\u003eARHGEF6\u003c/em\u003e (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eE). Previous studies have linked ELF4 to stemness maintenance in GBM, and ARHGEF6 to tumor progression in various cancers[\u003cspan additionalcitationids=\"CR24 CR25\" citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e], suggesting that the amplification-driven upregulation of these oncogenic drivers may support the aggressive resistant state.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec21\" class=\"Section2\"\u003e \u003ch2\u003eTranscriptomic and proteomic signatures of FGFRi response\u003c/h2\u003e \u003cp\u003eTo identify high-confidence molecular signatures associated with FGFRi response, we performed an integrated analysis of pharmacogenomic data from the GDSC2 and PRISM cohorts. We specifically screened for candidate genes that demonstrated consistent drug response associations between the two datasets within each specific cancer lineage, thereby yielding a robust panel of biomarkers (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eA, Figure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003eA, B; Table S4).\u003c/p\u003e \u003cp\u003eTo further assess their clinical relevance, we mapped these candidate genes to patient data from the TCGA and CPTAC cohorts (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eB). This comprehensive landscape revealed that many identified markers exhibited tumor-specific differential expression and significant prognostic value. Specifically, a subset of resistance-associated transcripts\u0026mdash;highlighted by \u003cem\u003eUCP2\u003c/em\u003e, \u003cem\u003eCOL11A1\u003c/em\u003e, \u003cem\u003ePRAME\u003c/em\u003e, and \u003cem\u003eUNC5A\u003c/em\u003e\u0026mdash;demonstrated a striking multi-dimensional consistency. These genes were not only correlated with AZD4547 resistance but were also identified as prognostic risk factors in patient cohorts. Notably, markers such as \u003cem\u003eCOL11A1\u003c/em\u003e and \u003cem\u003ePRAME\u003c/em\u003e were also significantly upregulated in tumor tissues compared to normal controls. Conversely, we identified a panel of sensitivity-associated markers, such as \u003cem\u003eATOH8\u003c/em\u003e, \u003cem\u003eATP8B1\u003c/em\u003e, \u003cem\u003eLIMCH1\u003c/em\u003e, and \u003cem\u003eNTN4\u003c/em\u003e. These genes formed a distinct cluster characterized by negative drug response correlations and favorable patient prognosis.\u003c/p\u003e \u003cp\u003eComplementing the transcriptomic landscape, we further leveraged proteomic profiling to capture functional alterations. Notably, the identified proteomic signatures showed minimal overlap with the transcriptomic markers (Figs.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eC, D, Figure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003eC), indicating the prevalence of post-transcriptional regulation in driving resistance. This analysis highlighted distinct protein-level markers, such as TM9SF4 and ZMPSTE24, which were consistently upregulated in resistant phenotypes across both datasets (Figs.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eC, D; Table S5). Clinical validation in the CPTAC dataset confirmed these candidates were significantly overexpressed in tumor tissues (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eE). Crucially, our proteomic screen also highlighted ADAM17 as a robust resistance marker. Given its established role in shedding EGFR ligands, the upregulation of ADAM17 provides proteomic evidence corroborating the \"RTK switching\" mechanism toward EGFR signaling, which we further validate in downstream analyses.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec22\" class=\"Section2\"\u003e \u003ch2\u003eLineage-specific and convergent functional landscapes of FGFRi resistance\u003c/h2\u003e \u003cp\u003eTo transition from discrete molecular markers to systems-level resistance mechanisms, we performed GO and Reactome functional enrichment analyses on the mRNA and protein signatures identified from our integrative screen. This initial profiling revealed that the molecular landscape of resistance is characterized by lineage-dependent adaptive programs rather than a uniform pan-cancer mechanism (Figure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003eD-G). For instance, transcriptomic signatures in COAD were predominantly associated with metabolic pathways, such as 'Respiratory electron transport' (Figure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003eE), whereas ESCC and BLCA exhibited proteomic enrichment in structural modules like 'Assembly of collagen fibrils' (Figure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003eG).\u003c/p\u003e \u003cp\u003eTo ensure the reliability of these lineage-specific findings and mitigate potential platform- or omics-level discordance, we performed GSEA across four independent discovery layers: GDSC2-mRNA/protein and PRISM-mRNA/protein. To prioritize high-confidence biological modules, we implemented a stringent consensus criterion, requiring significant enrichment in at least two data layers with consistent functional polarity. This integrative approach identified six convergent functional modules associated with FGFRi resistance: metabolic reprogramming, cell-cycle dysregulation, estrogen response, immunity \u0026amp; inflammation, signaling transduction, and invasion \u0026amp; migration (Figs.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eA; Table S6).\u003c/p\u003e \u003cp\u003eFurther analysis revealed that these resistance programs are highly context-dependent, manifesting as distinct molecular signatures across specific cancer lineages. In COAD, the resistant phenotype displayed a multidimensional functional profile encompassing metabolic reprogramming (e.g., adipogenesis), signaling transduction (e.g., mTORC1 signaling), and immunity \u0026amp; inflammation (e.g., interferon-gamma response) (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eA, Figure \u003cspan refid=\"MOESM2\" class=\"InternalRef\"\u003eS2\u003c/span\u003eA). Within these pathways, ATP1B3 and UCP2 were prioritized as high-confidence regulatory components representing the mTORC1 signaling and adipogenesis programs, respectively. These markers exhibited concordant associations with resistance, characterized by positive drug-response correlations in GDSC2 and significantly elevated expression in the PRISM resistant cohort (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eB). Furthermore, UCP2 was validated as a significant prognostic risk factor in both CPTAC and TCGA datasets (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eC), aligning with its documented role in modulating mitochondrial bioenergetics and reinforcing the clinical relevance of mitochondrial metabolic adaptation[\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e, \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e]. Collectively, these findings suggest that resistant cells in COAD may exploit metabolic plasticity and signaling rewiring to elevate the threshold for drug-induced cell death.\u003c/p\u003e \u003cp\u003eConversely, the resistance landscape in BLCA was marked by an invasion \u0026amp; migration signature, primarily involving Epithelial-Mesenchymal Transition (EMT) and apical junctions (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eA, Figure \u003cspan refid=\"MOESM2\" class=\"InternalRef\"\u003eS2\u003c/span\u003eB). The actin-bundling protein FSCN1 was identified as a core component of this phenotype; its expression was positively correlated with FGFRi resistance and was identified as a significant prognostic risk factor in bladder cancer patients (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eD, E). The prominence of FSCN1 suggests that resistant cells may leverage cytoskeletal remodeling to enhance cellular plasticity and invasive potential, thereby facilitating therapeutic evasion. Additionally, the resistant phenotype in BRCA was significantly linked to estrogen response pathways and elevated MLPH expression (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eA, Figure \u003cspan refid=\"MOESM2\" class=\"InternalRef\"\u003eS2\u003c/span\u003eC, D).\u003c/p\u003e \u003cp\u003eDespite these lineage-specific variations, a convergent reliance on cell-cycle progression was identified as a shared hallmark of resistance in PAAD and GBM. These resistant phenotypes were significantly enriched for MYC targets, E2F targets, and G2M checkpoint programs (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eA, Figures \u003cspan refid=\"MOESM2\" class=\"InternalRef\"\u003eS2\u003c/span\u003eE, F). In PAAD, MCM7 expression was positively correlated with FGFRi resistance and identified as a significant prognostic risk factor (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eF, G). Similarly, the resistant state in GBM was characterized by the upregulation of genes involved in cell-cycle progression and nucleotide metabolism, most notably MCM4 and HPRT1 (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eH, Figures \u003cspan refid=\"MOESM2\" class=\"InternalRef\"\u003eS2\u003c/span\u003eG, H). Both markers exhibited striking associations with unfavorable outcomes in CPTAC and TCGA-GBM cohorts (Figs.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eI, J). Given the fundamental role of the Minichromosome Maintenance (MCM) family in DNA replication initiation[\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e], these findings point toward a link between FGFRi resistance and a hyper-proliferative state in PAAD and GBM, likely supported by the coordinated activation of cell-cycle and metabolic programs.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cdiv id=\"Sec23\" class=\"Section3\"\u003e \u003ch2\u003eDistinct multi-omics regulatory landscapes of FGFRi resistance\u003c/h2\u003e \u003cp\u003eTo elucidate how genomic aberrations propagate across molecular hierarchies to shape the FGFRi resistance phenotype, we systematically analyzed the correlations between high-confidence somatic alterations (mutations, CNVs) and epigenetic features (DNA methylation) with downstream transcriptome and proteome profiles. This analysis revealed highly heterogeneous regulatory patterns, where distinct modes of signaling rewiring explain the lineage-specific phenotypes observed earlier.\u003c/p\u003e \u003cp\u003eIn GBM, we uncovered an extensive regulatory network integrating both CNV effects and transcription factor (TF) associations. The analysis detected 410 CNV-mRNA and 1,677 CNV-protein regulatory pairs (Table S7), underscoring the pervasiveness of CNV impact in GBM. Beyond this genomic baseline, the network was further orchestrated by specialized transcriptional programs. Notably, the TFs CTCF, HDAC1, and RBMX emerged as central regulatory hubs. Among them, RBMX exhibited strong positive correlations with both prognostic risk factors MCM4 and HPRT1, whereas CTCF and HDAC1 were primarily linked to the regulation of MCM4 (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eA). This hierarchical flow suggests that CNV-TF coordination may sustain the hyper-proliferative state and nucleotide metabolic reprogramming associated with the FGFRi-resistant phenotype in GBM.\u003c/p\u003e \u003cp\u003eIn contrast, other cancer cohorts exhibited more focused regulatory profiles. In BLCA, we observed a discrete multi-dimensional landscape involving 14 CNV-mRNA, 9 CNV-protein, and 8 TF-mRNA associations. A notable genomic-transcriptomic link was the significant negative correlation between \u003cem\u003eCHEK2P2\u003c/em\u003e copy number and the expression of the resistance marker NT5E (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eB). Furthermore, we identified the transcription factor EHMT2 as a potential regulator of the actin-bundling protein FSCN1 (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eB). This EHMT2-FSCN1 axis may provide a mechanistic basis for the EMT and pro-invasive phenotype previously linked to FGFRi resistance in bladder cancer patients. Meanwhile, BRCA resistance was uniquely characterized by epigenetic modulation. Specifically, the expression of the resistance-associated marker \u003cem\u003eMLPH\u003c/em\u003e exhibited significant positive correlations with the methylation levels of \u003cem\u003eHAAO\u003c/em\u003e and \u003cem\u003eEN1\u003c/em\u003e (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eC). In PAAD, a convergent regulatory mode was identified, featuring 10 TF-mRNA associations. Notably, the transcriptional co-activator TRRAP emerged as a pivotal regulator, significantly orchestrating the mRNA expression of \u003cem\u003eABL1\u003c/em\u003e and MCM7 (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eD).\u003c/p\u003e \u003cp\u003eFinally, these core regulatory molecules were synthesized into a unified multi-omics interaction network (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eE), providing a holistic view of the coordinated rewiring of genomic, epigenetic, and transcriptional landscapes in FGFRi resistance.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec24\" class=\"Section2\"\u003e \u003ch2\u003eSingle-cell Profiling Resolves Clonal Selection and Conserved Metaprograms Underlying FGFRi Resistance\u003c/h2\u003e \u003cp\u003eTo dissect the cellular-level heterogeneity and evolutionary dynamics governing FGFRi resistance, we transitioned from bulk-level profiling to scRNA-seq analysis across a pan-cancer atlas. Following a stringent quality control pipeline, a high-resolution landscape of 9,997 individual cells was established, encompassing diverse lineages and distinct drug-response phenotypes (Figs.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eA-C). To identify conserved biological states across these heterogeneous cell lines, we utilized NMF to deconvolve the transcriptomic data into robust gene metaprograms (MPs). This analysis derived four recurrent MPs representing distinct oncogenic processes: tumor immunity (MP1), cell cycle (MP2), DNA damage repair (MP3), and telomerase activity (MP4) (Figs.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eD, E).\u003c/p\u003e \u003cp\u003eNotably, the activity landscapes of these MPs exhibited lineage-dependent characteristics. Specifically, MP2 was activated in MEL and GBM cells, while MP3 showed prominent activation in MEL and BRCA lineages (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eF), defining their baseline transcriptomic features. Crucially, the statistical enrichment of these MPs in resistant phenotypes was highly context-dependent. MP2 was consistently enriched in the resistant subcohorts of PAAD and MEL across two independent scoring algorithms, whereas MP3 activity was over-represented in resistant populations of MEL and GBM (Figs.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eG, H). The enrichment of MP2\u0026mdash;characterized by core replication drivers such as CDK1 and BUB1\u0026mdash;within resistant PAAD and MEL subcohorts suggests that FGFRi resistance is predominantly underpinned by the clonal selection of pre-existing subpopulations harboring intrinsic cell-cycle dysregulation. Furthermore, the upregulation of MP3 in resistant MEL and GBM indicates that enhanced genomic maintenance serves as a critical survival adaptation under therapeutic stress. Together, these findings demonstrate that FGFRi resistance arises from the clonal selection of pre-existing cell states rather than uniform transcriptomic adaptation, and diverse lineages employ distinct strategies\u0026mdash;from hyper-proliferation to fortified DNA repair\u0026mdash;to collectively elevate the cellular survival threshold against FGFR inhibition.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cdiv id=\"Sec25\" class=\"Section3\"\u003e \u003ch2\u003eDevelopment of a Biologically Grounded Machine Learning model for Predicting FGFRi Response\u003c/h2\u003e \u003cp\u003eTo translate our discovery into clinical utility, we developed a robust transcriptome-based predictive model to stratify FGFRi sensitivity. The PRISM discovery cohort was randomly partitioned into a training set (70%), a validation set (20%), and an independent test set (10%) to ensure a rigorous and unbiased evaluation of model performance (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eA). Utilizing LASSO regression, we distilled the high-dimensional transcriptional landscape into a parsimonious 52-gene signature (Figs.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eB, C). We then benchmarked seven mainstream machine learning paradigms to select the optimal algorithm. The Random Forest ensemble model was identified as the superior classifier, achieving top-performing discriminative power with the highest AUC in both the internal validation set (AUC\u0026thinsp;=\u0026thinsp;0.782) and the independent test set (AUC\u0026thinsp;=\u0026thinsp;0.714) (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eD, E). Beyond its remarkable AUC, the model consistently excelled across a comprehensive suite of evaluation metrics\u0026mdash;including sensitivity, precision, recall, and accuracy\u0026mdash;within both the validation and test cohorts (Figs.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eF, G). The generalizability of this framework was further corroborated via external validation in the GDSC2 database, where cell lines predicted as resistant by the model exhibited significantly higher drug response (AUC) values than those predicted as sensitive (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.021) (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eH). This multi-dimensional superiority underscores the predictive robustness and translational potential of the 52-gene signature in identifying FGFRi-responsive phenotypes across diverse malignancies.\u003c/p\u003e \u003cp\u003eTo deconstruct the algorithmic interpretability of the Random Forest model and ensure its biological fidelity, we employed SHAP to quantify the contribution of individual features to the predictive output (Figs.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eI, J). The prioritization of top-ranking features, including \u003cem\u003eUSP18\u003c/em\u003e, \u003cem\u003eHIP1R\u003c/em\u003e, and \u003cem\u003eRASGRF2\u003c/em\u003e, underscores the model\u0026rsquo;s alignment with established oncogenic signaling dependencies. Specifically, USP18 has been documented to fuel tumor growth and proliferation by activating the AKT signaling pathway[\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e, \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e], while while HIP1R modulates cellular survival and invasive potential primarily via the PI3K/AKT axis[\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e, \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e], and RASGRF2 orchestrates both Src/PI3K and NF-κB cascades to promote metastatic phenotypes[\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e]. Notably, the convergence of these top-ranked molecules on the PI3K/AKT axis\u0026mdash;a pathway we subsequently characterize as a core synthetic lethal vulnerability (see Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003e)\u0026mdash;suggests that our framework accurately captures the fundamental signaling rewiring that facilitates therapeutic evasion.\u003c/p\u003e \u003cp\u003eFurthermore, the inclusion of \u003cem\u003eIFNA13\u003c/em\u003e, a mediator of interferon signaling, and GALT, a key enzyme involved in galactose metabolism, within the top feature set reflects the model's capacity to integrate inflammatory response and metabolic reprogramming. These features represent two convergent adaptive strategies that we previously identified through GSEA and proteogenomic profiling as hallmarks of the FGFRi-resistant phenotype. This mechanistic consistency emphasizes that the 52-gene signature is not merely a collection of statistical correlates but a biologically grounded representation of the systemic network rewiring underlying pan-cancer FGFRi resistance.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec26\" class=\"Section3\"\u003e \u003ch2\u003eIdentification of Synthetic Lethal Vulnerabilities Linked to the FGF/FGFR Axis\u003c/h2\u003e \u003cp\u003eTo uncover potential therapeutic targets that synergize with FGFR inhibition, we implemented a systematic computational framework to screen for synthetic lethal (SL) partners of the FGF/FGFR axis (comprising 4 FGFRs and 22 FGF ligands) across eight cancer types. By aggregating the significant pairs identified independently across the cancer types, we constructed a comprehensive pan-cancer SL landscape for the FGF/FGFR family (Table S8). For instance, in ESCC, we identified 25 SL pairs, while 39 pairs were identified in PAAD.\u003c/p\u003e \u003cp\u003eTo elucidate the biological implications of these SL interactions, we performed an integrated pathway enrichment analysis combining both GO and KEGG databases. This multi-dimensional annotation revealed a convergent functional landscape focused on five core biological modules: signaling transduction, metabolic reprogramming, cell cycle and proliferation, invasion and migration, and immunity (Figs.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003eA, B).\u003c/p\u003e \u003cp\u003eFirst, Signal Transduction Reactivation. Both databases consistently highlighted a heavy reliance on canonical survival cascades, specifically \"Positive regulation of MAPK/ERK cascade\" (GO) and \"PI3K-Akt signaling pathway\" (KEGG). These findings resonate with our earlier multi-omics GSEA results (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e), where the KRAS and mTORC1 signaling pathways were significantly enriched in resistant phenotypes. This convergence suggests that when the FGFR axis is compromised, tumor cells strategically rewire their signaling networks to exploit these parallel hubs.\u003c/p\u003e \u003cp\u003eSecond, Cell Cycle Dysregulation. Terms related to \"Positive regulation of cell cycle\" and \"G1/S transition\" were prominent, which aligns seamlessly with our single-cell analysis that pinpointed the aberrant activation of the cell-cycle metaprogram (MP2) in resistant subpopulations (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003e). This indicates that targeting SL partners involved in cell cycle checkpoints (e.g., CDKs) could specifically eliminate the hyper-proliferative resistant clones.\u003c/p\u003e \u003cp\u003eThird, RTK Switching. Most notably, we observed a significant enrichment of the 'EGFR tyrosine kinase inhibitor resistance' pathway (KEGG). This finding provides robust computational validation for the \"RTK switching\" hypothesis introduced earlier, confirming that FGFR inhibition triggers a compensatory upregulation of alternative RTKs, such as EGFR, to sustain downstream signaling. Consequently, dual blockade of FGFR and these identified SL targets represents a rational synthetic lethal strategy to overcome resistance.\u003c/p\u003e \u003cp\u003eConsequently, dual blockade of FGFR and these identified SL targets (e.g., EGFR or MEK/ERK pathway components) could represent a promising synthetic lethal strategy to overcome resistance.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eDespite the initial clinical efficacy of FGFRi such as AZD4547, the inevitable emergence of acquired resistance remains a major bottleneck in precision oncology[\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]. Current research has largely been confined to isolated resistance mechanisms within specific cancer types, failing to capture the systemic network rewiring that drives therapeutic failure. In this study, we bridged this gap by constructing a comprehensive pan-cancer proteogenomic atlas of FGFRi resistance. By integrating data from 312 cell lines across the GDSC2 and PRISM databases, we deciphered the multi-layered heterogeneity of resistance and identified actionable synthetic lethal vulnerabilities that pave the way for next-generation combination therapies.\u003c/p\u003e \u003cp\u003eA key finding of our study is the context-dependent reliance of resistant cells on metabolic reprogramming. In colorectal cancer (COAD), we observed a striking activation of the adipogenesis pathway and upregulation of \u003cem\u003eUCP2\u003c/em\u003e. UCP2 is known to reduce mitochondrial oxidative stress and inhibit apoptosis, thereby raising the threshold for drug-induced cell death[\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e, \u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e]. Consistent with this, our synthetic lethality screening further highlighted \"Choline metabolism\" and \"Lipid and atherosclerosis\" pathways as critical vulnerabilities. This suggests that under FGFRi blockade, tumor cells may undergo a metabolic switch\u0026mdash;shifting from glycolysis to fatty acid oxidation or altering membrane lipid composition\u0026mdash;to sustain survival signals. This finding aligns with the \"warburg effect\" plasticity and suggests that combining FGFRi with metabolic inhibitors (e.g., inhibitors of fatty acid synthesis) could be a viable strategy for specific tumor subtypes[\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e, \u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e]. Since UCP2-mediated metabolic adaptation is often tumor-specific and less prominent in normal tissues [\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e, \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e], targeting this vulnerability offers a potential therapeutic window with reduced toxicity.\u003c/p\u003e \u003cp\u003eBeyond bulk omics, our single-cell analysis provided high-resolution insights into the evolutionary dynamics of resistance. We identified a specific cell-cycle metaprogram (MP2) that is aberrantly activated in resistant subpopulations across melanoma and glioblastoma. This observation is consistent with a model of clonal selection rather than purely acquired plasticity: FGFR inhibition likely eliminates sensitive clones while sparing pre-existing subpopulations with intrinsic cell-cycle dysregulation. The enrichment of MCM family genes (e.g., \u003cem\u003eMCM7\u003c/em\u003e, \u003cem\u003eMCM4\u003c/em\u003e) in resistant phenotypes further corroborates this. However, this interpretation requires validation through longitudinal tracking or functional experiments.\u003c/p\u003e \u003cp\u003ePerhaps the most translationally significant discovery of our study is the support of the \"RTK switching\" hypothesis. Our GSEA analysis initially revealed that resistant cells maintain high activity of downstream KRAS and mTORC1 signaling despite FGFR blockade, hinting at the activation of alternative upstream drivers. This hypothesis was supported by our synthetic lethality screening, which identified the \"EGFR tyrosine kinase inhibitor resistance\" pathway as a top-ranked vulnerability. This indicates that upon FGFR inhibition, tumor cells dynamically rewire their signaling networks to exploit EGFR as a compensatory bypass track to reactivate the MAPK/PI3K cascade. Furthermore, our functional enrichment revealed potential engagement of autophagy and PD-L1 checkpoint pathways, suggesting that this network rewiring extends beyond kinases to include survival-mediated autophagy and immune evasion. These results provide a compelling rationale for dual-blockade strategies (e.g., FGFRi\u0026thinsp;+\u0026thinsp;EGFRi or FGFRi\u0026thinsp;+\u0026thinsp;anti-PD-1) to dismantle these compensatory survival loops.\u003c/p\u003e \u003cp\u003eTranslating these molecular findings into clinical utility, we developed a machine learning model based on 52 transcriptomic features. Unlike static biomarkers (e.g., genomic mutations), our model integrates dynamic mRNA signatures to accurately predict the probability of FGFRi response. This tool holds significant potential for clinical utility, serving as a robust stratification tool to distinguish patients likely to benefit from AZD4547 monotherapy from those requiring immediate upfront combination strategies. We propose a tiered precision medicine framework: Specifically, patients predicted to be 'sensitive' could proceed with monotherapy, thereby sparing them the added toxicity of combination regimens. Conversely, those predicted to be 'resistant' would be flagged as high-priority candidates for the combination therapies we identified (e.g., FGFRi plus EGFR or metabolic inhibitors). For this resistant subgroup, subsequent profiling of specific markers (e.g., \u003cem\u003eUCP2\u003c/em\u003e levels or EGFR activation) would guide the precise selection of the secondary agent. Mechanistically, the predictive power of our model is biologically grounded. The top features such as USP18 and HIP1R are established regulators of the PI3K/AKT axis[\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e, \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e]. This aligns seamlessly with our synthetic lethality findings, where the PI3K/AKT pathway emerged as a convergent survival bypass downstream of EGFR, further validating the biological interpretability of our computational approach.\u003c/p\u003e \u003cp\u003eOur study has limitations inherent to cell line-based analyses. First, \u003cem\u003ein vitro\u003c/em\u003e models lack the complex tumor microenvironment (TME), which may influence the translatability of immune-related findings, such as PD-L1 associations. Second, the technical limitations of scRNA-seq in capturing transient metabolic states. The cell dissociation may lead to the loss of certain mRNA transcripts or alter the expression of genes sensitive to cellular stress, which could potentially impact the detection of subtle metabolic reprogramming signatures. To address this, future studies incorporating spatial metabolomics or integrated functional experiments are warranted to validate the metabolic reprogramming patterns identified in this study. Third, the unequal sample sizes across cancer types may introduce bias. Fortunately, our cross-platform validation across bulk and clinical cohorts ensures the robustness of the core findings. Forth, this study focused on a single FGFR inhibitor (AZD4547), and the generalizability to other FGFR inhibitors requires validation. Finally, computational synthetic lethal candidates require experimental validation in patient-derived organoids (PDOs) or xenograft (PDX) models to confirm their clinical efficacy.\u003c/p\u003e"},{"header":"Conclusions","content":"\u003cp\u003eIn summary, this study presents a first-of-its-kind pan-cancer atlas of FGFRi resistance. By integrating multi-omics profiling with synthetic lethality screening, we unveiled a paradigm where resistance is driven by a dynamic interplay of RTK switching (EGFR), metabolic rewiring (UCP2), and clonal selection of cell-cycle variants. These insights not only clarify the heterogeneous nature of resistance but also offer a concrete, actionable framework for the development of precise, mechanism-based combination therapies to improve patient outcomes.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cdiv class=\"DefinitionList\"\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eACC\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eAccuracy\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eAUC\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eArea Under the Curve\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eBLCA\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eBladder Urothelial Carcinoma\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eBRCA\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eBreast Invasive Carcinoma\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eCNV\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eCopy Number Variation\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eCOAD\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eColon Adenocarcinoma\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eDT\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eDecision Tree\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eEMT\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eEpithelial-Mesenchymal Transition\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eESCC\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eEsophageal Squamous Cell Carcinoma\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eFC\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eFold Change\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eFDR\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eFalse Discovery Rate\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eFGF\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eFibroblast Growth Factor\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eFGFR\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eFibroblast Growth Factor Receptor\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eFGFRi\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eFibroblast Growth Factor Receptor Inhibitor\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eGBM\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eGlioblastoma\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003ekNN\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003ek-Nearest Neighbor\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eLASSO\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eLeast Absolute Shrinkage and Selection Operator\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eLFC\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eLog-Fold Change\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eLogReg\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eLogistic Regression\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eLUAD\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eLung Adenocarcinoma\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eMEL\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eMelanoma\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eMLP\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eMultilayer Perceptron\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eMP\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eMetaprogram\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eNES\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eNormalized Enrichment Score\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eNMF\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eNon-negative Matrix Factorization\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eNPV\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eNegative Predictive Value\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003ePAAD\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003ePancreatic Adenocarcinoma\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003ePCA\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003ePrincipal Component Analysis\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003ePDO\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003ePatient-Derived Organoid\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003ePDX\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003ePatient-Derived Xenograft\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003ePPV\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003ePositive Predictive Value\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eRF\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eRandom Forest\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eROC\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eReceiver Operating Characteristic\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eRTK\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eReceptor Tyrosine Kinase\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eSHAP\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eShapley Additive Explanations\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eSL\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eSynthetic Lethal\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eSNV\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eSingle Nucleotide Variant\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eTF\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eTranscription Factor\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eTKI\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eTyrosine Kinase Inhibitor\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eTME\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eTumor Microenvironment\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eTPM\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eTranscripts Per Million\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003et-SNE\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003et-Distributed Stochastic Neighbor Embedding\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eWT\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eWild-Type\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eXGBoost\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eExtreme Gradient Boosting\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003c/div\u003e"},{"header":"Declarations","content":"\u003cp\u003e \u003cstrong\u003eEthics approval and consent to participate\u003c/strong\u003e \u003cp\u003eNot applicable.\u003c/p\u003e \u003c/p\u003e \u003cp\u003e \u003cstrong\u003eConsent for publication\u003c/strong\u003e \u003cp\u003eNot applicable.\u003c/p\u003e \u003c/p\u003e\u003cp\u003e \u003ch2\u003eCompeting interests\u003c/h2\u003e \u003cp\u003eThe authors declare that they have no competing interests.\u003c/p\u003e \u003c/p\u003e\u003ch2\u003eFunding\u003c/h2\u003e \u003cp\u003eNot applicable.\u003c/p\u003e\u003ch2\u003eAuthors' contributions\u003c/h2\u003e \u003cp\u003eLT conceived the study and performed the bioinformatic analysis. TH and PY assisted with data collection and visualization. XW, HJ and JL supervised the study, provided resources, and revised the manuscript. All authors read and approved the final manuscript.\u003c/p\u003e\u003ch2\u003eAcknowledgements\u003c/h2\u003e \u003cp\u003eWe thank the authors of the GDSC, PRISM, TCGA, CPTAC, and GEO datasets for making their data publicly available. We also acknowledge the BioEnricher and Seurat development teams for their open-source tools. We would like to thank our friend Jinyang Liu for providing valuable guidance and support during the data analysis process.\u003c/p\u003e\u003ch2\u003eAvailability of data and materials\u003c/h2\u003e \u003cp\u003eThe datasets analyzed in the study are available in the following public repositories:\u003c/p\u003e \u003cp\u003eDepMap: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://depmap.org/portal/\u003c/span\u003e\u003cspan address=\"https://depmap.org/portal/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e \u003cp\u003eGDSC2: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.cancerrxgene.org/\u003c/span\u003e\u003cspan address=\"https://www.cancerrxgene.org/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e \u003cp\u003ePRISM: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://depmap.org/portal/prism/\u003c/span\u003e\u003cspan address=\"https://depmap.org/portal/prism/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e \u003cp\u003eTCGA: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://gdc.cancer.gov/\u003c/span\u003e\u003cspan address=\"https://gdc.cancer.gov/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e \u003cp\u003eCPTAC: The data was derived from the publication by Li et al. [\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e], available at \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubmed.ncbi.nlm.nih.gov/37582357/\u003c/span\u003e\u003cspan address=\"https://pubmed.ncbi.nlm.nih.gov/37582357/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e or via the CPTAC data portal (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pdc.cancer.gov/\u003c/span\u003e\u003cspan address=\"https://pdc.cancer.gov/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eGEO: Accession number GSE157220 (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE157220\u003c/span\u003e\u003cspan address=\"https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE157220\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThe code used for the machine learning model and analysis is available from the corresponding author upon reasonable request.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eXie Y, Su N, Yang J, Tan Q, Huang S, Jin M, et al. FGF/FGFR signaling in health and disease. Signal Transduct Target Ther. 2020;5:181. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/s41392-020-00222-7\u003c/span\u003e\u003cspan address=\"10.1038/s41392-020-00222-7\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKrook MA, Reeser JW, Ernst G, Barker H, Wilberding M, Li G, et al. Fibroblast growth factor receptors in cancer: genetic alterations, diagnostics, therapeutic targets and mechanisms of resistance. Br J Cancer. 2021;124:880\u0026ndash;92. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/s41416-020-01157-0\u003c/span\u003e\u003cspan address=\"10.1038/s41416-020-01157-0\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTernier G, Shahzad K, Edirisinghe O, Okoto P, Alraawi Z, Sonnaila S, et al. Fibroblast Growth Factors: Roles and Emerging Therapeutic Applications. Curr Drug Targets. 2025;26:551\u0026ndash;70. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.2174/0113894501351461250301072444\u003c/span\u003e\u003cspan address=\"10.2174/0113894501351461250301072444\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHung KL, Luebeck J, Dehkordi SR, Col\u0026oacute;n CI, Li R, Wong IT-L, et al. Targeted profiling of human extrachromosomal DNA by CRISPR-CATCH. Nat Genet. 2022;54:1746\u0026ndash;54. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/s41588-022-01190-0\u003c/span\u003e\u003cspan address=\"10.1038/s41588-022-01190-0\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePaik PK, Shen R, Berger MF, Ferry D, Soria J-C, Mathewson A, et al. A Phase Ib Open-Label Multicenter Study of AZD4547 in Patients with Advanced Squamous Cell Lung Cancers. Clin Cancer Res. 2017;23:5366\u0026ndash;73. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1158/1078-0432.CCR-17-0645\u003c/span\u003e\u003cspan address=\"10.1158/1078-0432.CCR-17-0645\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAggarwal C, Redman MW, Lara PN, Borghaei H, Hoffman P, Bradley JD, et al. SWOG S1400D (NCT02965378), a Phase II Study of the Fibroblast Growth Factor Receptor Inhibitor AZD4547 in Previously Treated Patients With Fibroblast Growth Factor Pathway-Activated Stage IV Squamous Cell Lung Cancer (Lung-MAP Substudy). J Thorac Oncol. 2019;14:1847\u0026ndash;52. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.jtho.2019.05.041\u003c/span\u003e\u003cspan address=\"10.1016/j.jtho.2019.05.041\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGonzalez-Ericsson PI, Unni N, Jhaveri K, Stringer-Reasor E, Liu Q, Wang Y, et al. Phase Ib Trial of Fulvestrant, Palbociclib, and Erdafitinib, a pan-FGFR Tyrosine Kinase Inhibitor, in HR+/HER2- Metastatic Breast Cancer. Clin Cancer Res. 2025;31:3652\u0026ndash;61. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1158/1078-0432.CCR-24-3803\u003c/span\u003e\u003cspan address=\"10.1158/1078-0432.CCR-24-3803\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFranza A, Pirovano M, Giannatempo P, Cosmai L. Erdafitinib in locally advanced/metastatic urothelial carcinoma with certain FGFR genetic alterations. Future Oncol. 2022;18:2455\u0026ndash;64. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.2217/fon-2021-1151\u003c/span\u003e\u003cspan address=\"10.2217/fon-2021-1151\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePeng M, Deng J, Li X. Clinical advances and challenges in targeting FGF/FGFR signaling in lung cancer. Mol Cancer. 2024;23:256. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1186/s12943-024-02167-9\u003c/span\u003e\u003cspan address=\"10.1186/s12943-024-02167-9\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePowles T, Carroll D, Chowdhury S, Gravis G, Joly F, Carles J, et al. An adaptive, biomarker-directed platform study of durvalumab in combination with targeted therapies in advanced urothelial cancer. Nat Med. 2021;27:793\u0026ndash;801. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/s41591-021-01317-6\u003c/span\u003e\u003cspan address=\"10.1038/s41591-021-01317-6\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSaborowski A, Lehmann U, Vogel A. FGFR inhibitors in cholangiocarcinoma: what\u0026rsquo;s now and what\u0026rsquo;s next? Ther Adv Med Oncol. 2020;12:1758835920953293. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1177/1758835920953293\u003c/span\u003e\u003cspan address=\"10.1177/1758835920953293\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKatoh M, Loriot Y, Brandi G, Tavolari S, Wainberg ZA, Katoh M. FGFR-targeted therapeutics: clinical activity, mechanisms of resistance and new directions. Nat Rev Clin Oncol. 2024;21:312\u0026ndash;29. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/s41571-024-00869-z\u003c/span\u003e\u003cspan address=\"10.1038/s41571-024-00869-z\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGoyal L, Saha SK, Liu LY, Siravegna G, Leshchiner I, Ahronian LG, et al. Polyclonal Secondary FGFR2 Mutations Drive Acquired Resistance to FGFR Inhibition in Patients with FGFR2 Fusion-Positive Cholangiocarcinoma. Cancer Discov. 2017;7:252\u0026ndash;63. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1158/2159-8290.CD-16-1000\u003c/span\u003e\u003cspan address=\"10.1158/2159-8290.CD-16-1000\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRyan MR, Sohl CD, Luo B, Anderson KS. The FGFR1 V561M Gatekeeper Mutation Drives AZD4547 Resistance through STAT3 Activation and EMT. Mol Cancer Res. 2019;17:532\u0026ndash;43. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1158/1541-7786.MCR-18-0429\u003c/span\u003e\u003cspan address=\"10.1158/1541-7786.MCR-18-0429\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWu Q, Zhen Y, Shi L, Vu P, Greninger P, Adil R, et al. EGFR Inhibition Potentiates FGFR Inhibitor Therapy and Overcomes Resistance in FGFR2 Fusion-Positive Cholangiocarcinoma. Cancer Discov. 2022;12:1378\u0026ndash;95. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1158/2159-8290.CD-21-1168\u003c/span\u003e\u003cspan address=\"10.1158/2159-8290.CD-21-1168\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eQuintanal-Villalonga A, Molina-Pinelo S, Cirauqui C, Ojeda-M\u0026aacute;rquez L, Marrugal \u0026Aacute;, Suarez R, et al. FGFR1 Cooperates with EGFR in Lung Cancer Oncogenesis, and Their Combined Inhibition Shows Improved Efficacy. J Thorac Oncol. 2019;14:641\u0026ndash;55. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.jtho.2018.12.021\u003c/span\u003e\u003cspan address=\"10.1016/j.jtho.2018.12.021\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDiPeri TP, Zhao M, Evans KW, Varadarajan K, Moss T, Scott S, et al. Convergent MAPK pathway alterations mediate acquired resistance to FGFR inhibitors in FGFR2 fusion-positive cholangiocarcinoma. J Hepatol. 2024;80:322\u0026ndash;34. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.jhep.2023.10.041\u003c/span\u003e\u003cspan address=\"10.1016/j.jhep.2023.10.041\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSilverman IM, Hollebecque A, Friboulet L, Owens S, Newton RC, Zhen H, et al. Clinicogenomic Analysis of FGFR2-Rearranged Cholangiocarcinoma Identifies Correlates of Response and Mechanisms of Resistance to Pemigatinib. Cancer Discov. 2021;11:326\u0026ndash;39. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1158/2159-8290.CD-20-0766\u003c/span\u003e\u003cspan address=\"10.1158/2159-8290.CD-20-0766\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJogo T, Nakamura Y, Shitara K, Bando H, Yasui H, Esaki T, et al. Circulating Tumor DNA Analysis Detects FGFR2 Amplification and Concurrent Genomic Alterations Associated with FGFR Inhibitor Efficacy in Advanced Gastric Cancer. Clin Cancer Res. 2021;27:5619\u0026ndash;27. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1158/1078-0432.CCR-21-1414\u003c/span\u003e\u003cspan address=\"10.1158/1078-0432.CCR-21-1414\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGuercio BJ, Sarfaty M, Teo MY, Ratna N, Duzgol C, Funt SA, et al. Clinical and Genomic Landscape of FGFR3-Altered Urothelial Carcinoma and Treatment Outcomes with Erdafitinib: A Real-World Experience. Clin Cancer Res. 2023;29:4586\u0026ndash;95. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1158/1078-0432.CCR-23-1283\u003c/span\u003e\u003cspan address=\"10.1158/1078-0432.CCR-23-1283\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLi Y, Porta-Pardo E, Tokheim C, Bailey MH, Yaron TM, Stathias V, et al. Pan-cancer proteogenomics connects oncogenic drivers to functional states. Cell. 2023;186:3921\u0026ndash;e394425. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.cell.2023.07.014\u003c/span\u003e\u003cspan address=\"10.1016/j.cell.2023.07.014\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHao Y, Hao S, Andersen-Nissen E, Mauck WM, Zheng S, Butler A, et al. Integrated analysis of multimodal single-cell data. Cell. 2021;184:3573\u0026ndash;e358729. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.cell.2021.04.048\u003c/span\u003e\u003cspan address=\"10.1016/j.cell.2021.04.048\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKosti A, Chiou J, Guardia GDA, Lei X, Balinda H, Landry T, et al. ELF4 is a critical component of a miRNA-transcription factor network and is a bridge regulator of glioblastoma receptor signaling and lipid dynamics. Neuro Oncol Engl. 2023;25:459\u0026ndash;70. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1093/neuonc/noac179\u003c/span\u003e\u003cspan address=\"10.1093/neuonc/noac179\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhuang Z, Zhang C, Tan Y, Zhang J, Zhong C. ELF4 was a prognostic biomarker and related to immune infiltrates in glioma. J Cancer. 2024;15:5101\u0026ndash;17. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.7150/jca.96886\u003c/span\u003e\u003cspan address=\"10.7150/jca.96886\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZheng T, Zhou H, Zheng Z, Guo Y, Liu J, Zhang J, et al. The pathological significance and potential mechanism of ARHGEF6 in lung adenocarcinoma. Comput Biol Med. 2023;158:106894. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.compbiomed.2023.106894\u003c/span\u003e\u003cspan address=\"10.1016/j.compbiomed.2023.106894\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLi K, Wang H, Yang C, Li C, Xue B, Zhou J. Clinical implication and potential function of ARHGEF6 in acute myeloid leukemia: An in vitro study. PLoS ONE. 2023;18:e0283934. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1371/journal.pone.0283934\u003c/span\u003e\u003cspan address=\"10.1371/journal.pone.0283934\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYu J, Shi L, Lin W, Lu B, Zhao Y. UCP2 promotes proliferation and chemoresistance through regulating the NF-κB/β-catenin axis and mitochondrial ROS in gallbladder cancer. Biochem Pharmacol. 2020;172:113745. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.bcp.2019.113745\u003c/span\u003e\u003cspan address=\"10.1016/j.bcp.2019.113745\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang K, Zhang L, Deng B, Zhao K, Chen C, Wang W. Mitochondrial uncoupling protein 2: a central player in pancreatic disease pathophysiology. Mol Med. 2024;30:259. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1186/s10020-024-01027-y\u003c/span\u003e\u003cspan address=\"10.1186/s10020-024-01027-y\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCosta A, Diffley JFX. The Initiation of Eukaryotic DNA Replication. Annu Rev Biochem. 2022;91:107\u0026ndash;31. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1146/annurev-biochem-072321-110228\u003c/span\u003e\u003cspan address=\"10.1146/annurev-biochem-072321-110228\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTan Y, Zhou G, Wang X, Chen W, Gao H. USP18 promotes breast cancer growth by upregulating EGFR and activating the AKT/Skp2 pathway. Int J Oncol. 2018;53:371\u0026ndash;83. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3892/ijo.2018.4387\u003c/span\u003e\u003cspan address=\"10.3892/ijo.2018.4387\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDiao W, Guo Q, Zhu C, Song Y, Feng H, Cao Y, et al. USP18 promotes cell proliferation and suppressed apoptosis in cervical cancer cells via activating AKT signaling pathway. BMC Cancer. 2020;20:741. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1186/s12885-020-07241-1\u003c/span\u003e\u003cspan address=\"10.1186/s12885-020-07241-1\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhu S, Xu H, Chen R, Shen Q, Yang D, Peng H, et al. DNA methylation and miR-92a-3p-mediated repression of HIP1R promotes pancreatic cancer progression by activating the PI3K/AKT pathway. J Cell Mol Med. 2023;27:788\u0026ndash;802. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1111/jcmm.17612\u003c/span\u003e\u003cspan address=\"10.1111/jcmm.17612\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhu J, Wang X, Guan H, Xiao Q, Wu Z, Shi J, et al. HIP1R acts as a tumor suppressor in gastric cancer by promoting cancer cell apoptosis and inhibiting migration and invasion through modulating Akt. J Clin Lab Anal. 2020;34:e23425. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1002/jcla.23425\u003c/span\u003e\u003cspan address=\"10.1002/jcla.23425\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLu P, Chen J, Yan L, Yang L, Zhang L, Dai J, et al. RasGRF2 promotes migration and invasion of colorectal cancer cells by modulating expression of MMP9 through Src/Akt/NF-κB pathway. Cancer Biol Ther. 2019;20:435\u0026ndash;43. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1080/15384047.2018.1529117\u003c/span\u003e\u003cspan address=\"10.1080/15384047.2018.1529117\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNesci S, Rubattu S. UCP2, a Member of the Mitochondrial Uncoupling Proteins: An Overview from Physiological to Pathological Roles. Biomedicines. 2024;12:1307. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/biomedicines12061307\u003c/span\u003e\u003cspan address=\"10.3390/biomedicines12061307\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMart\u0026iacute;nez-Reyes I, Chandel NS. Cancer metabolism: looking forward. Nat Rev Cancer. 2021;21:669\u0026ndash;80. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/s41568-021-00378-6\u003c/span\u003e\u003cspan address=\"10.1038/s41568-021-00378-6\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFaubert B, Solmonson A, DeBerardinis RJ. Metabolic reprogramming and cancer progression. Science. 2020;368:eaaw5473. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1126/science.aaw5473\u003c/span\u003e\u003cspan address=\"10.1126/science.aaw5473\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eY S, J C, Q D, Z T. Molecular characteristics of early- and late-onset ovarian cancer: insights from multidimensional evidence. J ovarian Res [Internet] J Ovarian Res; 2025 [cited 2026 Jan 26];18. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1186/s13048-025-01664-9\u003c/span\u003e\u003cspan address=\"10.1186/s13048-025-01664-9\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eIm FP-V, Jp O-C, D M-L L-A Jr. Epicatechin Decreases UCP2 Gene Expression in MDA-MB-231 Breast Cancer Cells by the Presence of a Regulatory Element in the Promoter. International journal of molecular sciences [Internet]. Int J Mol Sci. 2025. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/ijms26094102\u003c/span\u003e\u003cspan address=\"10.3390/ijms26094102\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. [cited 2026 Jan 26];26.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":true,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"journal-of-translational-medicine","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"jtrm","sideBox":"Learn more about [Journal of Translational Medicine](http://translational-medicine.biomedcentral.com)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/jtrm/default.aspx","title":"Journal of Translational Medicine","twitterHandle":"@BioMedCentral","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"BMC/SO AJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"FGFR inhibitor, Drug resistance, Pan-cancer analysis, Multi-omics integration, Synthetic lethality","lastPublishedDoi":"10.21203/rs.3.rs-8869688/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8869688/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e \u003cp\u003eAlthough fibroblast growth factor receptor (FGFR) inhibitors (FGFRi) have demonstrated clinical promise, the inevitable emergence of acquired resistance remains a distinct bottleneck, severely compromising their long-term clinical efficacy. The pan-cancer molecular landscape and heterogeneous mechanisms driving this resistance, ranging from genetic alterations to dynamic network rewiring, remain poorly understood.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e \u003cp\u003eWe integrated large-scale pharmacogenomic profiling (GDSC2 and PRISM) with single-cell RNA sequencing to dissect the proteogenomic landscape of FGFRi resistance across 312 cell lines from 8 cancer types, complemented by machine learning modeling and systematic synthetic lethality screening to uncover actionable therapeutic targets.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eOur dual-database analysis unveiled a multi-dimensional atlas of FGFRi resistance. We identified cancer-specific genomic drivers, such as \u003cem\u003eELF4\u003c/em\u003e amplification in glioblastoma, alongside key transcriptomic markers including \u003cem\u003eUCP2\u003c/em\u003e and \u003cem\u003eFSCN1\u003c/em\u003e, highlighting a shift towards metabolic reprogramming and epithelial-mesenchymal transition (EMT). Single-cell resolution analysis unveiled that resistance is predominantly associated with the enrichment of subpopulations harboring aberrant cell-cycle dysregulation (MP2), suggesting a model of clonal selection rather than purely transcriptional plasticity-driven adaptation. Furthermore, a Random Forest model based on 52 mRNA features was constructed, demonstrating robust predictive capability for FGFRi sensitivity (AUC\u0026thinsp;\u0026gt;\u0026thinsp;0.7). Most notably, our synthetic lethal screening revealed a convergent reliance on compensatory RTK signaling (specifically EGFR pathway enrichment) and downstream MAPK/PI3K cascades in resistant phenotypes, providing robust evidence for an \"RTK switching\" mechanism.\u003c/p\u003e\u003ch2\u003eConclusions\u003c/h2\u003e \u003cp\u003eThis study establishes a high-resolution proteogenomic atlas of FGFRi resistance, identifying a convergent evolution towards metabolic reprogramming and EGFR-mediated bypass signaling. Our findings characterize resistance as a dynamic network rewiring and propose rational combination therapies (e.g., FGFRi combined with EGFR or metabolic inhibitors) to overcome resistance.\u003c/p\u003e","manuscriptTitle":"Integrative Proteogenomic and Single-Cell Analysis Reveals RTK Switching and Metabolic Reprogramming as Synthetic Lethal Vulnerabilities in FGFR Inhibitor Resistance","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-02-26 19:39:39","doi":"10.21203/rs.3.rs-8869688/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"reviewerAgreed","content":"","date":"2026-02-23T22:01:50+00:00","index":0,"fulltext":""},{"type":"reviewersInvited","content":"","date":"2026-02-23T21:32:54+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-02-18T11:22:25+00:00","index":"","fulltext":""},{"type":"submitted","content":"Journal of Translational Medicine","date":"2026-02-18T03:32:04+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"journal-of-translational-medicine","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"jtrm","sideBox":"Learn more about [Journal of Translational Medicine](http://translational-medicine.biomedcentral.com)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/jtrm/default.aspx","title":"Journal of Translational Medicine","twitterHandle":"@BioMedCentral","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"BMC/SO AJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"a07339b3-ca1a-4978-9a44-e5817dc32b3e","owner":[],"postedDate":"February 26th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2026-05-08T20:01:56+00:00","versionOfRecord":[],"versionCreatedAt":"2026-02-26 19:39:39","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8869688","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8869688","identity":"rs-8869688","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}