Resistance to immunomodulatory drugs in multiple myeloma: the cereblon pathway and beyond

Abstract

Acquired resistance to immunomodulatory drugs (IMiD) remains a significant unmet need in the treatment landscape of multiple myeloma (MM). The cereblon (CRBN) pathway-dependent mechanisms are known to be vital contributors to IMiD resistance; however, they may account for only a small proportion. Recent research has unveiled additional mechanisms of acquired IMiD resistance that are independent of the CRBN pathway. In this review, we provide a comprehensive overview of the existing work on IMiD resistance in MM, focusing specifically on the emerging evidence of CRBN pathway-independent mechanisms. Finally, we discuss the plausible actionable strategies and outlook for IMiD-based therapies moving forward.

Introduction

Immunomodulatory drugs (IMiD) are a major class of drugs that have changed the treatment paradigm for multiple myeloma (MM). Thalidomide, the first-in-class IMiD, was introduced in the late 1950s as a sedative and anti-emetic during pregnancy, but was soon withdrawn due to neuropathy and teratogenicity.¹ However, a breakthrough study in 1999 reported promising efficacy in MM patients, leading to accelerated approval for MM treatment in May 2006^2,3 (Figure 1). This success spurred the development of thalidomide analogs, namely, lenalidomide and pomalidomide, to enhance therapeutic effectiveness while reducing toxicities. Lenalidomide received US Food and Drug Administration (FDA) approval in June 2006 for use in MM patients who have had at least one prior line of therapy.⁴ Pomalidomide was granted FDA approval seven years later for relapsed / refractory MM patients (RRMM) who had undergone at least two prior therapies, including lenalidomide and bortezomib.³ While thalidomide is now less prescribed, lenalidomide is widely used as the backbone of numerous combination treatments for newly diagnosed MM (NDMM), as post-transplant maintenance, and in RRMM, whereas pomalidomide is commonly used for treatment of RRMM, especially for lenalidomide-refractory patients.⁵

The clinical benefit of IMiD in MM is well established; however, their therapeutic efficacy and durability are significantly limited by primary and acquired drug resistance. In MM, approximately 5% of patients demonstrate primary resistance to IMiD, while those who initially responded to IMiD-based regimens eventually acquire resistance over time.^6-9 Moreover, recent analyses suggest efficacy and key subset differences depend on age, particularly in older patients where clinical benefit is more limited. There is, therefore, a significant unmet need in understanding the underlying mechanisms of resistance to IMiD.

Landmark studies have revealed that IMiD exert their activity by binding to a specific tri-tryptophan pocket of cereblon (CRBN), a substrate adaptor protein of the CRL4^CRBN E3 ubiquitin ligase complex, which consists of DNA damage-binding protein 1 (DDB1), cullin-4A/B (CUL4A/B), and regulator of cullins 1 (ROC1).^10,11

Figure 1.Timeline of the development of immunomodulatory drugs. The historical timeline of immunomodulatory drug (IMiD) development over the years from multiple myeloma (MM) to its therapeutic application in different malignancies. FDA: US Food and Drug Administration.

Multiple studies on CRBN pathway abnormalities have facilitated our understanding of IMiD resistance.^12-14 Resistance mechanisms beyond the CRBN pathway, on the other hand, are gradually emerging, but their relative significance, and how one study relates to another, is not fully understood. In this current review, we explore in depth the evidence on CRBN pathway-independent mechanisms of IMiD resistance, dissect the details of the studies, and systematically describe the evidence based on how one may be supporting the findings of another. Additionally, we discuss areas of future research that may hold promise in advancing our understanding of IMiD resistance and propose plausible therapeutic strategies to overcome IMiD resistance in the clinic.

Mechanism of action of immunomodulatory drugs

By binding to CRBN, IMiD redirect the CRL4^CRBN E3 ubiquitin ligase machinery to target and induce proteasomal degradation of a range of neo-substrates, including the transcription factors Ikaros and Aiolos, encoded by the IKZF1 and IKZF3 genes, respectively^10,11,15 (Figure 2A). Both IKZF1 and IKZF3 (IKZF1/3) are regulators of B-cell differentiation and were described to be essential genes in MM.^16,17 IMiD-induced-degradation of IKZF1/3 is, therefore, crucial for the anti-neoplastic effects in MM, mediating the cellular toxicity and/or induction of immunomodulatory responses (Figure 2B). The direct anti-MM effect is mainly attributed to the downregulation of two MM essential genes, IRF4 and c-MYC, causing disruption of their oncogenic drive and hence cytotoxicity.^18,19 At the microenvironment level, IMiD enhance immunomodulatory responses by: 1) promoting immune recognition through increased antigen presentation by dendritic cells; 2) increasing production of anti-tumor cytokines such as interferon-γ (IFN-γ) and interleukin-2 (IL-2) to drive T-cell expansion and natural killer (NK)-cell activation; 3) reducing adhesion molecules such as VCAM-1 and ICAM-1 on bone marrow stromal cells (BMSC) to impede tumor cell-BMSC interactions; 4) inhibiting immunosuppressive T-regulatory cells (Tregs); and 5) impairing signaling of angiogenic factors, vascular endothelial growth factors (VEGF) and fibroblast growth factors (FGF) in the BM niche.^20-22

The distinct clinical efficacy of each thalidomide derivative reflects the differences in its CRBN binding affinity and the subtly different spectrum of neo-substrate degradation.²³ Thalidomide, lenalidomide, and pomalidomide share common phthalimide and glutarimide moieties, but all differ, respectively, in a carboxy and an amino group at the phthalimide ring (Figure 2C). These small but key structural variations lead to significant differences in clinical efficacy, with increased potency observed from thalidomide to lenalidomide and then to pomalidomide.¹ Compared with thalidomide, lenalidomide was documented to be 50-2,000 times more potent in inducing T-cell proliferation and 300-1,200 times more potent in augmenting T-cell activity due to increased IL-2 and IFNγ production. Pomalidomide is 10 times more efficient than its predecessor in stimulating T cell and inducing pro-inflammatory cytokines from Type 1 helper (Th1) cells, while reducing anti-inflammatory cytokines from Th2 cells.²⁴

Figure 2.Modes of action of immunomodulatory drugs. (A) Immunomodulatory drugs (IMiD) act as a ‘molecular glue’ to mediate recruitment of neo-sub-strates to the CRL4^CRBN E3 ubiquitin ligase which results in neo-substrate ubiquitylation and proteasome degradation. (B) Summary of the effects of IMiD-induced neo-substrate degradation in multiple myeloma (MM), which includes direct anti-neoplastic and immunomodulatory effects. (C) Chemical structures of cereblon (CRBN) binders. IMiD and CELM-oD share a glutarimide ring that binds to CRBN; however, CELM-oD possess more extended structures on their phthalimide ring (boxed in red). CELMoD: cereblon E3 ligase modulatory drugs; Treg: regulatory T cell.

Immunomodulatory drug resistance associated with the cereblon pathway, and its paradox

Owing to the core function of CRBN in the activity of IMiD, disruption to the CRL4^CRBN E3 ubiquitin ligase components has been the most commonly reported mode of resistance to this group of drugs (Table 1). Here, we document the key findings from previous studies on the genomic and non-genomic abnormalities of CRBN and its pathway genes, and their association, or not, with patients’ responses to IMiD.^25-28

Genomic alterations in the CRBN gene that have been reported in MM patients include most commonly, single nucleotide variation (SNV) and copy number loss, while the non-genomic events involved epigenetic and transcriptomic aberrations that affect its stability and expression, including the abnormal exon 10 splicing.^25,26,29-31 Somatic SNV in CRBN are not frequent among newly diagnosed MM (<1%); however, their prevalence significantly increases to 9-12% of IMiD-refractory patients.^25,26 Genetic mapping identified that these SNV were predominantly located within the IMiD-binding domain,²⁶ and their ectopic introduction into MM cell lines obliterated responses to lenalidomide.^26,32 There was also an increase in the frequency of CRBN copy number loss, from 1.5% in NDMM to 7.9% in lenalidomide (Len)-refractory and a significant 24% in pomalidomide (Pom)-refractory patients.²⁵ In addition, higher levels of alternative splicing of exon-10 in CRBN, which prevents IMiD-binding, has an incidence reaching up to 10% of Len-refractory patients, and has been consistent in predicting poor responses to IMiD.^25,33 These reports underscore the biological role of genomic and non-genomic lesions of CRBN in IMiD resistance. However, it is noteworthy that they represent only a small proportion of IMiD-refractory patients (up to 20% for lenalidomide and up to 30% for pomalidomide), suggesting that the majority of acquired IMiD resistance cases (i.e., >70-80%) are unaccounted for by CRBN abnormalities. Studies from smaller independent cohorts have not reported changes in the frequency of CRBN mutations in Len-refractory versus NDMM patients,^34-36 but are mainly small and/or limited to SNV detection.

Table 1.Immunomodulatory drug resistance associated with cereblon pathway.

Pre-clinical modeling of acquired IMiD-resistance demonstrated that resistant cell lines had depleted CRBN expression.²⁹ This is consistent with clinical observations, whereby high CRBN expression correlated with improved progression-free survival (PFS) in IMiD-treated-patients, while the IMiD-non-responders exhibited reduced CRBN expression.^30,37 However, it is also notable in other studies that CRBN levels were not predictive for IMiD responses.^34,38 Importantly, amongst all types of CRBN abnormalities described in IMiD-RRMM, there was no one mechanism that rendered a complete loss of CRBN expression.^25,29,31 For example, neither the cases with CRBN gene copy loss nor those with an aberrant exon 10 splicing demonstrated changes/reduction in CRBN expression compared to their counterparts without these aberrations.²⁵ More interestingly, a substantial proportion of the IMiD-RRMM (32%) paradoxically exhibited increased CRBN expression with no loss-of-function variant detected.³⁹ These data suggest that in the cases without genetic loss-of-function, the activity of the CRBN pathway was putatively retained in the IMiD-resistant MM cells. This supports the observation that len-refractory patients were responsive to subsequent pomalidomide treatment,^5,40,41 suggesting that low but intact CRBN expression does not abolish the functional CRL4^CRBN E3 ubiquitin ligase activity and residual CRBN signaling may still mediate responsiveness to the more potent pomalidomide.

Investigations into CRBN axis genes have also yielded inconclusive results. For example, IMiD-refractory disease had an increased mutation frequency in IKZF1, IRF4 and CUL4B compared to NDMM,²⁶ but another study showed no difference in the mutation status of DDB1, CUL4A, CUL4B, IKZF1, IKZF2, and IKZF3.³⁶ High IKZF1/3 expression has been associated with poorer PFS in lenalidomide-treated patients,⁴² and, in contrasting data, with a favorable response to lenalidomide and better PFS.^43,44 In another patient cohort, IKZF1/3 protein levels were non-prognostic.⁴⁵ Further upstream in the CRBN pathway, COP9 signalosome (CSN) and E2 ubiquitin ligase proteins (UBE) are required for the maintenance of CRL4^CRBN E3 ubiqitin ligase activity. Whole-genome sequencing (WGS) of MM patients (N=522) revealed increased incidence of copy number loss of chromosomal 2q37, the region containing CSN members (COPS7B and COPS8) in the lenalidomide-refractory, and lenalidomide-then-pomalidomide-refractory patients compared to the NDMM patients.²⁸ Importantly, however, the enrichment of this abnormality was again observed only in a small subset of the refractory patients (10-16%) and mutational analysis revealed low frequency for CSN and UBE members, implicating that once those with CRBN aberrations were also removed, >60% of IMiD-refractory patients are still unaccounted for by the aberrancy in these CRBN pathway proteins.²⁸

In summary, CRBN pathway abnormalities are not a universal event in IMiD-refractory patients and the mechanism of resistance is likely to extend beyond this canonical IMiD-pathway.

Immunomodulatory drug resistance independent of the cereblon pathway

Supporting this notion, evidence on IMiD resistance mechanisms independent of the CRBN pathway are gradually emerging. These have been described to include myeloma cell-intrinsic factors and myeloma cell-extrinsic factors (Table 2).

Intrinsic mechanisms

The main myeloma cell-intrinsic mechanisms of IMiD resistance in MM involve the dysregulation of various oncogenic mediators, including known MM essential genes and other under-studied MM drivers. Some of these studies reported direct mechanistic evidence, while others showed clinical association without clear mechanisms (Figure 3).

Interleukin-6 (IL-6)/STAT3 signaling, a crucial MM driver, was found to be up-regulated in an acquired lenalidomide resistant MM cell line, XG1 (XG1-LenRes).^46,47 The authors identified autocrine production of IL6 in the XG1-LenRes, which was further enhanced in the presence of lenalidomide treatment. Stimulation of parental-XG1 cells with IL6 induced lenalidomide resistance, whereas inhibition of STAT3 with a selective STAT3 inhibitor (PB-1-102) re-sensitized its isogenic resistant counterpart to lenalidomide. Constitutive IL6/STAT3 activity in XG1-Len-resistant (LenRes) was associated with sustained expression of IRF4. Notably, XG1-LenRes did not have any accompanying abnormalities in CRBN and its downstream components. There was no change in CRBN expression, and neither were there any differential effects on lenalidomide-induced IKZF1/3 degradation. In agreement with this, introduction of exogenous CRBN also failed to restore lenalidomide sensitivity in XG1-LenRes,⁴⁷ indicating the involvement of CRBN pathway-independent mechanisms.

Table 2.Cereblon pathway-independent evidence in Immunomodulatory drug resistance.

Dysregulation of another oncogenic pathway, the Wnt/β-catenin, was also observed in LenRes MM cells. Through gene expression profiling of lenalidomide-resistant-U266, -ANBL-6, -KAS-6 and -MM1.S versus their parental cells, several Wnt/β-catenin intermediates (Wnt-3, Fzd-4, β-catenin) were found to be up-regulated. This increase in Wnt/β-catenin activity led to stabilization of cytoplasmic β-catenin and upregulation of MM drivers CyclinD1 and c-Myc. Knocking down β-catenin, in turn, restored MM cell sensitivity to lenalidomide.⁴⁸ Another report showed that CD44, a downstream transcriptional target of β-catenin, was also associated with IMiD resistance. The authors found that increased CD44 expression in the LenRes cells enhanced MM cell adhesion to BMSC to promote cell survival. Inhibition of β-catenin, and consequently CD44, with all-trans retinoic acid (ATRA) successfully re-sensitized resistant MM cells to lenalidomide.⁴⁹

The main players in the oncogenic MAPK pathway such as NRAS, KRAS and, to a lesser extent BRAF, are the most frequently mutated genes in MM patients, with NDMM and RRMM cases bearing a high 20-50% and 45-80% frequency, respectively.^26,50-52 BRAF/KRAS/NRAS are upstream mediators of the mitogen-activated protein kinase (MEK)/extracellular signal-regulated kinase (ERK) kinases and activating mutations of BRAF/KRAS/NRAS genes trigger these kinases to up-regulate a series of proliferative and cell cycle signals.⁵¹ A mouse xenograft study bearing MM1.S plasmacytomas with acquired IMiD resistance showed that resistance onset was accompanied by hyperactivity of the MEK1/ERK pathway (increased pMEK1/2 and pERK1/2). The addition of selumetinib, a small-molecule MEK inhibitor, effectively reinstated IMiD sensitivity, both in vivo and ex vivo,⁵³ hence suggesting the role of the BRAF/KRAS/NRAS /MEK/ERK signaling cascade in mediating IMiD resistance. Nevertheless, it should be noted that BRAF/KRAS/NRAS gene mutations are a general predictor of poor clinical outcome and are observed widely in all RRMM states.⁵⁰ There is a diverse range of SNV reported⁵⁴ and the functional impact of these different BRAF/KRAS/NRAS point mutations on the activation of MEK/ERK pathway leading specifically to IMiD resistance requires more study.

The biological role of MEK/ERK signaling was further demonstrated in a genome-wide CRISPR-Cas9 knockout screen in pomalidomide-treated MM.1s cells, in which TNF receptor-associated factor 2 (TRAF2) appeared as a modulator of resistance.⁵⁵ In cells co-cultured with BMSC or BMSC supernatants, the authors identified that IL-6 directly activates MEK/ERK signaling while triggering proteasomal degradation of TRAF2 to stimulate nuclear factor κB (NFκB) and ERK signaling. MM1.S cells with TRAF2 knockout exhibited significant resistance to lenalidomide and pomalidomide, alongside activation of NF-κB and MEK/ERK pathways, independently of the CRBN-IKZF1/3 axis. Consistent with the former study, inhibition of MEK with selumetinib effectively overcame IMiD resistance in TRAF2 knockout MM cells; however, the authors also identified that TRAF2 knockout conferred higher resistance to dexamethasone and melphalan treatments, indicating that TRAF2 knockout-induced drug resistance may not be specific to IMiD.⁵⁵ Epigenetic alterations are widely implicated in cancer drug resistance. In the case of IMiD in MM: 1) a global increase in DNA methylation with 2) a reciprocal decrease in chromatin accessibility, and 3) a dominance of gene downregulation, were observed in acquired IMiD-resistant-OPM2 and -H929, with the main components of the CRBN pathway (CRBN, IKZF1/3, and IRF4) being unaffected. This is consistent with reports that promoter silencing of CRBN and its pathway genes were not associated with CRBN pathway deficiency and IMiD resistance.^39,56,57 In this study, the authors, instead, identified SMAD3 (a transcription factor and cell signaling regulator) as the novel gene commonly down-regulated in the resistant counterpart of both the cell lines. Treatment with a combination of 5-azacytidine (Aza) and the EZH2 inhibitor (EPZ-6438) reverted chromatin repression, increased SMAD3 expression, and ultimately re-sensitized the resistant cells to IMiD.⁵⁷ In support of this pre-clinical finding, Kalff et al. have reported some clinical efficacy in combining oral Aza with lenalidomide-dexamethasone (Rd) in heavily treated LEN-resistant RRMM patients (overall response rate [ORR] 37.5%, clinical benefit rate 50%).⁵⁸ Although Khouri et al.’s Rd-Aza (subcutaneous) treatment protocol in another patient cohort yielded a lower response rate (ORR 22%, clinical benefit rate 32%), the authors propose that Rd-Aza may overcome some IMiD refractoriness with careful regime optimization and correct patient selection.⁵⁹ Further evidence of epigenetic involvement in IMiD resistance was demonstrated in a chromatin immunoprecipitation sequencing (ChIP-seq) study across 16 MM cell lines, comparing LenRes versus len-sensitive cells.⁶⁰ At the gene specific level, the authors reported that in the lenalidomide-resistant cells, the promoter regions of ANKRD30B and SLAMF6 exhibited the highest occupancy of the active H3K4me3 mark, while the promoter regions of GPR15 and NKX6-1 demonstrated a marked depletion. Among the CRBN pathway genes, only CUL4B displayed enriched H3K4me3 at its promoter region in the len-sensitive cells. Nonetheless, the underlying mechanism by which IMiD induce epigenetic reprogramming, and the extent to which changes in epigenetics contribute to IMiD loss of function (LOF), was not described, and this represents an essential area of study as work moves forward.

Figure 3.Intrinsic cereblon-independent immunomodulatory drug resistance mechanisms. Intrinsic factors that contribute to immunomodulatory drug (IMiD) resistance beyond the cereblon (CRBN) pathway include (A) aberrant activation of signaling pathways (such as IL-6/STAT3, Wnt/β-catenin and MEK/ERK), (B) epigenetic alterations and transcription factor redundancy, and (C) high-risk prognostic markers (such as gain/amp(1p21) and del(17p)), differential expression of circRNA, CDK6 overexpression, and dysregulated cell surface receptors. BM: bone marrow; MM: multiple myeloma.

Furthermore, NCOR2, an epigenetic remodeling gene, has been implicated in multi-drug resistance in MM, including to IMiD.⁶¹ The authors identified that NCOR2 was interacting with nucleosome remodeling and deacetylase (NuRD) complex, to repress the expression of CD180 by directly binding to its promoter and that it resulted in the down-regulation of MYC. They showed in IMiD-resistant cells that low NCOR2 and CD180 expression was associated with increased MYC expression. There was no change in CRBN and IKZF1 expression, and thus they concluded that high MYC in IMiD-resistance was induced by deregulation of the NCOR2-CD180 pathway, independently of CRBN. It is noteworthy that, in this study, NCOR2 knockout also led to resistance to BET and HDAC inhibitors; thus, the therapeutic implication of loss of NCOR2 in MM was not specific to IMiD. A more specific epigenetic dissection of IMiD resistance has been recently reported.⁶² Out of 48 MM cell lines challenged with pomalidomide, 44 (92%) remained viable, despite significant depletion of IKZF1/3. These MM cells displayed high growth rate with most of the IMiD-resistant cell lines retaining high levels of MYC and IRF4 expression.⁶² Further investigations using an assay for transposase-accessible chromatin with sequencing (ATAC-Seq) revealed reduced chromatin accessibility for IKZF1-binding in the pomalidomide- versus DMSO-treated cells. Notably, the sites that lost chromatin accessibility for IKZF1 were enriched for BATF, IRF4 and FOX bHLH binding motifs. RNA-sequencing of 66 MM cell lines revealed that the inherently IMiD-resistant cell lines (e.g., KMS-12BM, RPMI-8226) expressed high levels of BATF. They identified that BATF heterodimerization was compensating for the IMiD-induced loss of IKZF1/3 to sustain IRF4 overexpression, ultimately leading to IMiD resistance. In the analysis of paired NDMM and RRMM patients treated with IMiD (N=35), a significant upregulation of BATF upon relapse was observed. Cross-referencing these findings in the CoMMpass dataset, the authors identified that high BATF expression did indeed confer poorer survival outcome (N=484) in IMiD-treated patients.⁶²

IKZF1/3 redundancy in IMiD responses was also highlighted in a complementary study whereby two other factors, EP300 and BRD4, compensated for the IMiD-mediated loss of IKZF1/3.⁶³ On ChIP-seq analysis, half the chromatin-bound IKZF1/3 sites overlapped with EP300 and BRD4 binding sites. While lenalidomide universally depleted chromatin-bound IKZF1 in both IMiD-sensitive MM1.S and IMiD-resistant RPMI-8226 cells, the IMiD-resistant MM cells maintained BRD4 and P300 super-enhancer occupancy. Further investigation revealed that this was acting through transcription factor ETV4, which co-binds the enhancers with IKZF1 to induce IMiD resistance. They also reported that ETV4 expression was associated with poorer PFS and overall survival (OS) for CoMMpass patients treated with IMiD, and for POLLUX (NCT02076009) patients treated with Rd. Analysis of 36 paired CoMMpass patients and 14 paired POLLUX patients showed that ETV4 was significantly up-regulated at relapse, whereas no change was observed in the expression of IKZF1, IKZF3, IRF4 or MYC.⁶³ Alongside the immediate former publication,⁶² this work identifies that CRBN-mediated-degradation of IKZF1/3 can be bypassed in sustaining the oncogenic IRF4-MYC axis to drive IMiD resistance.

High-risk MM markers, specifically t(4;14), t(14;16), del(17p), and gain/amp(1q21), have been associated with early relapses following IMiD-based therapy.⁶⁴ A longitudinal genomic analysis of RRMM patients (N=386) highlighted the enrichment of gain/amp(1q21) and del(17p) in IMiD-refractory cases, underscoring the potential impact of genes up-regulated in chr1q and deleted in chr17p on IMiD responses.⁶⁵ In agreement with this, the Myeloma XI trial (N=556) revealed in their multivariate analysis that isolated gain(1q21) and double-hit cases (defined as two concomitant high-risk features) derived no survival benefit from lenalidomide maintenance.⁶⁶ We note, however, that 1q21(gain/amp) and 17p13(del) are poor predictive markers for a broad range of MM therapeutics, and, therefore, interactions between mechanisms specific to IMiD versus general drivers of resistance / early relapse need to be further interrogated. Notably, adenosine deaminase acting on RNA (ADAR1), encoding an RNA editing enzyme, is located in the amplified chr1q21 region. Our work has shown a close association between high ADAR1 expression and hyperedited MM transcriptome with reduced responsiveness to IMiD,⁶⁷ implicating the involvement of aberrant RNA editing in the mechanism of IMiD resistance. Another study has also shown that ADAR1-mediated editing of glioma-associated oncogene homolog 1 (GLI1), a Hedgehog pathway transcriptional activator and self-agonist, promotes malignant regeneration and IMiD resistance in MM.⁶⁸ In view of the growing interest and the biological relevance of RNA abnormalities in MM, our team is currently investigating the mechanism by which ADAR1 and its aberrant activity regulate IMiD responses in MM. We have identified a novel mechanism involving the ADAR1-regulated-dsRNA sensing pathway in modulating IMiD resistance.⁶⁹

Further evidence of RNA-related aberrations in IMiD resistance was reported in a recent circular RNA (circRNA) profiling study.⁷⁰ Totals of 200 and 277 differentially expressed circRNA were observed, in H929-LenRes and H929-Pom-resistant cells, respectively, compared to their sensitive counterpart. The authors identified ciRS-7 to be consistently down-regulated, while circIKZF3 was commonly up-regulated in the lenalidomide- and pomalidomide-resistant cells. The depletion of ciRS-7 correlated with increased methylation levels of the promoter CpG island of its host gene, LINC00632. Combination treatment of an EZH2 inhibitor (EPZ-6438) and a DNA methyl transferase inhibitor (5-azacytidine) partially restored the expression of LINC00632 and ciRS-7 and the IMiD sensitivity of the cells. Nevertheless, silencing ciRS-7 in sensitive parental cells did not increase resistance to IMiD, potentially suggesting indirect modes of action. In the case of circIKZF3, its underlying mechanism remains elusive, due to the challenges faced by the authors in knocking down circIKZF3 in both LenRes and Pom-resistant cells.⁷⁰

Investigation of the proteomics and RNA-sequencing analyses of RRMM patients treated with lenalidomide-based therapy has identified CDK6 upregulation as a driver of IMiD resistance.⁷¹ Overexpression of CDK6 in IMiD-sensitive MM cell lines resulted in reduced IMiD sensitivity, while the inhibition of CDK6 through Palbociclib or CDK6-specific proteolysis-targeting chimeras (PROTAC) (BSJ-03-123 or CST528) demonstrated synergy with IMiD both in vitro and in MM1.S xenografts. In their patient cohort, they did not detect any CRBN pathway abnormalities (RNA, protein, or phosphoprotein), genetic alterations to the other genes in the CRL4^CRBN E3 ligase complex, or association between CRBN and CDK6 protein expression. CRISPR / Cas9-knock-out of CRBN in MM lines led to no change in the expression of cyclin dependent kinase 6 (CDK6). Although CDK6 inactivation in conjunction with IMiD resulted in significant inhibition of MYC, downstream functions of CDK6 in RRMM remain unclear.⁷¹

CD138 is a marker for terminally differentiated plasma cells during normal B-cell development and serves as a specific surface antigen for MM cells.⁷² A significant increase in CD138-negative MM cells has been observed in relapsed or progressing patients (N=15) compared to NDMM patients (N=90).⁷³ Characterization of two MM cell lines (KYMM-1 and KYMM-2) established from a single patient showed that the cell population with decreased CD138 surface expression had higher lenalidomide resistance. The downregulation of IRF4 and upregulation of BCL6 was suggested as the mechanism for this resistance,⁷³ citing another study that documented high IRF4 expression was correlated with increased lenalidomide sensitivity.⁷⁴

In another study on surfaceome, glycoprotein cell surface capture (CSC) proteomics on LenRes OPM2 and H929 showed a common signature of increased CD33 and CD45/ PTPRC, when compared to their sensitive counterparts.⁷⁵ Analysis of the Multiple Myeloma Research Foundation (MMRF) CoMMpass dataset from paired diagnosis and first-relapse tumor cells (verIA14, N=50, where 94% of patients had received lenalidomide and dexamethasone with a proteasome inhibitor [PI] as part of their induction regimen) revealed that both CD33 and PTPRC transcripts were significantly increased at first relapse.⁷⁵ The authors noted that plasma cell expression of either of these markers has been associated with poor prognosis for NDMM, who exhibited more aggressive disease upon lenalidomide resistance.^76,77 They, however, did not describe, nor further investigate, the basic or plausible mechanisms underlying CD33 and PTPRC associations with disease progression. Lastly, various other genes have appeared in genome-wide CRISPR-screens in cell lines as regulators of IMiD sensitivity, for example, TOP2B, EDC4, RARA, SNRNP25, OTUB1, PLAA, DEPDC5, SRP14, XRN1, EIF4A1, ARID2, MBTPS1/2, and SCAP.^78-80 Whether they have any relevance to clinical IMID resistance, and if so how, remains a topic for future research.

Extrinsic mechanisms

The interaction of MM cells with BM components such as secreted growth factors/cytokines, BMSC, and immune cells can promote growth, survival, and drug resistance of MM cells (Figure 4).

Characterization of the MM tumor microenvironment (TME) landscape in patients from the MANHATTAN trial (clinical-trials.gov 03290950; N=49 patients) who did not achieve minimal residual disease (MRD) negativity and have shorter PFS, demonstrated TME dysregulations including low population of CD14⁺ monocytes, increased frequency of T-cell exhaustion, persistence of IFNγ-expressing NK cells, and decreased T-cell receptor (TCR) diversity.⁸¹ These were observed alongside genomic defects such as high APOBEC mutational activity, 1p22 (RPL5) deletions and IKZF3 loss, suggesting an interplay between tumor genomic features with the composition of TME in driving drug resistance.

In a prospective observational clinical trial, immune-profiling of lenalidomide-RRMM, non-refractory MM and non-MM-healthy individuals revealed that the lenalidomide-RRMM patients had a significant expansion of effector T-cell populations that express elevated levels of checkpoint molecules, lymphocyte activating 3 (LAG3) and programmed cell death 1 (PD-1). Their frequency was positively correlated with increased serum inflammatory cytokines, IL6, IL17, and TNFα.⁸² High levels of PD-1 and LAG3-positive T cells were predictive of inferior survival and clinical outcomes in these RRMM patients, as in previous studies.^83,84

The chemokine (C-C motif) ligand 20 (CCL20) was found to be down-regulated in an acquired lenalidomide-resistant MM cell line (U266-LenRes), as well as in the MM cells and plasma of lenalidomide-treated RRMM patients (N=5). At both the in vitro and in vivo level, the addition of CCL20 was able to re-sensitize MM cells to lenalidomide.⁸⁵ The authors postulated that CCL20 plays a role in increasing lymphocyte chemotaxis to the tumor areas and in assisting the cell-mediated immunity.

Our group has recently reported metabolic reprogramming, with the release of metabolic waste product such as lactate into the TME, to be associated with IMiD resistance.⁸⁶ Elevated lactate secretion is known to promote acidosis in the TME, driving metastasis, angiogenesis, and drug resistance.⁸⁷ t(4;14) is a high-risk MM marker with a prevalence rate of 15-20% that drives overexpression of the histone methyltransferase, nuclear receptor binding SET domain protein 2 (NSD2).⁸⁸ NSD2 promotes plasma cell transformation by catalyzing the active histone mark H3K36me2. We identified that protein kinase C alpha (PKCα) is an epigenetic target of NSD2.⁸⁶ Through metabolomics analysis, we found that lactate was a differential metabolite associated with PKCα. High lactate levels were associated with reduced responsiveness to lenalidomide. Knockdown of PKCα resulted in reduced intracellular and extracellular lactate levels, consequently increasing cellular sensitivity to lenalidomide, independent of the CRBN-IKZF1/3 axis. Clinically, t(4;14) MM patients demonstrated elevated plasma lactate levels compared to non-t(4;14) patients and did not derive significant benefits from lenalidomide-based regimens.⁸⁶

Multiple myeloma cells with an IMiD-resistance phenotype have also been associated with increased secretion of extracellular vesicles (EV) and enhanced adherence abilities.⁸⁹ Through a comprehensive transcriptomic analysis of acquired lenalidomide-resistant MM cell lines (KMS-21, KMS-27, KMS-34), core regulatory genes governing EV secretion, including sortilin 1 (SORT1) and lysosomal associated membrane protein 2 (LAMP2), were found to be significantly up-regulated compared to their sensitive counterparts. Knockdown of SORT1 or LAMP2 reduced EV secretion, decreased cell adhesion and restored lenalidomide sensitivity in lenalidomide-resistant cells without affecting CRBN expression. Further analysis of publicly available clinical data revealed that high SORT1 and LAMP2 expression were associated with poor survival in MM patients (GSE19784, N=300 patients) and in patients treated with lenalidomide (GSE136324, N >200 patients).⁸⁹

Figure 4.Extrinsic mechanisms independent of the cereblon pathway. Potential tumor microenvironment (TME) mechanisms of immunomodulatory drug (IMiD) resistance which include (A) composition of the immune cells in the TME with tumor-acquired genetic features, (B) expansion of exhausted T-cell population, (C) downregulation of CCL20, (D) elevated lactate levels in the TME, and (E) increased secretion of extracellular vesicle (EV)-mediated multiple myeloma (MM) cell adhesion and resistance.

Taken together, the above findings suggest that aberrations at the TME may impinge on the efficacy of IMiD in MM, further highlighting that IMiD resistance is likely to involve an interplay of many biological factors.

Future perspectives

The discovery of CRBN as a pivotal target of IMiD has been instrumental in advancing our understanding of the molecular mechanism of these therapeutic agents.^10,11 Given the inevitable occurrence of IMiD-resistance in MM, the identification of biomarkers that can accurately predict for IMiD responses is of paramount importance to increase the prospects of therapeutic efficacies.

At present, exploring CRBN status as a potential biomarker for predicting IMiD responses and resistance seems a conceivable strategy; however, as reviewed in the earlier sections, the results reported hitherto have been rather inconclusive. In light of this, several critical limitations should be addressed to ensure the reliability of using CRBN as a biomarker. Firstly, standardized assays for quantifying functional levels of CRBN expression in clinical samples are currently lacking.⁹⁰ The development of robust and reproducible measurement techniques is crucial if CRBN expression is to have any role as a biomarker in clinical practice. Next, determining the optimal approach for assessing functional CRBN levels, whether at the mRNA or protein level or both, and evaluating their genomic aberrations are equally important. Identifying the cancer clonal fraction (CCF) harboring these genomic events at relapse versus diagnosis will provide insights into the clonal selection of IMiD-resistant subclones and whether longitudinal exposure to the therapies could drive clonal selection. In addition, determining the threshold level at which the CCF is deemed prognostic for patients will be essential in guiding personalized treatment strategies. Considering that CRBN expression may be down-regulated but not completely abrogated in RRMM, it will also be paramount to consider whether a threshold of expression, and which and how different transcript splice variants should be measured to identify a non-functional CRBN activity, could predict treatment failure. It is likely that any such biomarker parameters would also vary between different IMiD and cereblon E3 ligase modulators (CELMoD).

Similarly, the approach to measure other CRBN pathway proteins, such as IKZF1/3, should be undertaken in greater depth. For example, the rate rather than the magnitude of IKZF1/3 degradation was found to be the more important determinant for modulating IRF4 expression, and, thus, the efficacy of IMiD.⁹¹ This highlights the need to investigate the timepoint and protein level at which IKZF1/3 cease to be sufficient to support transcription of the downstream IRF4 oncogenic events, and whether any compensatory mechanisms such as BATF and ETV4 may assume the transcriptional regulation roles of IKZF1/3. Further, IMiD-bound CRBN has binding affinity for IKZF1/3 and other reported competing neo-substrates, which, at variable levels between different myelomas, may be differentially responsible for response and resistance. These are some of the interesting questions that could form the basis of future research.

High-risk copy number alterations in gain/amp(1q21) and del(17p) involve a large number of genes. It is plausible that some of them may play a role in driving IMiD resistance. For example, in chromosome 17p, a few IMiD-response pathway genes (UBE2G1, NCOR1, and COPS3) reside close to tumor suppressor TP53. The functional impact of the loss of these genes in del(17p) MM has not been investigated. Whether IMiD resistance might be driven by these genes independently or by their co-deletion with TP53 awaits further investigation.

The growing body of evidence on the significance of non-canonical pathways and CRBN-independent mechanisms in regulating the efficacy of IMiD have been demonstrated mostly in cell line models (Table 2), which means that clinical evidence is not well-established. This highlights the need to determine how these alternative genes/modes of action are translationally relevant, given that many of the CRBN-independent abnormalities seem to be enriched also in patients resistant to non-IMiD treatment regimens. The proposed associations should, therefore, be validated in clinical samples, alongside the well-studied CRBN pathway genes, to determine the prognostic value of any associations found, and, if they prove worthwhile, to be further explored as a novel predictor and specific biomarker for IMiD treatment.

As described, there are also quite a number of studies on clinical samples that report associations with IMiD response but do not yet have direct mechanistic explanations. Greater effort into elucidating these mechanisms and how they modulate IMiD responses in MM is of paramount importance as we move forward in identifying information that can be used to aid clinical decision making.

Previous genome-wide CRISPR/Cas9-mediated studies were conducted with the typical aim of identifying genes that regulate IMiD sensitivity.^78-80,92 It is also worthwhile to validate these findings in a converse manner, i.e., CRISPR studies on IMiD-resistant cells to decipher their novel dependencies and therapeutic vulnerabilities. This might identify genes that, when knocked out or activated, will result in the re-sensitization or killing of IMiD-resistant MM cells.

Tumor heterogeneity leads to the development of multiple mechanisms of resistance to IMiD.⁹³ Rapid advancements in single cell profiling technology have enabled us to dissect the heterogeneity of cells at both single and spatial resolution. Spatial single-cell transcriptomics is a burgeoning tool to decipher tumor architecture and TME,^94,95 and studies utilizing this cutting-edge technology in MM is still in its infancy. Given the current lack of knowledge surrounding the association between cell-intrinsic and cell-extrinsic mechanisms in IMiD resistance, it will be pertinent to adopt spatial single-cell technology in future investigations to enable detailed characterization of the MM cell-TME interaction.

It is becoming evident that the mechanisms underpinning IMiD resistance in MM involves a complex interplay of genomic, transcriptomic, and proteomic events both in tumor and immune cells. Future approaches to overcoming IMiD resistance may, therefore, call for targeting a combination of CRBN pathway-dependent and pathway-independent mechanisms. This would entail adopting personalized multi-drug strategies that concurrently target individually-relevant pathways. With the immense array of possible target combinations, predicting the optimal drug combination for an individual patient presents a future trajectory for optimal therapy. In this era of artificial intelligence, it is likely to require deep machine learning techniques to effectively map individualized drug-drug interactions, rank different drug combinations, and determine a more accurate drug dosage for optimal clinical efficacy. Such an approach has been successfully adopted in our previous proof-of-concept study in MM and lymphoma.^96,97

Next-generation cereblon targeting agents, CELMoD (iberdomide and mezigdomide), are in ongoing clinical trials in MM as a means to overcome IMiD resistance. They promise higher potency, more robust degradation of known targets, and an expanded repertoire of neo-substrate targets.^98,99 Compared to IMiD, the binding affinity of iberdomide for CRBN was 20-fold higher (IC₅₀ for CRBN binding was approx. 3 uM and approx. 0.15 uM, respectively) which leads to a more rapid IKZF1/3 degradation and, therefore, enhanced treatment efficacy.¹⁰⁰ Iberdomide and mezigdomide have both shown meaningful clinical outcome in heavily pre-treated IMiD-refractory MM patients with ORR 26% for the former and 40% for the latter.^101,102 However, this also implies that up to 60% of IMiD-refractoriness cannot yet be rescued by next-generation counterparts. Given that CELMoD share the same mechanisms of action as IMiD, it remains to be seen whether any particular CRBN-dependent or -independent mechanisms also contribute to the therapeutic insufficiency of CELMoD.^103-105 The comprehensive understanding of the range of biology of IMiD resistance, therefore, holds the key in facilitating the successful integration of these new agents into clinical practice.

Conclusions

Immunomodulatory drug resistance associated with the CRBN pathway has been extensively reviewed previously.^12-14 In this review, we extended our discussion into several other important aspects including potential CRBN pathway-independent IMiD resistance mechanisms, ranging from the cell intrinsic IMiD-resistance to the extrinsic components in the TME. The transformation of IMiD from a teratogenic ‘dark remedy’ to the pioneering standard-of-care treatment today in MM marks a significant shift in treatment paradigms, but lasting responses are inevitably hindered by acquired resistance, which represents an unmet therapeutic need. As we move forward, a clearer understanding of which resistance mechanisms are clinically relevant and why will lead us to new avenues for personalized and effective therapeutic interventions as we manoeuvre the challenging IMiD landscape, with IMiD resistance and the differential impact of new agents across age gaps⁴¹ in particular, and so better translate these promising results from the clinical trial to real-world practice.¹⁰⁶

Footnotes

• Received April 19, 2024
• Accepted November 8, 2024

Correspondence

*PJT and MYK contributed equally as first authors

P.J. Teoh mdctpju@nus.edu.sg

W.J. Chng mdccwj@nus.edu.sg

Disclosures

No conflicts of interest to disclose.

Funding

This work was supported by the Cancer Science Institute of Singapore, National University of Singapore, through the National Research Foundation Singapore and the Singapore Ministry of Education under its Research Centres of Excellence initiative.

Acknowledgments

Figures were created with BioRender.com.