Genomic profiling enables personalized strategies to overcome drug resistance in multiple myeloma

Abstract

Multiple myeloma (MM) is a hematologic malignancy characterized by the clonal proliferation of malignant plasma cells within the bone marrow. It is a complex and heterogeneous disease that primarily affects older adults and is associated with significant morbidity and mortality. Despite advances in treatment, including proteasome inhibitors, immunomodulatory drugs, and monoclonal antibodies, MM remains largely incurable, with most patients experiencing relapse and disease progression. The disease’s biological complexity is driven by a multitude of genetic and molecular alterations, which contribute to its clinical variability and resistance to therapy. The development of drug resistance is a major obstacle in the management of MM, often leading to treatment failure and disease relapse. Resistance mechanisms can be intrinsic, present before therapy initiation, or acquired, emerging during the course of treatment. These mechanisms involve complex molecular pathways, genetic mutations, and interactions within the bone marrow microenvironment that enable malignant plasma cells to evade therapeutic effects. A comprehensive understanding of these resistance pathways is crucial for designing strategies to overcome or prevent resistance, thereby improving patient outcomes. Integrating insights from molecular pathways and genomic profiling into clinical practice holds promise for tailoring personalized therapies that can effectively target resistant disease clones and prolong remission. Additionally, exploring combination therapies that target multiple pathways simultaneously, informed by molecular profiling, holds promise for overcoming intrinsic and acquired resistance. Continued innovation in predictive modeling and functional assays will facilitate the translation of molecular insights into effective, individualized treatment plans.

Introduction

Multiple myeloma (MM) is genomically heterogeneous and prone to therapeutic escape. Objective: this narrative review tests the hypothesis that genomic and multi-omic profiling are not only descriptive tools but actionable decision aids that (1) identify specific molecular mechanisms of resistance, (2) enable real-time monitoring of emergent resistant clones, and (3) map directly to concrete therapeutic choices (drug switches, rational combinations, or enrollment in targeted trials). To be distinct from prior reviews that catalog mechanisms, we synthesize genomic examples into a practical, clinician-oriented framework that shows exactly how a genomic finding changes management (e.g., PSMB5 mutations → PI selection; CRBN truncation → avoid IMiD retreatment). Evidence that genomic readouts can directly inform decisions and monitoring is rapidly maturing [1–3]. Importantly, the clinical and biological mechanisms of drug resistance in plasma-cell disorders unfold along a disease continuum, from monoclonal gammopathy of undetermined significance (MGUS) and smoldering myeloma (SMM) to newly-diagnosed symptomatic MM, early relapse and finally heavily pretreated, multi- or triple-class refractory disease. Resistance emerging at different points on this timeline is often biologically distinct: early relapse (commonly defined as progression after 1–3 prior lines) frequently reflects selection or expansion of pre-existing subclones and reversible transcriptional/adaptive programs, whereas late, heavily pretreated or triple-class refractory disease more commonly harbors complex polyclonal genomic architecture, structural variation, antigen loss and profound microenvironmental and immune-exhaustion signatures that together limit the activity of later-line targeted and immune therapies. Explicitly framing resistance by disease stage is therefore essential to translate genomic readouts into stage-appropriate clinical choices (for example, early targeted therapy or immune-based consolidation in genomically simple high-risk SMM vs. combinatorial strategies or antigen-agnostic approaches in polyclonal triple-refractory disease) [4–6].

Modern therapies have substantially improved response rates, but MM’s intra- and inter-patient heterogeneity often produces primary and acquired resistance. Primary resistance stems from pre-existing, treatment-insensitive clones; acquired resistance reflects selection of resistant subclones during therapy. Resistance arises from tumor-intrinsic alterations and protective signals supplied by the bone-marrow microenvironment [7–9].

Cell-autonomous resistance mechanisms encompass alterations within the MM cells themselves that confer survival advantage in the presence of therapeutic agents. Genetic alterations play a pivotal role, including mutations and copy number variations (CNVs) in genes frequently implicated in MM pathogenesis or drug response [10]. For instance, mutations in TP53 are often associated with high-risk MM and resistance to various agents, while alterations in the RAS/RAF pathway can lead to constitutive activation of survival signaling [11]. Similarly, specific chromosomal translocations, such as t(4;14) or t(14;16), are linked to aggressive disease and resistance. Epigenetic modifications, including DNA methylation and histone acetylation/deacetylation, can alter gene expression profiles, leading to the silencing of pro-apoptotic genes or upregulation of resistance-associated genes. For example, hypermethylation of tumor suppressor genes can contribute to drug resistance [12, 13].

Another critical mechanism involves alterations in drug metabolism and efflux. Upregulation of ATP-binding cassette (ABC) transporters, such as P-glycoprotein (MDR1/ABCB1) or Breast Cancer Resistance Protein (BCRP/ABCG2), can actively pump drugs out of MM cells, reducing intracellular drug concentrations below therapeutic thresholds [14]. This is a common mechanism observed across various cancer types and therapeutic agents. Furthermore, MM cells can develop profound changes in their drug targets. For PIs, mutations in proteasome subunits (e.g., PSMB5) can reduce drug binding affinity, thereby abrogating their cytotoxic effects [15]. Alternatively, activation of parallel or bypass signaling pathways, such as the PI3K/AKT/mTOR, MAPK/ERK, or NF-κB pathways, can circumvent drug-induced apoptosis. These pathways provide alternative survival cues, allowing MM cells to proliferate even when a primary pro-survival pathway is inhibited [16]. Enhanced DNA repair mechanisms and the activation of autophagy or unfolded protein response (UPR) pathways can also contribute to cellular resilience and survival under drug-induced stress [17].

Microenvironment-mediated resistance mechanisms highlight the crucial role of the BMM in shielding MM cells from therapeutic insult, a phenomenon often termed “cell adhesion-mediated drug resistance” (CAM-DR). The BMM is a complex ecosystem comprising various cell types, including stromal cells, osteoclasts, osteoblasts, endothelial cells, and immune cells, along with an intricate extracellular matrix [18]. MM cells adhere to BMM components via adhesion molecules like VCAM-1, ICAM-1, and fibronectin, triggering downstream signaling pathways (e.g., NF-κB, JAK/STAT) that promote MM cell survival, proliferation, and drug resistance [19]. This adhesion can also physically protect MM cells from drug exposure. Beyond direct cell adhesion, the BMM secretes a plethora of soluble factors that foster MM cell survival and resistance [20]. Cytokines such as IL-6, IGF-1, APRIL, and BAFF, produced by stromal cells and other BMM components, act as potent growth and survival factors for MM cells, activating various pro-survival signaling cascades [21]. For example, IL-6 is a well-established driver of MM cell proliferation and can render cells resistant to chemotherapy. The BMM also plays a critical role in angiogenesis, providing essential nutrients and oxygen, and modulating the immune response, often creating an immunosuppressive environment that further contributes to MM cell escape from both host immunity and immune-targeting therapies [22, 23].

The profound complexity of these resistance mechanisms necessitates a sophisticated approach to patient management. This is where the integration of genomic profiling has emerged as a transformative tool. Next-generation sequencing (NGS) technologies, including whole exome sequencing (WES), whole genome sequencing (WGS), RNA sequencing (RNA-seq), and single-cell sequencing, allow for an unprecedented interrogation of the MM genome and transcriptome at high resolution [24]. The persistent challenge of drug resistance in MM necessitates a thorough synthesis of current knowledge regarding its mechanisms. Although novel therapeutic agents have extended patient survival, many individuals ultimately develop resistance, highlighting a critical gap in long-term disease management [25].

This narrative review has three explicit aims and one distinict contribution: (1) Concise catalogue: summarize resistance mechanisms that are detectable by current genomic and multi-omic assays; (2) Actionability: evaluate, with clinical examples, how specific genomic findings change treatment choice or sequencing (notably PSMB5, PSMG2, CRBN and pathway activations); (3) Operational guidance: provide a practical framework for clinicians and investigators to report genomic results and use them to select targeted or combination strategies. Distinct contribution: rather than only listing mechanisms, the manuscript translates genomic findings into decision rules (e.g., which mutations should prompt therapy change, which require functional confirmation, and how to integrate ctDNA for serial monitoring) [26, 27].

Methods, literature search and selection

This manuscript is a narrative review. To increase transparency and reduce selection bias, a concise, reproducible approach to literature identification and synthesis was followed. A structured search of PubMed/MEDLINE, Embase, Web of Science and Google Scholar was performed for publications in English from 1 January 2010 through 31 December 2025. Search terms combined subject headings and free-text words such as “multiple myeloma”, “myeloma”, “drug resistance”, “proteasome inhibitor”, “bortezomib”, “carfilzomib”, “IMiD”, “lenalidomide”, “pomalidomide”, “CRBN”, “proteasome”, “genomic profiling”, “next generation sequencing”, “clonal evolution”, “bone marrow microenvironment”, and related synonyms. Titles and abstracts were screened for relevance and full texts were reviewed when the abstract suggested a substantive contribution to mechanisms of resistance, genomic profiling, or genomically informed therapeutic strategies. The review prioritized (a) original research articles, (b) clinical trials, (c) recent systematic reviews and meta-analyses, and (d) translational/preclinical studies that clarify molecular mechanisms; letters without data and conference abstracts without full text were excluded. As this is a narrative (not systematic) review, no formal risk-of-bias or meta-analytic procedures were applied. Where appropriate, seminal older works and high-impact recent papers published after the search cutoff were included to ensure conceptual completeness. This pragmatic, minimally structured approach follows current recommendations for transparent reporting of narrative reviews.

Clinical overview of MM

Multiple myeloma is a clinically heterogeneous bone-marrow plasma-cell malignancy with diverse genetic subgroups; understanding how genotype translates into differential drug sensitivity is essential for tailoring therapy and addressing relapse [28]. Treatments include a range of agents such as alkylating agents (e.g., melphalan, cyclophosphamide), corticosteroids (e.g., dexamethasone), immunomodulatory drugs (e.g., thalidomide, lenalidomide, pomalidomide), and proteasome inhibitors (e.g., bortezomib, carfilzomib, ixazomib) [29]. Despite important therapeutic advances, MM remains largely incurable for many patients because of relapse driven by drug resistance. Median overall survival with modern therapies is on the order of 6–8 years, with higher long-term survival in transplant-eligible patients; population 5-year survival has risen substantially in recent decades [30, 31].

At its core, MM originates from a malignant transformation of a post-germinal center B-lymphocyte, leading to the uncontrolled expansion of a single clone of plasma cells. Importantly, this malignant clone evolves within a dynamic bone-marrow ecosystem, a continuum that begins in premalignant states (MGUS and smoldering myeloma) and progresses to active MM under the combined influence of tumor-intrinsic changes and progressive niche remodeling; recognizing this MGUS → MM continuum is essential for identifying early, microenvironment-directed interventions that may prevent or delay the emergence of therapy-resistant disease [32]. These clonal plasma cells typically reside in the bone marrow, where they interact intimately with the BMM [33]. This complex interplay involves various cellular and soluble factors, including stromal cells, osteoclasts, osteoblasts, cytokines (e.g., IL-6, TNF-alpha, VEGF), and adhesion molecules, which collectively promote myeloma cell growth, survival, and drug resistance [34]. A hallmark of MM is the production of a monoclonal immunoglobulin, or “M-protein,” by the aberrant plasma cells. This M-protein can be an intact immunoglobulin (IgG, IgA, IgD, IgE, rarely IgM) or, more commonly, free light chains (kappa or lambda) [35]. The accumulation of these proteins, along with the direct effects of plasma cell infiltration, is responsible for the characteristic end-organ damage [36]. Genetically, MM is highly diverse, often involving complex karyotypes, chromosomal translocations (e.g., t(11;14), t(4;14), t(14;16)), and deletions (e.g., del(17p)). These genetic aberrations influence disease progression, prognosis, and therapeutic response, underscoring the molecular heterogeneity of the disease [37].

The clinical presentation of MM is highly variable, ranging from asymptomatic disease detected incidentally to severe, rapidly progressive symptoms. The classic features of active MM, often remembered by the acronym CRAB, reflect the direct consequences of plasma cell proliferation and M-protein production: Calcium Elevation (Hypercalcemia): Occurs in approximately 20% of patients, primarily due to increased osteoclastic activity mediated by myeloma cells, leading to bone resorption. Symptoms include fatigue, confusion, polyuria, and constipation; Renal impairment in MM most commonly results from light-chain cast nephropathy (LCCN), intratubular precipitation of filtered free light chains with uromodulin causing tubular obstruction, and from light-chain deposition disease or AL amyloidosis. Contributing/exacerbating factors include hypercalcemia, dehydration, and exposure to nephrotoxic agents (e.g., NSAIDs, radiographic contrast), which can reduce renal perfusion or tubular flow and thereby worsen light-chain-mediated injury, but nephrotoxic drugs per se generally exacerbate rather than independently cause LCCN. Rapid reduction of serum free light chains and correction of reversible factors are central to renal recovery [38, 39]; Anemia: Present in nearly 70% of patients, typically normocytic and normochromic. It is caused by bone marrow infiltration, suppression of erythropoiesis, chronic inflammation, and renal insufficiency. Symptoms include fatigue, weakness, and dyspnea; Bone Lesions: The most common and often devastating complication, manifesting as osteolytic lesions (punched-out lesions), generalized osteopenia, pathological fractures, and severe bone pain. These lesions are a result of uncoupled bone remodeling, where osteoclast activity is increased and osteoblast activity is suppressed. Spinal cord compression is a serious neurological emergency that can arise from vertebral collapse or extramedullary plasmacytomas [40–44]. Beyond the CRAB criteria, patients may experience recurrent infections due to humoral and cellular immune dysfunction, neuropathy (often due to direct nerve infiltration or amyloidosis), hyperviscosity syndrome (rare, more common with IgA or IgM paraproteins), and amyloidosis (AL type), which can affect various organs including the heart, kidneys, and nervous system [45].

Treatment strategies: a rapidly evolving landscape

The treatment landscape for MM has undergone a revolutionary transformation in the past two decades, leading to significant improvements in patient outcomes. Treatment selection is highly individualized, based on patient fitness, age, disease risk, and prior therapies. The primary goals are to achieve deep and durable responses, improve quality of life, and extend survival [46].

Induction Therapy: Typically involves a 3 or 4-drug regimen combining a proteasome inhibitor (PI, e.g., bortezomib), an immunomodulatory drug (IMiD, e.g., lenalidomide), and a corticosteroid (dexamethasone). Daratumumab, a CD38 monoclonal antibody, is increasingly incorporated into frontline regimens (e.g., daratumumab-VRd) [47]. High-Dose Chemotherapy with Autologous Stem Cell Transplantation (ASCT): Following induction, eligible patients undergo melphalan-based high-dose chemotherapy followed by ASCT. This remains a cornerstone of therapy, offering deep and durable remissions [48]. Consolidation/Maintenance Therapy: Post-ASCT, maintenance therapy, typically with lenalidomide, is often prescribed to prolong remission duration [49]. Consolidation refers to a short, fixed-duration therapy given after ASCT (or after induction in non-transplant patients) with the aim of deepening response (increasing CR/MRD-negativity) before transitioning to longer-term maintenance. Typical consolidation regimens mirror induction (for example 2–4 cycles of bortezomib/lenalidomide/dexamethasone or bortezomib-based triplets) and have shown improvements in depth of response and progression-free survival in several trials (e.g., EMN02/HOVON95 and other studies), although the OS benefit is variable and practice varies by risk group and region. Consolidation is therefore considered an optional strategy to deepen remission, particularly in higher-risk patients or those with suboptimal MRD after ASCT [50, 51]. In transplant-ineligible patients, VMP (bortezomib-melphalan-prednisone) remains a widely used fixed-duration frontline option (VISTA trial established VMP’s benefit vs. MP), particularly in parts of Europe and in patients where continuous lenalidomide is less suitable. For transplant-ineligible patients, daratumumab plus lenalidomide and dexamethasone (D-Rd / DRD) has emerged as a frontline preferred option because the MAIA phase-III trial demonstrated substantial improvements in PFS and OS compared with Rd alone [52, 53]. By contrast, high-dose melphalan (HDT) remains the alkylator used for conditioning prior to ASCT. Melphalan should not be used as part of induction therapy for transplant-eligible patients. Its role in current practice is limited to (a) high-dose melphalan as conditioning prior to autologous stem-cell transplantation (ASCT) in transplant-eligible patients, and (b) oral melphalan in fixed-duration VMP (bortezomib–melphalan–prednisone) regimens for transplant-ineligible patients [54, 55]. The management of RRMM is complex, leveraging a growing armamentarium of drugs. The strategy depends on prior therapies, response duration, and tolerability. Retreating with agents from a different drug class or combination, or using newer drugs, is standard [56]. Figure 1 provides a schematic overview of the major drug classes used in MM treatment.

Fig. 1.

Schematic illustrating major drug classes used in MM treatment, including proteasome inhibitors, immunomodulatory agents, monoclonal antibodies, corticosteroids, alkylating agents, histone deacetylase inhibitors, nuclear export inhibitors, and bispecific antibodies. Each class exerts distinct mechanisms contributing to frontline and relapsed/refractory MM therapy

Disease continuum and stage-specific resistance

A practical understanding of resistance requires explicit attention to the disease stage and prior treatment exposures. Precursor states (MGUS, SMM) frequently harbor limited mutational burden and may be driven by a few early driver events; progression to overt MM is often associated with acquisition of driver mutations (KRAS/NRAS/BRAF, MAPK pathway), copy-number changes and structural variants that increase clonal complexity. In newly relapsed disease (early relapse), resistance commonly represents selection of a dominant resistant subclone or reversible transcriptional adaptation; in contrast, heavily pretreated and triple-class refractory disease typically shows polyclonal resistance mechanisms (multiple co-occurring mutations, structural rearrangements, antigen loss, and microenvironmental remodeling) that reduce the likelihood of single-agent re-sensitization. Recognising these stage-dependent patterns helps prioritise diagnostic assays (e.g., focused targeted panel or ctDNA for early relapse versus deep WGS/single-cell profiling in complex, late-line disease) and informs whether to pursue antigen-directed or antigen-agnostic strategies [5, 6, 57]. Table 1 summarizes the key therapeutic agents used in MM.

Table 1.

Key therapeutic agents for MM

Drug class	Examples	Mechanism of action	Role in therapy
Proteasome inhibitors	Bortezomib, Carfilzomib, Ixazomib	Inhibit the proteasome, leading to accumulation of unfolded proteins, ER stress, and apoptosis in myeloma cells.	Frontline, relapsed/refractory (RRMM) [47]
Immunomodulatory drugs	Lenalidomide, Pomalidomide, Thalidomide	Act via cereblon, leading to degradation of protein targets, immunomodulation (e.g., increased T/NK cell activity), and anti-angiogenesis.	Frontline, maintenance, RRMM [48]
Monoclonal antibodies	Daratumumab, Isatuximab (anti-CD38); Elotuzumab (anti-SLAMF7)	Target specific surface antigens on myeloma cells, inducing cell death via various mechanisms (e.g., ADCC, CDC, direct apoptosis).	Frontline, RRMM [12, 48]
Corticosteroids	Dexamethasone	Induce apoptosis in myeloma cells, reduce inflammation, and enhance efficacy of other agents.	Cornerstone of virtually all MM regimens [58]
Alkylating agents	Melphalan (high-dose for ASCT); Cyclophosphamide (less common now)	Cross-link DNA, inhibiting DNA replication and transcription.	High-dose for ASCT, rarely in specific regimens for transplant-ineligible or RRMM [56, 58]
Histone deacetylase inhibitors	Panobinostat	Inhibit HDACs, leading to chromatin remodeling, altered gene expression, and apoptosis.	Approved for RRMM in combination with bortezomib and dexamethasone (less commonly used due to toxicity) [56]
Nuclear Export inhibitors	Selinexor	Inhibits XPO1, leading to nuclear accumulation of tumor suppressor proteins and apoptosis.	Approved for highly refractory RRMM [56, 58]
Bispecific antibodies	Teclistamab (BCMA x CD3)	Engages T-cells to target BCMA-expressing myeloma cells.	Emerging/approved for highly refractory RRMM [56]
CAR-T Cell therapy	Idecabtagene vicleucel (ide-cel), Ciltacabtagene autoleucel (cilta-cel)	Genetically engineered T-cells targeting B-cell maturation antigen (BCMA) on myeloma cells.	Approved for highly refractory RRMM [56]

Principles of drug resistance in oncology

Resistance arises from (1) cell-intrinsic genomic/epigenetic alterations, (2) adaptive stress responses that rewire proteostasis and metabolism, and (3) microenvironmental protection. Importantly, many of these mechanisms are detectable by genomic or multi-omic assays (e.g., PSMB5 and PSMG2 variants for PI resistance; CRBN truncations for IMiD resistance; transcriptional/proteomic signatures for metabolic compensation) [31, 59–61]. We emphasize that drug resistance in multiple myeloma should be regarded as an evolutionary process rather than a series of isolated, static events. Under this view, therapy imposes selective pressure on a genetically and phenotypically heterogeneous population of myeloma cells; resistant subclones expand, sensitive clones contract, and phenotypic states can interconvert through reversible epigenetic and transcriptional reprogramming. Clonal competition, microenvironmental constraints and therapy-induced adaptive transcriptional programs therefore jointly shape the trajectories of relapse and refractory disease. Modern single-cell and longitudinal genomic assays are uniquely able to resolve these dynamics, revealing both genetic selection and non-genetic (plastic) adaptive responses that contribute to treatment failure [62–64]. The cellular and molecular mechanisms underlying drug resistance are diverse and often interconnected. They involve alterations at multiple levels, from gene expression and protein function to cellular metabolism and interactions with the tumor microenvironment. These mechanisms can emerge or become dominant through genetic mutations, epigenetic modifications, or adaptive responses [65]. Table 2 outlines the key mechanisms of drug resistance in oncology.

Table 2.

Key mechanisms of drug resistance in oncology

Mechanism category	Specific mechanism & description	Examples
Drug inactivation/efflux	Increased Drug Efflux: Overexpression of ATP-binding cassette (ABC) transporters actively pumps drugs out of the cell. Drug Metabolism: Increased activity of enzymes (e.g., cytochrome P450) that metabolize and inactivate drugs.	P-glycoprotein (MDR1/ABCB1), MRP1 (ABCC1), BCRP (ABCG2) overexpression reducing intracellular concentrations of chemotherapy drugs (e.g., doxorubicin, paclitaxel) and tyrosine kinase inhibitors (TKIs). Glutathione S-transferase (GST) systems detoxifying alkylating agents [31]
Target alteration/bypass	Target Mutations/Amplifications: Mutations in the drug target prevent drug binding or reduce its efficacy; amplification of the target can overcome drug inhibition. Target Downregulation: Reduced expression of the drug target. Bypass Signaling Pathways: Activation of alternative signaling pathways that compensate for the inhibited pathway.	EGFR mutations (e.g., T790M in NSCLC resisting first-generation EGFR TKIs), BRAF mutations (e.g., V600E in melanoma) developing secondary resistance mutations [61] ER downregulation in breast cancer becoming resistant to endocrine therapy. Activation of MET or HER2 signaling pathways in EGFR-mutant NSCLC cells treated with EGFR TKIs; activation of PI3K/AKT/mTOR pathway in cancers treated with MEK inhibitors.
DNA damage response	Enhanced DNA Repair: Increased capacity to repair DNA damage induced by chemotherapy or radiotherapy. Apoptosis Inhibition: Upregulation of anti-apoptotic proteins or downregulation of pro-apoptotic proteins, rendering cells resistant to cell death.	Overexpression of PARP1 contributing to resistance to PARP inhibitors; increased activity of homologous recombination and non-homologous end-joining pathways. Upregulation of Bcl-2 or Mcl-1, or loss of p53 function, leading to resistance to various cytotoxic agents [59, 60]
Cellular state & microenvironment	Epithelial-Mesenchymal Transition (EMT): Cells acquire stem-cell like features, increased motility, and resistance to apoptosis. Cancer Stem Cells (CSCs): Intrinsic resistance due to quiescence, enhanced DNA repair, and high expression of efflux pumps. Tumor Microenvironment (TME): Stromal cells, hypoxia, and extracellular matrix components provide protective signals and physical barriers. Metabolic Reprogramming: Altered metabolic pathways support survival under drug pressure.	Acquisition of mesenchymal markers (e.g., vimentin) and loss of epithelial markers (e.g., E-cadherin) in various cancers. CD133+, ALDH+, or CD44+/CD24- CSC populations demonstrating intrinsic resistance to numerous agents. Fibroblasts, macrophages, and endothelial cells secreting growth factors (e.g., HGF) that activate bypass pathways; hypoxic conditions leading to selection of resistant clones; acidic pH reducing drug uptake. Glycolytic shift (Warburg effect) providing energy and biosynthetic precursors for resistant cell growth [60]
Immune evasion	Loss of Antigen Presentation: Decreased expression of MHC-I or tumor antigens. Immune Checkpoint Upregulation: Increased expression of inhibitory checkpoint ligands (e.g., PD-L1). Immunosuppressive TME: Recruitment of regulatory T cells (Tregs) and myeloid-derived suppressor cells (MDSCs).	Loss of MHC-I expression compromising T-cell recognition. Upregulation of PD-L1 on tumor cells and TME cells leading to resistance to immune checkpoint blockade. Increased Treg or MDSC infiltration suppressing anti-tumor immune responses, limiting immunotherapy efficacy [65]

Addressing drug resistance requires a multifaceted and adaptive approach, informed by a deep understanding of its underlying mechanisms. Table 3 describes emerging strategies to combat drug resistance.

Table 3.

Emerging strategies to combat drug resistance

Strategy category	Approach & rationale	Examples & development focus
Combination therapies	Simultaneously targeting multiple pathways or using agents with complementary mechanisms to prevent the emergence of single-mechanism resistance.	Combining targeted agents (for MM: e.g., BRAF/MEK inhibitors in BRAF V600E-mutated relapsed/refractory MM; or MEK-directed combinations when RAS pathway activation is dominant) to block compensatory pathways and overcome lesion-specific resistance [56, 61]
Sequential/rotational therapies	Administering different drugs or drug combinations in a sequence to prevent the dominance of any single resistant clone.	Cycling through different agents or regimens. Utilizing liquid biopsies to monitor resistance evolution and switch therapies proactively [59, 61]
Next-generation inhibitors	Developing drugs that specifically overcome known resistance mutations or target alternative pathways activated in resistance.	Third-generation EGFR TKIs (e.g., osimertinib) designed to overcome the T790M resistance mutation in NSCLC. Fourth-generation BTK inhibitors for CML or lymphoma with specific BTK mutations [31, 60]
Targeting resistance mechanisms	Directly inhibiting efflux pumps, enhancing DNA repair sensitivity, or reversing EMT-driven resistance.	ABC transporter inhibitors (e.g., verapamil, tariquidar) to restore sensitivity to chemotherapy (though clinical success has been limited due to toxicity and drug-drug interactions). Chk1 inhibitors or WEE1 inhibitors to target enhanced DNA repair pathways. Agents targeting EMT transcription factors or associated signaling pathways [59, 65]
Tumor microenvironment modulation	Targeting components of the TME that protect cancer cells from therapy or promote resistance.	Targeting stromal fibroblasts (e.g., with inhibitors of FAP or CXCR4) to reduce pro-tumorigenic signaling. Anti-angiogenic therapy (e.g., bevacizumab) to normalize tumor vasculature and improve drug delivery [65] Hypoxia-activated prodrugs or agents that re-oxygenate the TME [56, 65]
Immunotherapy augmentation	Combining immune checkpoint inhibitors with other agents to overcome primary or acquired resistance to immunotherapy.	Combining PD-1/PD-L1 inhibitors with CTLA-4 inhibitors, chemotherapy, radiotherapy, or novel agents (e.g., oncolytic viruses, bispecific antibodies) to enhance T-cell infiltration and activity. Targeting immunosuppressive cells within the TME (e.g., MDSCs, Tregs) [65]
Adaptive therapy	Adjusting treatment dose or schedule based on tumor response and resistance evolution, aiming to manage rather than eradicate the tumor.	Metronomic chemotherapy (low-dose, frequent administration) to target endothelial cells and reduce toxicity. Treatment holidays or intermittent dosing to allow sensitive clones to rebound, out-competing resistant ones and re-sensitizing the tumor to treatment. Utilizing mathematical modeling to predict optimal dosing strategies [65]
Biomarker-driven approaches	Using molecular profiling (genomic, transcriptomic, proteomic) to identify specific resistance mechanisms in individual patients and guide personalized treatment decisions.	Liquid biopsies (circulating tumor DNA/ctDNA) for real-time monitoring of resistance mutations and clonal evolution. Next-Generation Sequencing (NGS) to identify actionable mutations and predict response or resistance. Proteomic analysis to identify pathway activation or protein expression changes indicating resistance [65]

Advances in genomic profiling technologies

Genomic, transcriptomic and proteomic platforms are complementary diagnostic tools rather than interchangeable alternatives: each assay answers a different clinical question and together they create a more actionable picture of resistance. Targeted NGS panels and whole-exome sequencing (WES) efficiently detect coding somatic mutations and can rapidly identify many clinically actionable variants; whole-genome sequencing (WGS) adds non-coding and structural variant resolution that is sometimes critical for complex rearrangements or copy-number events that drive resistance. Transcriptomics (bulk RNA-seq) reports pathway activity and fusion transcripts, while single-cell RNA/DNA profiling resolves intratumoral subclones and cellular states that are invisible to bulk assays. Finally, minimally-invasive circulating tumor DNA (ctDNA) and circulating cell-free DNA allow longitudinal, repeatable sampling to monitor temporal evolution under therapy. Using these platforms together, for example, a diagnostic WES or panel at baseline, single-cell or spatial profiling for focal lesion assessment, and ctDNA for serial monitoring, is a pragmatic clinical algorithm for detecting emergent resistance and directing adaptive therapy [66–70].

Limitations and practical caveats of current genomic and multi-omic platforms

Despite their power, genomic and multi-omic platforms have important limitations that must be acknowledged when translating data into treatment decisions. First, spatial heterogeneity is widespread in MM: single-site bone marrow aspirates may miss clones that seed focal lesions or extramedullary sites, producing false reassurance about the absence of resistance mutations. Imaging-guided or multi-site sampling and evolving spatial transcriptomics / single-cell workflows can partially mitigate this but are not yet routine [71–73]. Second, temporal evolution under therapy can rapidly change the relevance of a baseline genomic snapshot: in precursor states such as MGUS or SMM small malignant clones may already carry driver events but lack the structural complexity observed in late-line MM, and conversely, repeated therapy in relapsed patients can generate new subclonal mutations, structural variants and microenvironmental adaptations. Serial sampling (repeat marrow, lesion-directed biopsies and/or ctDNA) is therefore essential both to (a) identify early genomic features that predict progression (important for SMM → early intervention strategies) and (b) detect emergent resistance mechanisms in relapsed/refractory disease. Platform choice should be stage-aware: targeted panels or high-sensitivity ctDNA may suffice in many early relapse settings, whereas whole-genome or single-cell approaches are often needed to resolve the polyclonal complexity of heavily pretreated, triple-class refractory disease [6, 74]. Third, sampling bias and sensitivity differ across platforms: WES/WGS require sufficient tumor fraction in marrow DNA, single-cell assays can be influenced by cell recovery and selection, and ctDNA sensitivity falls when tumor burden is low; each method therefore has blind spots that must be recognized during clinical interpretation [70, 75]. Fourth, genotype ≠ function: many genomic variants do not translate into altered protein abundance or activity because of post-transcriptional regulation, protein stability and post-translational modification; proteomics and functional assays are often necessary to confirm that a detected mutation is the driver of resistance. Integrative proteogenomics thus strengthens causal inference and prioritization of therapeutic targets [76, 77]. Finally, cost, turnaround time, and variable clinical utility remain real barriers: WGS is increasingly informative but expensive; single-cell and spatial assays are powerful investigational tools but not yet standardized for routine clinical decision-making. These constraints mean genomic data should guide, but not unilaterally determine, therapy selection until prospective trials validate genomics-driven adaptive strategies [67, 78].

Cellular mechanisms underlying drug resistance in MM

Drug resistance in MM cells is a complex phenotype emerging from various cellular adaptations. These adaptations encompass altered drug pharmacokinetics, dysregulation of intrinsic survival pathways, and profound influences from the microenvironment. Understanding these cellular mechanisms is fundamental to counteracting therapeutic failure [79].

Alterations in drug transport and efflux systems

ATP-binding cassette (ABC) transporters including P-glycoprotein (P-gp/ABCB1) can mediate drug efflux and have been implicated in multidrug resistance in MM in preclinical studies, particularly with anthracycline substrates (e.g., doxorubicin). However, their clinical relevance in the era of proteasome inhibitors, IMiDs and monoclonal antibodies is more limited, and P-gp–mediated resistance tends to be most important when older cytotoxic agents (notably anthracyclines) are used or after selection by such agents. Thus, while detection of transporter upregulation can explain cross-resistance phenotypes and remains biologically informative, P-gp is not the dominant resistance mechanism for many current frontline and backbone therapies [80–82]. Clinically, transporter overexpression is rarely a lone actionable marker in MM; however, its detection (e.g., gene expression or copy-number gain) can explain cross-resistance phenotypes and supports strategies that avoid substrates of ABC pumps or add agents that bypass efflux mechanisms [83].

Changes in apoptotic and survival pathways

MM cells acquire drug resistance by both (A) genetic/constitutive drivers that directly block apoptosis (for example, BCL-2 family overexpression or TP53 loss) and (B) adaptive stress responses that are induced by therapy (for example, the unfolded protein response (UPR) and autophagy). The two categories have different therapeutic implications: true resistance drivers (e.g., MCL-1 upregulation or CRBN truncation) are often fixed features of resistant clones and predict failure of the cognate drug class, whereas adaptive responses (e.g., transient activation of PERK–eIF2α signaling or increased autophagic flux after proteasome inhibition) are dynamic, therapy-induced dependencies that can be exploited therapeutically. Distinguishing these is critical: drivers argue for switching drug class or using direct inhibitors of the driver, while adaptive dependencies favor rational combinations that block the survival program induced by treatment (for example, pairing proteasome inhibitors with autophagy or UPR inhibitors) [84, 85]. For clarity, examples help. Resistance drivers include (i) genetic alterations that directly weaken drug binding or apoptotic execution (e.g., PSMB5 or PSMG2 mutations that reduce PI binding; CRBN truncations that cause IMiD refractoriness), and (ii) stable changes in anti-apoptotic balance such as persistent MCL-1 or BCL-XL overexpression that render cells intrinsically apoptotic-resistant. Adaptive dependencies are pathways upregulated after drug exposure (and therefore detectable by time-matched transcriptome/proteome assays), such as IRE1–XBP1 activation, PERK-mediated translational attenuation, and increased autophagic flux following proteasome blockade. These adaptive programs often create targetable liabilities, e.g., combining PIs with autophagy inhibitors or targeting MCL-1 in MCL-1-addicted clones can convert an adaptive rescue into a therapeutic vulnerability [84, 86, 87].

The BCL-2 family of proteins comprises both pro-apoptotic (e.g., Bax, Bak, Bim) and anti-apoptotic (e.g., BCL-2, BCL-XL, Mcl-1) members that regulate mitochondrial outer membrane permeabilization, a critical step in apoptosis [88]. In MM, overexpression of anti-apoptotic proteins, particularly Mcl-1, is a significant mechanism of drug resistance, including resistance to bortezomib and microenvironment-mediated protection. Strategies combining Chk1 and MEK1/2 inhibitors effectively overcome Mcl-1-dependent resistance by downregulating Mcl-1, upregulating Bim, and restoring Bak/Bax activation [89, 90]. Additionally, the expression of Homer1b/c, a pro-apoptotic protein, is diminished in MM cells, and its re-expression sensitizes cells to apoptosis and affects cell adhesion-mediated drug resistance [91]. Increased expression of BCL-2 by mesenchymal stem cells (MSCs) via the CXCL13-mediated signaling pathway also contributes to enhanced MM cell proliferation and drug resistance. Conversely, a tumor suppressor gene, WW domain-containing oxidoreductase (WWOX), induces apoptosis in U266 MM cells by activating the intrinsic apoptotic pathway, suggesting its potential role in restoring drug sensitivity [92].

UPR is a cellular stress response mechanism activated when misfolded proteins accumulate in the endoplasmic reticulum (ER) [93]. MM cells, characterized by high immunoglobulin production, inherently experience ER stress, making them particularly sensitive to agents that exacerbate this stress, such as proteasome inhibitors [94]. However, MM cells can adapt to chronic ER stress, developing mechanisms that contribute to drug resistance. Studies indicate that endoplasmic reticulum stress (ERS) activation can promote autophagy and apoptosis, while inhibiting proliferation through the PI3K/Akt/mTOR signaling pathway [62, 95]. Conversely, lower levels of glucose-regulated protein 78 (GRP78) and GRP94, key UPR chaperones, are found in resistant MM patients, suggesting that a dysregulated or insufficient UPR can contribute to resistance. The adaptive changes in MM cell metabolism, including those related to protein folding and degradation, can form the basis of resistance to proteasome inhibitors [96, 97].

Tumor microenvironment contributions

The bone marrow microenvironment provides a supportive niche for MM cells, shielding them from therapeutic agents and promoting survival and proliferation. Expanding on this, recent work emphasizes that the marrow niche is not merely a passive shelter but a dynamic, stage-dependent partner in myelomagenesis: even at the MGUS and smoldering stages, immune dysfunction, stromal activation and altered bone remodeling create permissive niches that favor clonal persistence and stepwise malignant progression; over time these niche alterations (chronic cytokine exposure, hypoxia, extracellular vesicle exchange and remodeling of adhesion cues) both select for pre-existing resistant subclones and actively promote genomic instability and epigenetic reprogramming in plasma cells, accelerating the transition to frank MM [32]. Beyond passive protection, accumulating evidence indicates that microenvironmental cues actively shape tumor genomics and epigenetics: bone-marrow stromal contact and soluble signals can induce chromatin remodeling in myeloma cells, and hypoxic or inflammatory niche conditions impair DNA repair pathways and increase mutation acquisition, processes that accelerate clonal diversification and drive resistance [98–100]. Interactions between MM cells and various components of this microenvironment, including stromal cells, extracellular matrix, and secreted cytokines, induce multiple pro-survival and drug-resistance pathways [101]. Bone marrow stromal cells (BMSCs) play a central part in MM pathogenesis and drug resistance. BMSCs secrete growth factors, cytokines, and extracellular vesicles, which protect MM cells from chemotherapy [102]. Exosomes derived from BMSCs increase MM cell growth, migration, survival, and induce resistance to agents like bortezomib, influencing pathways such as JNK, p38, p53, and Akt. This BMSC-induced drug resistance is often termed CAM-DR [103]. For example, Eph receptor A4 (EphA4) promotes MM cell adhesion and CAM-DR by enhancing Akt phosphorylation [104]. Similarly, annexin A7 (ANXA7) promotes proliferation and CAM-DR in MM cells by upregulating cell division cycle 5-like (CDC5L), influencing adhesion markers like CD44, ICAM1, and VCAM1 [105]. The long non-coding RNA PCAT-1 also contributes to CAM-DR by enhancing cell growth and inhibiting apoptosis via the p38 and JNK MAPK pathways. MSCs from MM patients secrete higher levels of CXCL13, which promotes MM cell invasion, proliferation, and bortezomib resistance through the CXCL13-CXCR5 axis, upregulating BTK, NF-κB, BCL-2, and MDR-1. The “activated phenotype” of the MM microenvironment is therefore crucial for supporting plasma cell survival and drug resistance [32, 106].

Cytokines and growth factors secreted within the bone marrow microenvironment establish a protective niche for MM cells, fostering their survival and contributing to drug resistance [107]. Interleukin-6 (IL-6) and Insulin-like Growth Factor-1 (IGF-1) are prominent examples that promote MM cell proliferation and reduce sensitivity to various anti-cancer agents by activating pro-survival pathways, including the upregulation of Mcl-1 [108]. Reelin, an extracellular matrix protein, enhances MM cell adhesion to fibronectin and protects against drug-induced apoptosis by activating integrin-β1 and the Src/Syk/STAT3 and Akt pathways. These signaling cascades contribute to the evasion of apoptosis and the maintenance of a drug-resistant phenotype [109]. Furthermore, the activated microenvironment induces stress-managing pathways and transcriptional rewiring, further contributing to drug resistance. Endogenous hydrogen sulfide (H₂S) also supports the survival of MM cells, including those resistant to bortezomib, primarily through cystathionine-β-synthase (CBS) activity, suggesting that CBS inhibitors could treat bortezomib-resistant MM [110]. Translational prioritization: among many BMME factors, five mechanisms deserve urgent prioritization for therapeutic development and biomarker evaluation. First, the IL-6 / STAT3 axis (central survival signal for MM) has direct clinical relevance: IL-6 blockade (siltuximab) has been evaluated in randomized studies combined with bortezomib and showed biologic activity, illustrating a pathway to combination strategies [111–113]. Second, the CXCL12/CXCR4 axis mediates marrow retention and chemoprotection; pharmacologic CXCR4 disruption (plerixafor) can mobilize malignant cells and has been tested as a chemosensitization strategy in hematologic malignancies [114]. Third, integrin (VLA-4)–VCAM-1 adhesion underlies CAM-DR and is actionable (imaging and preclinical targeting data support further translational development) [115, 116]. Fourth, niche hypoxia/HIF signaling promotes DNA-repair dysregulation and mutagenesis, linking metabolic microenvironments to genomic trajectories and suggesting hypoxia-targeted therapeutics as adjuncts [99, 117]. Fifth, stromal extracellular vesicles and organelle transfer (including mitochondrial transfer) reprogram tumor metabolism and drug tolerance and are emerging as targetable intercellular mechanisms [118]. Together, these prioritized targets form the most immediate translational bridge between the microenvironment and genomics-informed therapy: each has a plausible clinical strategy (drug, chemosensitization, or biomarker) that can be tested in combination with genomic profiling to prevent or overcome resistance.

To enhance the translational value of this section, we prioritize microenvironmental mechanisms by therapeutic relevance: (1) IL-6 → STAT3 signaling, a central driver of survival and PI/chemotherapy resistance and a clinically tested target (anti-IL-6 antibody siltuximab has undergone randomized evaluation paired with bortezomib) [111, 112]. (2) CXCL12 (SDF-1)/CXCR4 axis, a critical retention/chemoprotection pathway whose pharmacologic disruption (e.g., plerixafor) mobilizes malignant cells and can chemosensitize them in clinical studies [114]. (3) VLA-4/VCAM-1 and integrin-mediated adhesion, major mediators of CAM-DR with compelling preclinical data and emergent imaging/targeting strategies [115, 116]. (4) Hypoxia/HIF-driven signalling, shapes DNA repair, mutational processes and immune suppression, and therefore contributes to genomic instability and resistance trajectories [99, 117]. (5) Extracellular vesicle/mitochondrial transfer from stromal cells, a non-canonical mechanism that alters myeloma metabolism and drug tolerance and is targetable in the preclinical setting [118].

Prioritizing mechanisms in this way (IL-6/STAT3, CXCL12/CXCR4, integrin adhesion, hypoxia, exosome/mitochondrial transfer) clarifies which microenvironmental interactions have the strongest existing translational hooks (drugs or clinical strategies) and highlights where adjunctive trials or biomarker studies make the most sense clinically. Exosomes and other extracellular vesicles (EVs) are increasingly recognized as mediators of therapy resistance in multiple myeloma. Bone-marrow stromal cell–derived EVs transfer proteins, microRNAs and other bioactive cargo that activate PI3K/AKT and MAPK signalling in plasma cells, thereby reducing sensitivity to bortezomib and other agents and modulating immune and stromal compartments. Conversely, myeloma cell–derived EVs (“chemoexosomes”) deliver stress-response proteins and nucleic acids that reprogram niche cells to create a protective microenvironment. By shuttling functional RNAs (for example, miRNAs and mRNAs) and proteins between stromal and myeloma cells, EVs produce durable phenotypic changes that promote drug resistance and represent promising biomarker and therapeutic targets [119–122]. Figure 2 illustrates the integrated mechanisms of multidrug resistance in MM, including efflux pumps, survival pathways, and microenvironmental protection.

Fig. 2.

Schematic illustration showing how MM cells develop multidrug resistance through enhanced drug efflux (P-gp, ABCG2), activation of survival pathways (PI3K/AKT, Mcl-1, BCL-2), and protection from the bone marrow microenvironment via cytokines (IL-6, IGF-1, CXCL13). ER Endoplasmic reticulum, PI3K Phosphoinositide 3-kinase, RAF Rapidly accelerated fibrosarcoma, MEK Mitogen-activated protein kinase, ERK Extracellular signal-regulated kinase, JAK Janus tyrosine kinase, WWOX WW domain-containing oxidoreductase, BMSC Bone marrow stem cell

Molecular pathways driving drug resistance

Molecular alterations within MM cells are fundamental drivers of drug resistance. These include specific genetic mutations, chromosomal abnormalities, and dynamic epigenetic modifications that reshape cellular responses to therapeutic agents. The adaptive evolution of proteasome pathways also plays a role, especially in resistance to proteasome inhibitors [123].

Genetic mutations and chromosomal aberrations

MM is characterized by a high degree of genetic instability, manifesting as recurrent chromosomal abnormalities and somatic gene mutations that influence disease progression and drug sensitivity. These genetic events are often associated with intrinsic or acquired resistance to a variety of therapeutic agents [124]. Mutations in the TP53 tumor suppressor gene are frequently observed in advanced and drug-resistant MM, often correlating with poor prognosis. Loss of functional p53 compromises the cell’s ability to undergo apoptosis in response to DNA damage or other cellular stresses, thereby conferring resistance to many chemotherapeutic agents [125]. The RAS pathway, encompassing KRAS, NRAS, and BRAF genes, is also frequently mutated in MM. Activating mutations in these genes lead to constitutive activation of downstream signaling pathways, such as the mitogen-activated protein kinase (MAPK) pathway, promoting cell proliferation and survival [126]. For example, BRAF V600E mutations are found in approximately 10% of MM patients, and while targeted BRAF/MEK inhibitors can be effective, resistance frequently develops [127]. The acquisition of mutations in genes like CIC (capicua transcriptional repressor) can mediate acquired resistance to BRAF/MEK inhibitors in MM, demonstrating the complex genetic adaptations that occur [128]. A study identified KRAS G12C and ATM T1985I mutations in a patient with multi-drug refractory extramedullary disease, both impacting MEK signaling and suggesting MEK as a therapeutic target [129]. Recent prospective clinical data now show that biology-driven BRAF/MEK combinations can produce high response rates in selected, heavily pretreated patients with activating BRAF V600E mutations. In a phase-2 trial reported in Blood (Apr 6, 2023), encorafenib (450 mg once daily) plus binimetinib (45 mg twice daily) produced an overall response rate of 83.3% (10/12 evaluable patients) in relapsed/refractory BRAF V600E-mutated multiple myeloma, with a median PFS of ~ 5.6 months and encouraging 2-year OS signals. This trial provides direct clinical support for genomic profiling to identify BRAF V600E as an actionable lesion in MM and suggests that MAPK pathway-directed therapy is a practical precision-medicine approach in selected cases [130].

Recurrent chromosomal translocations are hallmarks of MM and are associated with distinct molecular subgroups and varying prognoses. Translocations involving the immunoglobulin heavy chain (IGH) locus on chromosome 14q32 are particularly significant [131]. In addition to IGH translocations, numerical and structural abnormalities of chromosome 1 are among the most frequent and clinically important events in MM. Gain and amplification of 1q (1q21+) occur in ~ 30–40% of newly diagnosed patients (and more frequently in relapsed/proliferative disease), and increasing copy number (gain → amplification) is associated with progressively worse progression-free and overall survival. Deletions on 1p (del(1p)) are also recurrent and have been linked to adverse prognosis and more aggressive disease biology. These 1q gains/amps and 1p losses commonly arise as secondary events that drive clonal evolution and treatment resistance, and therefore should be explicitly considered in genomic risk stratification [132–134]. Common translocations, such as t(11;14), t(4;14), t(14;16), and t(14;20), lead to the dysregulation of oncogenes and are implicated in diverse clinical outcomes and drug responses. For instance, the high-risk marker t(14;16) was identified in a patient with highly disturbed genome and multi-drug refractory disease. These translocations can influence the expression of genes involved in cell cycle regulation, apoptosis, and drug metabolism, thereby contributing to resistance [135].

Epigenetic modifications impacting drug response

Epigenetic modifications, including DNA methylation and histone modifications, represent heritable changes in gene expression without altering the underlying DNA sequence. These modifications exert profound influence on gene transcription and can reprogram MM cells to become drug-resistant [136]. Aberrant DNA methylation patterns, particularly hypermethylation of tumor suppressor genes, can silence critical genes involved in drug sensitivity or apoptosis, contributing to resistance. Similarly, histone modifications, such as acetylation and deacetylation, modulate chromatin structure and gene accessibility [137]. For example, inhibitors of histone deacetylase (HDAC) are emerging as therapeutic agents that can reverse certain resistance mechanisms. The regulation of glucocorticoid receptor (GR) expression, mediated by transcriptional and post-transcriptional mechanisms, involves epigenetic control. Resistance to glucocorticoids can stem from a block to transcriptional elongation within the NR3C1 gene (encoding GR), leading to decreased GR levels in resistant cell lines [138]. This transcriptional block, rather than gene mutation or promoter methylation, reduces GR expression. Another example is SIRT2, a histone deacetylase, which is expressed at higher levels in relapsed AML patients and contributes to drug resistance by modulating MRP1 levels and ERK1/2 signaling. These epigenetic changes offer opportunities for targeted therapeutic intervention [139].

Proteasome pathway adaptations

The proteasome is a multi-catalytic protein complex responsible for degrading ubiquitinated proteins, a crucial process for maintaining cellular homeostasis. Proteasome inhibitors (PIs) are cornerstone treatments in MM, inducing ER stress and apoptosis by blocking protein degradation [140]. However, MM cells can develop resistance to these agents through various adaptive mechanisms. Resistance to proteasome inhibitors remains clinically important. Recurrent point mutations in PSMB5 (the β5 catalytic subunit) and in proteasome-assembly factors have been experimentally linked to reduced bortezomib binding and may predict cross-resistance or preserved sensitivity to second-generation PIs; reporting such variants on diagnostic NGS panels can therefore directly inform retreatment choices [26]. For example, carfilzomib resistance in LP-1 cells is associated with coordinated activation of proteostasis programs, Nrf2 transcriptional activation, metabolic rewiring (increased fatty-acid oxidation), and pro-survival autophagy mediated downstream of PERK–eIF2α and SQSTM1/p62. Importantly, these features are adaptive and therefore actionable: several preclinical studies have shown that pharmacologic blockade of autophagy (chloroquine/hydroxychloroquine or specific autophagy inhibitors) or inhibition of UPR sensors (IRE1α or PERK inhibitors) sensitizes MM cells to proteasome inhibition. Conversely, direct genetic or pharmacologic targeting of anti-apoptotic drivers (for example, MCL-1 inhibitors such as S63845 or the newer clinical-stage MCL-1 antagonists) addresses a fixed driver rather than an induced dependency. These mechanistic distinctions support two rational strategies: (i) when a driver (e.g., MCL-1 overexpression, CRBN truncation, PSMB5 mutation) is present, change class or target the driver; (ii) when adaptive UPR/autophagy signatures emerge after therapy, combine the therapy with UPR/autophagy inhibitors to convert a survival response into cytotoxicity [84, 141, 142]. Upregulation of glycolysis and other metabolic adaptations are also linked to bortezomib resistance [143]. The hexosamine biosynthetic pathway (HBP) and increased mitochondrial fitness are implicated in bortezomib resistance, suggesting HBP as a therapeutic target. Additionally, activation of the SGK1/NF-κB pathway correlates with low sensitivity to bortezomib and ixazomib [144]. A truncating mutation in Cereblon (CRBN) and a point mutation in proteasome subunit G2 (PSMG2) were identified in a patient with bortezomib-refractory disease, representing direct molecular explanations for resistance to IMiDs and PIs, respectively. Snail1 also induces bortezomib resistance by upregulating MDR1 and downregulating P53 [144, 145].

A unified, clinically-oriented framework linking molecular pathways to therapy

Therapy selection in relapsed or refractory MM benefits from a framework that maps (A) molecular mechanism → (B) diagnostic genomics/proteomics → (C) clinical implication → (D) actionable strategy, and explicitly recognizes the bone-marrow microenvironment as an active modifier of the molecular mechanism step (A), since stromal cytokines, adhesion receptors, hypoxia and extracellular vesicles can alter genomic states, and functional pathway activation that the diagnostic assays measure [98, 99]. Examples: (a) Proteasome-pathway alteration (e.g., PSMB5 / PSMG2 variants) → detectable by targeted DNA panels or WES → predicts PI cross-resistance or retained sensitivity to alternative PIs → consider switching drug class or using non-PI combinations, or selecting next-generation PIs with activity against the altered subunit [26]. (b) CRBN truncation or loss-of-function mutations → detected by NGS and confirmed by immunohistochemistry/protein assays → explains IMiD refractoriness → favors non-IMiD immunotherapies (anti-CD38, BCMA-directed) or clinical trials of cereblon-independent approaches [146]. (c) Nrf2 / metabolic adaptations and increased mitochondrial fitness → identified by combined transcriptomic/proteomic signatures → indicate proteostasis-based resistance to PIs → supports combining PIs with metabolic inhibitors or agents that target mitochondrial stress response [143, 147]. This framework emphasizes that genomic/proteomic readouts are not ends in themselves, but decision tools that map directly to therapeutic choices, the central point this review advances.

Practical rule for clinicians and trial design. To guide rational therapy, we recommend explicitly classifying molecular readouts into (i) Resistance drivers: stable, often genomic/protein alterations that predict primary or persistent resistance (test: DNA sequencing, protein expression by IHC); and (ii) Adaptive dependencies: therapy-induced changes detectable by on-treatment transcriptome/proteome/functional assays (test: serial RNA-seq, phospho-proteomics, or ctDNA-guided timing). Clinical action: drivers → avoid or override with class-switching or direct driver antagonists; adaptive dependencies → combine the inducing therapy with inhibitors of the adaptive pathway (for example, PI + autophagy/UPR inhibitor; IMiD refractory due to CRBN loss → consider cereblon-independent immune therapies). Adding this explicit classification into trial protocols will improve the interpretability and utility of genomic profiling for adaptive combination strategies [85, 148].

Genomic profiling in elucidating resistance mechanisms

Genomic profiling provides added value beyond mechanistic description in three practical ways. First, diagnostic attribution: specific mutations can explain treatment failure (for example, PSMB5/PSMG2 variants, which have been experimentally and clinically linked to proteasome-inhibitor resistance and may predict cross-resistance to some PIs, and CRBN truncations or loss-of-function alteration associated with acquired IMiD refractoriness and therefore predict lack of benefit from further IMiD retreatment) [26, 149]. To aid clinical translation, the following pragmatic classification applies to commonly reported genomic findings in MM: (a) Prognostic markers, alterations strongly associated with overall outcome but not necessarily directing a specific drug choice (for example, del(17p)/TP53 loss, t(4;14), and t(14;16) which are repeatedly associated with high-risk disease and inferior survival) [150, 151]. (b) Predictive markers, genomic events that can inform a specific therapy (for example, activating BRAF V600E mutations which can predict response to BRAF/MEK inhibition in selected refractory patients; and acquired CRBN pathway alterations or PSMB5/PSMG2 variants that predict poor response to IMiDs or some PIs, respectively) [26, 130, 152, 153]. (c) Exploratory/research markers, genomic, transcriptomic or single-cell signatures that currently improve biological understanding or trial selection but are not yet standard for routine treatment decisions (examples: many complex transcriptomic/proteomic adaptive signatures, tumor mutational burden in MM, or single-cell plasticity states). Clinical implementation of many of these exploratory readouts requires further validation [154].

Second, therapeutic triage: identified alterations map to concrete choices, avoiding IMiD retreatment when CRBN loss is present, preferring non-PI classes or second-generation PIs when PSMB5 variants are detected, or prioritizing BCMA-directed approaches when other options are exhausted. Third, dynamic monitoring: serial ctDNA or circulating-cell sequencing enables earlier detection of emergent resistant clones and permits adaptive treatment changes before overt clinical relapse. Together these applications show that genomic readouts can function as decision tools, not only as descriptive evidence [26, 27, 155, 156]. Importantly, longitudinal and single-cell genomic/transcriptomic approaches do more than catalogue mutations, they can directly measure clonal competition, reveal reversible phenotypic plasticity, and identify adaptive transcriptional reprogramming that occurs within surviving clones under therapy. Recent single-cell studies in myeloma show how minor subclones or transcriptionally plastic states survive induction, adapt their transcriptional programs (for example upregulation of stress, metabolic, or immune-evasion pathways), and later re-expand as dominant, drug-resistant populations. Incorporating these dynamic readouts into clinical monitoring enables earlier detection of adaptive resistance and supports adaptive therapeutic strategies (e.g., sequential or combination regimens designed to limit selective sweeps) [63, 140, 157]. WES has confirmed the genetic heterogeneity of MM, revealing numerous gene mutations without a single unifying alteration, but frequently identifying changes in the MAPK pathway [158]. NGS can detect mutations linked to drug resistance, such as those in CRBN, PSMG2, and NR3C1, which can explain resistance to IMiDs, proteasome inhibitors, and corticosteroids, respectively [159]. Furthermore, NGS facilitates the identification of mutations in actionable pathways like PI3K/AKT and MEK, even in heavily pretreated patients, opening avenues for targeted therapies [160]. The ability of NGS to detail specific oncogenic pathways associated with drug resistance, such as TNFα, EGFR, IFNα, IFNγ, hypoxia, STAT3, and MYC in melanoma, illustrates its predictive power [161].

MM is characterized by significant clonal evolution and intra-tumor heterogeneity, where multiple genetically distinct subclones coexist and dynamically shift under selective pressure from therapy [162]. Genomic profiling has revealed that all MM patients exhibit subclonality, even involving supposed driver genes like KRAS, NRAS, and BRAF. This clonal diversity implies that some subclones may possess intrinsic resistance mechanisms, while others acquire them during treatment, ultimately leading to relapse [163]. Studies show that MM consists of hierarchically organized, clonally related subpopulations, with a low-frequency CD19⁻CD138⁻ “Pre-PC” population exhibiting up to 300-fold greater drug resistance than the dominant CD19⁻CD138⁺ plasma cells [164, 165]. This reversible, bidirectional phenotypic transition of myeloma-propagating cells directly links to clinical drug resistance and has implications for minimal residual disease assessment. Understanding this dynamic clonal selection is critical for developing strategies that target resistant subclones and prevent therapeutic escape [166].

Beyond single-omic analyses, integrative omics approaches combine data from genomics, transcriptomics, proteomics, and metabolomics to provide a holistic view of drug resistance mechanisms. This multi-layered approach helps to uncover complex interactions and compensatory pathways that contribute to resistance phenotypes [167]. For instance, proteomic analysis in methotrexate-resistant breast cancer cells identified 17 differentially expressed proteins, including nucleophosmin (NPM), which, upon knockdown, resensitized cells to methotrexate [168]. Similarly, transcriptomic and proteomic analyses in MM have revealed metabolic reprogramming, such as increased hexosamine biosynthetic pathway activity and mitochondrial fitness, in bortezomib-resistant cells. These integrated datasets illuminate the intricate adaptive responses of MM cells to therapy, offering new targets for intervention that might not be apparent from genomic data alone [169].

Personalized treatment strategies informed by genomic data

The detailed insights gleaned from genomic profiling are increasingly being translated into personalized treatment strategies for MM. This paradigm shift involves tailoring therapeutic approaches based on the unique molecular characteristics of an individual patient’s tumor, aiming to overcome drug resistance and improve outcomes [170]. Genomic data permits sophisticated risk stratification and molecular subgrouping of MM patients, moving beyond traditional clinical parameters. Cytogenetic data, particularly concerning ploidy and the presence of 14q32 translocations, have historically been used for subgrouping. However, advanced genomic analysis now identifies specific gene mutations and copy number abnormalities that predict prognosis and treatment response more accurately [171]. For example, detection of high-risk markers like t(14;16) or mutations in TP53 can guide clinicians toward more intensive or alternative therapies. Moreover, distinguishing between hyperdiploid and non-hyperdiploid MM, as identified by genetic profiling, offers a more refined approach to patient management [172].

Identification of actionable genomic alterations enables targeted, evidence-based treatment decisions. For example, del(17p) and TP53 inactivation are well-validated adverse prognostic markers associated with shorter progression-free and overall survival; their presence should inform risk-adapted monitoring and consideration of intensified therapy. Predictive examples include BRAF V600E, an actionable alteration found in a minority of patients that can direct BRAF/MEK inhibitor therapy in selected relapsed/refractory cases; CRBN truncations or other deleterious CRBN-pathway alterations, which predict poor response to IMiDs and therefore generally argue against IMiD retreatment; and BCMA (B-cell maturation antigen) is highly and broadly expressed on malignant plasma cells, making it an attractive immunotherapy target. However, BCMA-directed therapies, including CAR-T cells, bispecific antibodies, and antibody–drug conjugates, have so far been used mainly in relapsed/refractory multiple myeloma (including triple- or penta-refractory cases), rather than as routine first-line treatments [173]. Importantly, most pivotal trials and regulatory approvals did not require routine pre-treatment testing of tumour BCMA expression (for example by IHC), because BCMA is generally ubiquitously expressed in MM; therefore, routine genomic or proteomic BCMA testing is not currently required to select patients for these therapies. Finally, antigen loss or reduced BCMA expression can occur after prior BCMA-directed treatment and may influence retreatment strategies [130, 152, 153, 174–176]. BCMA is a surface protein highly expressed on malignant plasma cells and has emerged as a promising target for MM therapy. BCMA-directed therapies, such as antibody-drug conjugates (ADCs) and chimeric antigen receptor (CAR) T-cell therapies, leverage this specific expression to selectively eliminate MM cells [177]. These therapies represent a class of personalized medicine, offering potent anti-myeloma activity, particularly in relapsed/refractory settings. While not directly a genomic alteration, the high and consistent expression of BCMA on MM cells, which can be confirmed by genomic or proteomic profiling, makes it an effective target for precision medicine approaches [137]. IMiDs, such as thalidomide, lenalidomide, and pomalidomide, are critical components of MM treatment. However, resistance to IMiDs is common [178]. Genomic insights reveal mechanisms of resistance, such as acquired truncating mutations in CRBN, the primary target of IMiDs. This specific mutation explains the observed IMiD-refractory disease in some patients [179]. Genomic profiling can thus inform the selection of alternative therapies when such resistance mutations are detected. Furthermore, strategies designed to overcome IMiD resistance might involve combination approaches that bypass the CRBN pathway or target downstream effects of IMiD resistance [180]. For instance, in vitro, the combination of Chk1 and MEK1/2 inhibitors effectively overcomes Mcl-1-related resistance, which can be a consequence of microenvironmental protection, a factor that IMiDs also aim to disrupt [181].

Combination therapies and adaptive treatment models

Genomic profiling can indeed enable rational combinations, but the clinical translation of multi-agent strategies has been uneven and requires explicit discussion of feasibility, toxicity, and patient selection. Several high-profile combination approaches have delivered disappointing or inconclusive results (or unacceptable toxicity) when moved from preclinical promise into randomized trials, underscoring the need for caution. Notable examples include PD-1 checkpoint inhibitors combined with IMiDs, where randomized studies were suspended after excess mortality signals (KEYNOTE-183, KEYNOTE-185), highlighting unexpected immune-related harms when agents are combined without robust safety data [182]. Other combinations demonstrate efficacy but substantial tolerability challenges (for example, panobinostat-based regimens produce frequent grade 3–4 hematologic and gastrointestinal toxicities that limit long-term use) [183, 184]. Targeted immunotherapies (BCMA CAR-T and bispecific antibodies) show high response rates but carry predictable and sometimes severe toxicities, cytokine release syndrome (CRS), immune effector neurotoxicity syndrome (ICANS), and late neurologic events, which require specialized centers and careful patient selection [185, 186]. Finally, antibody–drug conjugates such as belantamab mafodotin can produce dose-limiting ocular toxicity (keratopathy) that forced REMS monitoring and changes in trial procedures, and in real-world series led to treatment discontinuations despite responses [187]. For instance, if a patient’s tumor harbors both BRAF V600E and a CIC mutation, a combination of BRAF/MEK inhibitors with an agent targeting the downstream effects of CIC downregulation might be considered. Similarly, an SGK1 inhibitor enhances the cytotoxic effects of bortezomib and ixazomib in MM cells with activated SGK1/NF-κB pathways, suggesting a potential combination strategy [188]. Adaptive treatment models involve dynamic monitoring of clonal evolution and resistance mechanisms throughout the course of therapy using liquid biopsies or repeat tumor sampling. This allows for timely adjustment of treatment regimens to counteract emerging resistance, effectively personalizing therapy in real-time [189].

Comparative assessment of resistance mechanisms across studies

A comparative analysis of drug resistance mechanisms in MM highlights both common themes and agent-specific adaptations. Overexpression of drug efflux pumps, such as P-gp and ABCG2, consistently emerges as a significant contributor to multidrug resistance across various chemotherapeutic agents [190]. The bone marrow microenvironment, through its stromal cells and secreted factors like CXCL13 and Reelin, creates a protective niche, promoting CAM-DR and activating pro-survival pathways (e.g., PI3K/Akt/mTOR, STAT3) that confer resistance to multiple drug classes, including proteasome inhibitors and immunomodulatory drugs [191]. Alterations in apoptotic pathways, particularly the upregulation of anti-apoptotic proteins like Mcl-1 and dysregulation of p53, are central to the survival of resistant clones [192]. Specific agents also face unique resistance mechanisms. For proteasome inhibitors, metabolic reprogramming, including increased hexosamine biosynthetic pathway activity and fatty acid oxidation, alongside Nrf2 activation, represents a distinct adaptive response. Genomic mutations, such as those in CRBN for IMiD resistance or PSMG2 for proteasome inhibitor resistance, directly compromise drug binding or function [193]. The dynamic nature of clonal evolution means that MM cell populations can shift their predominant resistance mechanisms over time or under selective pressure, leading to the emergence of novel drug-resistant subclones [194].

Current challenges in overcoming drug resistance

Despite significant advancements in MM treatment, overcoming drug resistance remains a formidable challenge. A primary difficulty stems from the inherent genomic instability and profound heterogeneity of MM, leading to clonal evolution and the emergence of diverse resistant subclones [195]. This complexity means that even highly effective initial treatments eventually exert selective pressure, favoring resistant populations. The tumor microenvironment also poses a significant barrier, providing sanctuary and pro-survival signals that shield MM cells from therapeutic agents [196]. Importantly for clinical translation, several combination strategies with strong preclinical rationale have suffered safety or efficacy setbacks in randomized trials, for example, PD-1 inhibitors combined with IMiDs were suspended after excess deaths in the KEYNOTE-183 and KEYNOTE-185 studies. Moreover, agents that demonstrate clear activity (for example, panobinostat, belantamab mafodotin, and BCMA-targeted CAR-T therapies) are limited by substantial toxicity profiles or practical feasibility constraints, which must be carefully weighed against their potential benefits [183, 187, 197]. Importantly, the microenvironment is not only a sanctuary: niche-derived signals and stresses (e.g., chronic cytokine exposure, hypoxia, direct stromal contact) can drive genomic instability and epigenetic remodeling, accelerating the emergence of therapy-resistant clones, a dynamic that argues for integrating microenvironment-targeting strategies alongside genomics-guided therapy [98, 99]. Furthermore, the lack of a suitable procedure to routinely monitor the development of clinical drug resistance in individual patients hinders timely therapeutic adjustments. Minimal residual disease (MRD) cells, often highly resistant, are difficult to detect and eradicate, contributing to relapse [198]. The identification of novel resistance mechanisms, such as specific mutations in CRBN or metabolic adaptations like Nrf2 activation, requires advanced diagnostic tools that are not yet universally implemented in clinical practice [199].

Future research in MM drug resistance centers on several promising trends. First, longitudinal genomic profiling, including liquid biopsies, will enable dynamic monitoring of clonal evolution and the emergence of resistance mutations, facilitating adaptive treatment strategies [200, 201]. Second, the development of novel targeted agents against newly identified resistance pathways, such as inhibitors of Nrf2, SGK1, or specific components of the hexosamine biosynthetic pathway, holds considerable potential [202]. Third, combination therapies designed to simultaneously target multiple resistance mechanisms or overcome microenvironmental protection are gaining traction [203]. This includes combining conventional agents with epigenetic modifiers, microenvironment disruptors, or metabolism-targeting drugs [204]. Fourth, immunotherapeutic approaches, such as CAR T-cells or bispecific antibodies, offer novel strategies to overcome resistance by engaging the patient’s immune system to eliminate malignant plasma cells, often by targeting antigens like BCMA. Finally, detailed functional genomics studies are essential to validate newly identified resistance genes and pathways, translating findings from sequencing into actionable therapeutic targets [205, 206].

Conclusion

Genomic and integrative multi-omic profiling now provide a practical, actionable window into the mechanisms that drive treatment failure in multiple myeloma, but their clinical promise will be realized only if three critical gaps are closed. First, functional validation must move from isolated reports into routine translational pipelines: genome-scale CRISPR / RNAi screens, patient-derived ex vivo drug sensitivity assays, and orthogonal proteogenomic measures are needed to show which variants are true drivers of resistance and which are passenger events. Second, longitudinal and spatial sampling, specifically scheduled ctDNA monitoring complemented by multi-site marrow or lesion sampling when feasible, must be embedded into clinical protocols and trials to detect emergent resistant clones early and to validate ctDNA as an actionable monitoring tool. Recent multi-omics serial studies demonstrate how serial sampling resolves subclone dynamics that single time-points miss. Third, integration of genomic readouts with standardized clinical phenotypes (treatment exposure, depth/duration of response, toxicity, imaging and functional measures) and federated clinical-genomic databases will be required to build robust predictive models and to translate molecular findings into patient-level decisions. Multi-omics subclone studies show the power of linking molecular trajectories to clinical outcome data and illustrate the path forward.

Concretely, we propose the following immediate priorities to convert profiling into clinical benefit: (1) establish prospective trial protocols that require pre-specified longitudinal sampling (baseline WES or targeted panel + periodic ctDNA + at-progression tissue), (2) mandate orthogonal functional tests for high-impact variants (CRISPR screens or standardized ex-vivo drug assays) before declaring a variant actionable, (3) adopt common reporting standards (variant tiering, VAF reporting, minimal dataset for proteogenomics), and (4) create linked clinical-genomic registries and interoperable data models to enable robust phenotype–genotype discovery and validation across centers. Implementation of these steps, backed by cross-institutional consortia and funding for integrated sample collection, will accelerate validation of genomics-guided adaptive strategies and permit meaningful prospective trials. Early data already suggest feasibility and clinical value from such integrated, longitudinal multi-omics designs.

Abbreviations

MM

Multiple myeloma

PIs

Proteasome inhibitors

IMiDs

Immunomodulatory drugs

BMM

Bone marrow microenvironment

CNVs

Copy number variations

ABC

ATP-binding cassette

UPR

Unfolded protein response

CAM-DR

Cell adhesion-mediated drug resistance

WES

Whole exome sequencing

WGS

Whole genome sequencing

RNA-seq

RNA sequencing

ASCT

Chemotherapy with autologous stem cell transplantation

MRP1

Multidrug resistance-associated protein 1

MDR

Multidrug resistance

P-gp

P-glycoprotein

ABCG2

ATP-binding cassette subfamily G member 2

SP

Side population

TAIII

Timosaponin A-III

MSCs

Mesenchymal stem cells

WWOX

WW domain-containing oxidoreductase

ER

Endoplasmic reticulum

ERS

Endoplasmic reticulum stress

GRP78

Glucose-regulated protein 78

BMSCs

Bone marrow stromal cells

EphA4

Eph receptor A4

ANXA7

Annexin A7

CDC5L

Cell division cycle 5-like

IL-6

Interleukin-6

IGF-1

Insulin-like growth factor-1

H₂S

Hydrogen sulfide

CBS

Cystathionine-β-synthase

MAPK

Mitogen-activated protein kinase

IGH

Immunoglobulin heavy chain

HDAC

Histone deacetylase

GR

Glucocorticoid receptor

HBP

Hexosamine biosynthetic pathway

CRBN

Cereblon

NPM

Nucleophosmin

BCMA

B-cell maturation antigen

ADCs

Antibody-drug conjugates

CAR

Chimeric antigen receptor

MRD

Minimal residual disease

Author contributions

Azin Taki wrote the main manuscript text, prepared the figures, and Fengbo Zeng review the manuscript.

Funding

No funding was received.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

Not applicable.

Generative AI

Generative AI tools (instatext, Wordvice, and DeepL) were used only to improve grammar, phrasing, and overall readability (language polishing and structural editing). No generative AI or large language model produced, verified, interpreted, or contributed substantive scientific content, data analysis, interpretations, conclusions, or clinical recommendations in this manuscript. All scientific content and final wording were created, reviewed, and approved by the named human authors, who take full responsibility for the content.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.