The drug resistance problem. Non-small cell lung cancer (NSCLC) accounts for roughly 85% of all lung cancer cases and remains the leading cause of cancer-related death globally. While treatment has advanced significantly - from surgery and chemotherapy to targeted therapy and immunotherapy - most patients eventually develop resistance to whatever treatment they receive, leading to disease recurrence and poor survival.
Single-omics approaches fall short. Traditional research into drug resistance has focused on individual molecular layers, such as genetic mutations. But drug resistance is not caused by a single change - it involves simultaneous dysregulation across multiple biological levels, including the genome, the transcriptome (RNA), the proteome (proteins), and the metabolome (small molecules). Studies relying on only one of these layers capture an incomplete picture of why and how tumors stop responding to treatment.
Proteomics as a critical connector. Among the various omics layers, proteomics - the large-scale study of proteins, including how they are modified and how they interact - is particularly valuable because proteins are the direct functional units of cellular biology. While a gene mutation may be detected by genomics and an altered RNA transcript by transcriptomics, it is the corresponding protein changes that actually execute the drug resistance mechanism. Proteomics thus bridges the gap between genetic information and biological function.
The promise of integration. Multi-omics integration combines data from genomics, transcriptomics, proteomics, and metabolomics to construct a holistic, systems-level view of drug resistance. This review synthesizes recent progress in this field, focusing specifically on how a proteomics-centered integrative framework can reveal molecular resistance networks that single-omics or non-integrated approaches miss entirely.
Genomics. Genomics analyzes the complete DNA sequence of cells, identifying mutations, copy number variations, and gene amplifications. In NSCLC, genomics has been foundational in discovering oncogenic driver mutations in EGFR, ALK, and KRAS genes that transformed treatment with targeted therapies. Large-scale studies like TRACERx, which analyzed 1,644 tumor regions from 421 NSCLC patients, have used genomics to reveal how intratumor heterogeneity - genetic differences between cells within a single tumor - drives metastasis and treatment failure.
Transcriptomics and single-cell analysis. Transcriptomics measures the RNA produced by cells, revealing which genes are actively expressed and how expression changes during drug resistance. Crucially, single-cell RNA sequencing (scRNA-seq) can detect rare resistant subpopulations within tumors that are invisible in bulk RNA measurements. Spatial transcriptomics maps gene expression to physical locations within tissue, identifying immune exclusion zones and region-specific resistance mechanisms that are lost when tissue is dissociated into single-cell suspensions.
Proteomics. Proteomics profiles proteins at scale using mass spectrometry, measuring not just protein abundance but also post-translational modifications like phosphorylation and ubiquitination that directly regulate protein activity. These modifications are critical because a protein that is present but not activated may look identical to one that is active at the mRNA or gene level - only proteomics can detect the functional difference. Stable isotope labeling techniques (SILAC, iTRAQ, TMT) allow precise comparison of protein expression between drug-resistant and drug-sensitive cells.
Metabolomics. Metabolomics profiles small-molecule metabolites - the downstream products of all upstream molecular activities. Drug-resistant NSCLC cells frequently show enhanced glycolysis (the Warburg effect), increased use of glutamine, and altered lipid metabolism that sustain proliferation under therapeutic stress. Metabolomics provides a functional readout of these adaptations, and integrating it with proteomics connects the enzymes driving metabolic reprogramming to the metabolic changes themselves.
EGFR-TKI resistance mechanisms. EGFR mutations are present in approximately 40-50% of NSCLC patients and are the most common target for tyrosine kinase inhibitor (TKI) therapy. Resistance to first-generation EGFR inhibitors (gefitinib, erlotinib) is dominated by the secondary T790M mutation, which physically blocks the drug from binding to EGFR. Third-generation drugs like osimertinib were designed to overcome T790M but now face their own resistance mechanisms, including C797S mutations, activation of AXL kinase, BRAF fusions, and recruitment of immunosuppressive cells to the tumor microenvironment.
Bypass signaling activation. A major general resistance strategy used by NSCLC tumors is activating a completely different growth signaling pathway that bypasses the one being blocked by the drug. MET amplification (in approximately 15% of EGFR-TKI resistant cases), HER2 amplification, and reactivation of other signaling networks allow tumors to maintain growth despite effective suppression of the primary drug target. These bypass mechanisms explain why single-target therapy is inherently limited.
KRAS and osimertinib resistance insights from proteomics. KRAS mutations are key drivers in lung cancer, and novel KRAS G12C inhibitors have entered clinical use. Proteomics has been essential in investigating resistance to these new drugs - TMT quantitative proteomics created a comprehensive library of how tumors respond to KRAS inhibition, elucidating resistance mechanisms and identifying potential combination therapies. A plasma proteomics study using 8 proteins including CCL23 and ARG1 built an I-SCORE model that achieved an AUC of 0.94 for predicting 12-month overall survival in osimertinib-treated patients.
EMT and histological transformation. Beyond molecular pathway changes, NSCLC tumors can undergo fundamental biological transformations to escape targeted therapy. Epithelial-mesenchymal transition (EMT), in which cancer cells adopt a more invasive, mesenchymal character, is associated with multiple drug resistance mechanisms. In rare but important cases, EGFR-mutant NSCLC can transform into small cell lung cancer - a completely different histological type with a distinct biology - as a mechanism of escape from EGFR-targeted therapy.
Platinum drug resistance. Platinum-based drugs (cisplatin, carboplatin) are cornerstones of NSCLC chemotherapy, acting by cross-linking DNA strands to block replication. Tumors resist these drugs by enhancing DNA repair mechanisms - particularly nucleotide excision repair driven by the ERCC1 protein, which is a validated proteomic biomarker of platinum resistance - as well as through mutations in DNA damage response pathways and alterations in cell cycle regulation (p53 mutations).
Taxane and antimetabolite resistance. Paclitaxel and docetaxel work by stabilizing microtubules and blocking cell division. Resistance involves altered expression of tubulin protein variants (identified by proteomics) and overexpression of anti-apoptotic proteins like Bcl-2 that prevent drug-induced cell death. For pemetrexed, an antimetabolite, resistance involves downregulation of the folate transporter needed to import the drug into cells, amplification of its target enzyme (thymidylate synthase), and epigenetic alterations.
Multi-drug resistance and ABC transporters. One of the broadest resistance mechanisms is multi-drug resistance (MDR), in which tumor cells develop the ability to actively pump multiple chemotherapy drugs out before they can act. This is mainly driven by ABC transporter proteins, particularly P-glycoprotein, whose expression is regulated by long non-coding RNAs (lncRNAs) that shield ABC transporter genes from suppression by microRNAs. Cancer stem cells, which are inherently resistant and can regenerate the tumor population, also contribute to MDR.
Epigenetic regulation of resistance. Drug resistance in NSCLC is also maintained by epigenetic mechanisms - changes in gene activity that do not alter the DNA sequence. Hypermethylation of the RASSF1A tumor suppressor gene promoter silences its expression, disrupting cell cycle control. Alterations in histone H3K27me3 marks change the expression of multidrug resistance proteins. A self-reinforcing loop exists: upregulated DNA methyltransferases in resistant cells cause further gene methylation, locking in the resistant state even in the absence of drug selection pressure.
Genomics plus proteomics. When genomics identifies a mutation like EGFR T790M, phosphoproteomic analysis can confirm the functional consequence: hyperphosphorylation of downstream effectors AKT and ERK, demonstrating that oncogenic signaling is maintained despite the presence of the drug. This integration connects the genetic change to the functional resistance mechanism and directly informs drug development - for example, the inhibitor YK-029A was designed to suppress both the mutation and downstream phosphorylation signaling simultaneously.
Transcriptomics plus proteomics - closing the mRNA-protein gap. A critical insight from integrating these two layers is that mRNA levels and protein levels frequently do not match. The tumor suppressor PTEN is often dramatically reduced in resistant cells even when its mRNA level is unchanged, indicating that post-transcriptional mechanisms - impaired translation or enhanced protein degradation - are responsible. The RNA-binding protein HuR stabilizes mRNAs encoding anti-apoptotic proteins like Bcl-2, increasing their translation and enabling cells to evade chemotherapy-induced cell death - a mechanism invisible to transcriptomics alone.
Metabolomics plus proteomics - mapping metabolic resistance. Drug-resistant cells upregulate glycolytic enzymes (hexokinase 2, pyruvate kinase M2) detected by proteomics, and corresponding metabolomic profiling confirms elevated lactate and glycolytic intermediates indicating a switch to aerobic glycolysis. Similarly, proteomic identification of elevated fatty acid synthase aligns with metabolomic detection of increased lipid species, confirming lipogenesis-driven chemoresistance. The transcriptional coactivator PGC-1alpha, identified proteomically in resistant cells, enhances mitochondrial metabolism and contributes to survival under therapeutic stress.
Single-cell multi-omics for heterogeneity. Machine learning-guided single-cell multi-omics recently uncovered an immunosuppressive niche driven by GDF15 in NSCLC, revealing a mechanism of resistance to anti-PD-1 checkpoint therapy. A study analyzing over 232,000 cells from 19 NSCLC patients found that interactions between tumor cells and specific macrophage and cancer-associated fibroblast subsets blocked T-cell infiltration - a finding correlated with worse patient outcomes. Such discoveries are only possible through simultaneous analysis of cell identity and molecular state.
Small molecule inhibitors of resistance proteins. Multi-omics-identified resistance proteins are being targeted with small molecule drugs. Osimertinib, which selectively inhibits EGFR carrying the T790M mutation, is a prototype: it was developed specifically to overcome resistance to first-generation EGFR inhibitors and substantially prolongs patient survival. Inhibitors of hexokinase 2 disrupt the hyperactive glycolytic metabolism identified by metabolomics in resistant cells. ABC transporter inhibitors like verapamil can restore drug accumulation in multidrug-resistant cells, though clinical translation has been limited by off-target toxicity.
Multi-omics-guided combination therapy. Because resistance in NSCLC rarely results from a single mechanism, multi-omics analyses that capture co-activated bypass pathways can rationally guide combination therapy design. When EGFR-TKI resistance involves MET amplification, adding a MET inhibitor (crizotinib) creates a combination that simultaneously suppresses both pathways. When HER2 amplification is the bypass mechanism, an EGFR-TKI plus a HER2 inhibitor (trastuzumab) combination may overcome resistance. This multi-omics-informed strategy is more mechanistically precise than trial-and-error drug combinations.
Targeting the tumor microenvironment. Multi-omics studies have revealed that the tumor microenvironment (TME) - the non-cancer cells and structures surrounding tumors - actively protects cancer cells from chemotherapy. Cancer-associated fibroblasts create physical barriers and protective signaling; abnormal vasculature restricts drug delivery; and immunosuppressive cells provide survival signals. Combining chemotherapy with VEGF inhibitors not only kills tumor cells directly but also normalizes vasculature and reduces protective stromal interactions, thereby overcoming this environmental form of resistance.
Prioritizing drug targets from multi-omics data. Multi-omics studies generate many candidate resistance proteins, so systematic prioritization is essential. The most actionable targets are those at central nodes of critical survival pathways, those with druggable protein structures (binding pockets accessible to small molecules or antibodies), those that are specifically overexpressed in tumors versus normal tissue, and those associated with resistant tumor subtypes lacking effective treatments. This framework focuses drug development resources on the highest-value opportunities.
Data integration barriers. Each omics technology produces data in different formats, scales, and units that are difficult to combine directly. RNA sequencing data is measured in expression values (FPKM, TPM), while proteomics yields peptide intensities. Systematic technical variations between laboratories and instruments can produce significantly different results from the same sample, undermining comparability. Emerging AI-driven tools like DeepOmics (a convolutional neural network for cross-modality normalization) and network frameworks like MOFA and WGCNA offer promising paths to standardized integration.
Cost and throughput limitations. High-resolution mass spectrometers required for proteomics represent major capital investments, with complex sample preparation that limits throughput to far fewer samples per run than sequencing technologies can process. These limitations restrict large-scale clinical cohort studies needed to validate potential biomarkers across diverse patient populations, creating a bottleneck between discovery and clinical translation.
The bench-to-bedside gap. Proteins and genes identified as resistance drivers in controlled laboratory models may not function the same way in actual patients, where the TME, individual genetic variation, and treatment history all influence drug response. Biomarkers showing promising predictive value in small studies often show reduced accuracy in larger multicenter trials due to differences in patient demographics, disease characteristics, and sequencing platforms. Better integration between researchers and clinicians is needed to ensure research priorities align with clinical reality.
Future priorities. The field is moving toward simultaneous multi-omics profiling from single samples to eliminate temporal inconsistency between measurements. AI and machine learning algorithms will be essential for automated data integration and pattern identification at scales beyond human analytical capacity. An international NSCLC multi-omics standardization committee - comprising researchers, clinicians, and bioinformaticians - has been proposed to develop unified protocols, enabling cross-institutional data sharing. The ultimate goal is multi-omics-based personalized medicine, where each patient's tumor is profiled across all molecular layers to identify their specific resistance mechanisms and inform customized treatment strategies.