Histone deacetylase‐based dual targeted inhibition in multiple myeloma

Despite enormous advances in terms of therapeutic strategies, multiple myeloma (MM) still remains an incurable disease with MM patients often becoming resistant to standard treatments. To date, multiple combined and targeted therapies have proven to be more beneficial compared to monotherapy approaches, leading to a decrease in drug resistance and an improvement in median overall survival in patients. Moreover, recent breakthroughs highlighted the relevant role of histone deacetylases (HDACs) in cancer treatment, including MM. Thus, the simultaneous use of HDAC inhibitors with other conventional regimens, such as proteasome inhibitors, is of interest in the field. In this review, we provide a general overview of HDAC‐based combination treatments in MM, through a critical presentation of publications from the past few decades related to in vitro and in vivo studies, as well as clinical trials. Furthermore, we discuss the recent introduction of dual‐inhibitor entities that could have the same beneficial effects as drug combinations with the advantage of having two or more pharmacophores in one molecular structure. These findings could represent a starting‐point for both reducing therapeutic doses and lowering the risk of developing drug resistance.


| INTRODUCTION
Multiple myeloma (MM) represents 10% of all hematological pathologies and it is the second most frequent one following non-Hodgkin lymphoma. It is characterized by an excess of abnormal plasma cells, predominantly localized in the bone marrow, which leads to the accumulation of monoclonal proteins in the serum and/or urine with associated multiorgan dysfunctions. 1 Median survival is 5-10 years and this is mostly due to the high rate of relapses and resistance. 2,3 Nevertheless, the standard of care has been significantly improved with the introduction of novel classes of drugs. Proteasome inhibitors (PIs), immunomodulatory drugs (IMiDs), histone deacetylase inhibitors (HDACi), monoclonal antibodies (mAbs), exportin-1 inhibitors and chimeric antigen receptor (CAR) T-cell therapy are all included in the armamentarium. [4][5][6] Peptide drug conjugates were also included into the list up until recently. However, despite the accelerated approval granted to the first-in-class peptide drug conjugate by the Food and Drug Administration (FDA) in February 2021, a safety alert arose in July of the same year as a consequence of the results of the phase III clinical trial OCEAN. [7][8][9] Among the available drugs, HDACi exert antimyeloma effects through targeting epigenetic and protein metabolism pathways. Overexpression of HDACs and their dysregulated functions have been observed in many cancer types, including MM. 10 HDACs are enzymes that regulate histone and nonhistone activity through removal of acetyl groups. 11,12 This causes chromatin condensation and transcriptional repression, together with effects on cell metabolism, apoptosis and autophagy. 11,12 The human HDAC family counts 18 HDACs, which are further classified into two categories based on their catalytic mechanism: 11 of them are zinc-dependent metalloenzymes (classes I, II, IV) and the remaining 7 (class III, also named sirtuins) require nicotinamide adenine dinucleotide as a co-factor for their enzymatic activity. [13][14][15][16] The four classes of HDACs are also divided according to their cellular localization, mechanism of catalysis, structure and expression pattern. [13][14][15][16] Class I HDACs are located in the nucleus and consist of HDACs 1, 2, 3, and 8. Class II HDACs, which are localized in both the nucleus and cytoplasm, comprise HDACs 4, 5, 6, 7, 9, and 10, and are further subdivided into classes IIA (HDACs 4, 5, 7, and 9) and IIB (HDACs 6 and 10). Class III HDACs are found in the nucleus, cytoplasm and mitochondria. HDAC11 belongs to class IV and is present in the nucleus and cytoplasm. [13][14][15][16] In MM, HDACi block myeloma cell viability and proliferation through various mechanisms of action, and either inhibit a broad range of HDACs (pan-HDACi) or are more selective. 16 Class I HDACs and HDAC6 are particularly deregulated in MM. 17 Panobinostat (PAN), an oral pan-HDACi, was the first-in-class approved drug for the treatment of MM by the FDA and the European Medicines Agency. 18,19 It was granted accelerated approval based on the outcomes of the progression-free survival (PFS) of a phase III study. After that, the required postapproval clinical studies could not be fulfilled and as a consequence PAN approval was withdrawn by the FDA in November 2021. However, it continues to be investigated in clinical trials for patients with relapsed/refractory MM (RRMM).
Other HDACi having shown promise against hematological malignancies have also been tested as anti-MM agents.
Among them, vorinostat (suberoylanilide hydroxamic acid, SAHA), a potent orally active pan-HDACi approved in 2006 by the FDA for advanced primary cutaneous T-cell lymphoma, is under investigation. 20 In addition, tinostamustine (EDO-S101), a fusion molecule of bendamustine (Benda) and SAHA, has been part of several preclinical studies reporting promising results and is discussed in a later section. 21 Despite showing improvements in MM patients, pan-HDACi treatment displays unpleasant side effects and toxicity. 22 Alternatively, the use of selective HDAC6i could be as beneficial as pan-HDACi with the additional advantage of a decreased toxicity profile. 23 Indeed, ricolinostat (RIC) is the first HDAC6i to reach clinical trials with results reporting an improved safety profile compared to pan-HDACi. 23 Notwithstanding the modest success based on selective HDACi, HDACi monotherapy has shown poor results overall. Interestingly, combination of these compounds with other standard treatments, especially with PIs, has shown synergistic effects both in preclinical and clinical settings. Almost 20 years ago, bortezomib (BTZ), a first generation PI, emerged and persisted as the backbone among anti-MM treatments in combination with a number of other drugs, including HDACi. 24 Since then, newer generation PIs, such as carfilzomib (CFZ), a second generation PI approved by the FDA in 2012, and ixazomib (IXZ), a third generation PI approved in 2015, have been evaluated as new treatment strategies in combination with HDACi in in vitro and in vivo studies, as well as in clinical trials. [25][26][27][28] It is well established that malignant cells exposed to proteasomal inhibition will clear misfolded proteins through the formation of aggresomes. Therefore, MM patients often develop resistance to PIs by means of the alternative aggresome pathway. 29 Aggresome formation occurs with the transport of cargo poly-ubiquitinated proteins along the microtubules network. Ubiquitin binding is mediated with high affinity by HDAC6. 29 Targeting both the proteasome and aggresome pathways by PIs and HDACi, respectively, induces the accumulation of polyubiquitinated proteins, which results in increased cellular stress and apoptosis. 30 In addition, IMiDs such as thalidomide (Thal), lenalidomide (Len) and pomalidomide (Pom) have made a profound difference in the treatment of MM and have also been lately combined with HDACi. 31 Likewise, standard chemotherapeutica drugs such as dexamethasone (Dex), doxorubicin (Dox) and MLP have been evaluated in combination with HDACi.
Drug combinations have become a well-established approach to maximize the efficacy of single drug treatments. Besides the drug combination approach, another strategy consists of designing dual-target agents such as HDACi/PI, with the added advantage of achieving inhibition of multiple pathways by a single molecular entity.
The use of multi-target drugs has shown synergistic or additive antitumor effects, and allowed to reduce the administered therapeutic doses, as well as led to a decrease in side effects and drug resistance development. 32,33 To the best of our knowledge, no other review has presented the key role of HDACi in combination with other drugs in MM treatment in the past couple of years. In vitro and in vivo studies, as well as clinical trials of drug combinations will be presented. The emerging use of dual-target inhibitors will also be discussed. Both strategies aim at targeting multiple pathways to improve the therapeutic effect and minimize the development of drug resistance.
T A B L E 1 Combination treatment consisting of HDAC and proteasome inhibitors.         | 2205 mitochondrial proteins (increased PARP cleavage, as well as cytosolic release of cytochrome c and Smac/ DIABLO). 21,41,42,44,45 Furthermore, the NF-ĸB signaling pathway was inhibited, aggresome and lysosome formation decreased, and ROS generation increased. 41,42,44 In RPMI 8226 or IM-9 cells, ATF4, Bim, BIP, CHOP, DR5, GADD34, GRP78, and XBP1 messenger RNA (mRNA) expression was upregulated. 21,44 Increased ac-H3, ac-α-tubulin, cleavage of p21 and p27, p53 and p-JNK protein expression were associated with SAHA/BTZ exposure. 21,41,42,45 Decreased protein expression of cyclin D1, Mcl-1 and XIAP was also reported. 41 In addition, Myc activation increased intracellular calcium levels induced by SAHA/BTZ leading to ER stress-mediated cell death. 42 (Table 1). Cell proliferation was greatly decreased upon treatment with BEL/BTZ compared to monotherapy, even when MM cells were co-cultured with patient BMSCs, demonstrating the ability of the combination to overcome the stimulatory effects of the BM microenvironment. 46 Moreover, the combination induced no toxicity in PBMCs. 46 For the majority of the combinations listed above, anti-MM activity was associated with increased apoptosis, mostly caspase-dependent. 41,46,49,54 In addition, increased ac-tubulin, Bim (EL, L, S), p21, p-p38 MAPK, p-p53, and p-H2AX protein expression and decreased Bcl-2 protein expression were noticed when BEL was used in combination with BTZ. 46 BEL/BTZ combination also showed synergistic inhibition of osteoclastogenesis and increased ROS production. 46 Furthermore, the combination of OLE/CFZ led to increased ac-H4 protein expression and to decreased HDACs 2, 3, 4, and 6, as well as Sp1 protein expression. 49 Enhanced ROS production was also reported for the combination NaB/BTZ, as well as increased cleavage of p21, p27, and p53  21 and oprozomib (OPZ), 21 but not with IXZ 21 (Table 1). However, EDO-S101/CFZ had antagonistic effects when tested in CFZ-resistant MM cells. 21 Decreased Bcl-2 and c-Myc mRNA expression and upregulated ATF4, BIP, CHOP, NOXA, p21, and XBP1 mRNA expression were observed upon EDO-S101/BTZ treatment. 21 In addition, increased levels of ac-H4, ac-α-tubulin, Bax/Bcl-2 ratio, BIP, cleaved PARP, CHOP, MAP1LC3A, MAP1LC3B, NOXA, p21, and p-H2AX protein expression, as well as decreased levels of p27 protein expression were also associated to the same combination. 21,55 EDO-S101/BTZ also induced cell cycle arrest in S phase, enhanced apoptosis and increased the accumulation of polyubiquitinated proteins in comparison to monotherapy. 21 Taken together, these results led to in vivo testing, whereby prolonged survival compared to single agent treatment was observed in a xenograft model 55 (Table 1).
Combinations between the class I selective HDACi entinostat (ENT) 32,56,57 or romidepsin (ROM) 58,59 and PIs (BTZ, CFZ, or K-7174) led to synergistic antiproliferative activity in MM cells, including those sensitive or resistant to BTZ and patient-derived MM cells (Table 1). ENT/BTZ led to cell cycle arrest in G2/M phase and enhanced apoptosis, while ROM/BTZ increased ac-H3, ac-H3K18, ac-H4, and ac-H4K12 protein expression. 32,56,58 Upregulated protein expression of ac-H3 was also seen for the combination of ROM with K-7174 as well as decreased HDACs 1, 2, and 3 protein expression. 59 ENT in combination with CFZ enhanced apoptosis as well as increased XBP-1 mRNA expression and CHOP protein expression. 57 The effectiveness of the combination of ROM/ BTZ was validated in an in vivo xenograft model resulting in the regression of BTZ-resistant tumors in comparison to single agent-treated group 58 (Table 1).
The selective class IIa HDACi TMP269 in combination with CFZ synergistically inhibited cell proliferation in various MM cell lines including patient-derived MM cells and cells sensitive or resistant to common treatments 60 (Table 1). Furthermore, induction of apoptosis by caspase activation and upregulation of proapoptotic protein expression such as ATF4, CHOP and cleaved PARP was also observed. 60 The HDACs 1, 2, 3 inhibitor BG45 enhanced the antiproliferative activity of BTZ in RPMI 8226 cells. 56   CIT/Pom, 70 EDO-S101/Len, 55 EDO-S101/Pom, 55  RIC/Pom 64,70 ) led to the suppression of MM cell proliferation, which was synergistic in all cases besides the SAHA/ IMID-1 and RIC/Len/Dex combinations (Table 2). Antiproliferative effects of the combinations were also seen in

patient-derived (IMiD-resistant or not) MM cells and in cells cultured with conditioned media harvested from
BMSCs. 64,68,69 CIT/Pom combination promoted cell cycle arrest compared to single agents, albeit to varying degrees in different MM cell lines. 70 Some of the combinations further increased apoptosis (synergistic in some cases) in MM cell lines and patient-derived cells, while no effect was observed in PBMCs. 38 Table 2).
HDACi have also been combined with chemotherapeutics. Combinations of PAN, 37 SAHA, 40 DAC, 73 and EDO-S101 55 with Dex reduced proliferation of MM cells, however, a synergistic effect was reported just for the latter combination. PAN, 72 SAHA, 40 and Quis 74 in combination with Dox showed increased antiproliferative activity, synergistic for PAN/Dox and Quis/Dox, with the former combination also leading to increased apoptosis.
In terms of combinations of HDACi and mAbs, two studies on the combination of RIC and daratumumab demonstrated that RIC enhanced antibody-dependent cellular toxicity in MM and patient-derived cells 76,77 ( Table 2).
HDACi have also been combined with several inhibitors of known targets. Combination of PAN with some of the compounds listed in Table 3, such (Table 2). Interestingly, decreased proliferation was also seen in MM.1S cells co-cultured with BMSCs, in the presence of PAN and AZD6244, as well as PAN and RAD001, suggesting that the BM microenvironment could not stimulate proliferation, while no effect of the combination was observed in BMSCs alone. 80 79,82 and was also observed when MM cells were co-cultured with stromal cells, 78,81,82 but was not seen in normal cells. 82 In the case of PAN/RAD001, increased mitochondrial damage and cytochrome c release into the cytosol were also observed. 87 The combination of PAN with FK506 inhibited osteoclast formation. 35  well as a reduction of human IgG levels, 79 decrease in β2-microglobulin, CHEK1, GPNMB, PLK1, and RANKL mRNA expression, 81 increased p-H2AX protein expression 82 and induction of apoptosis monitored by increased cleaved caspase 3. 35 Combinations of SAHA with agents such as GDC-0941, 89 JQ1, 90 NVP-AUY922, 83 P5091, 91 Pter, 86  Moreover, a synergistic inhibition was observed for the latter two combinations in some cell lines. 92,94 The VPA/Pio combination was reported to arrest cell cycle at G0/G1 phase, while also decreasing the expression of proteins related to the cell cycle such as cyclin A, cyclin D1 and p-Rb. Furthermore, VPA/Pio enhanced caspase-dependent apoptosis. 94 Potentiation of some of the VPA-induced effects by Pio could also be associated with higher acetylation levels of histones H3 and H4 compared to VPA alone or TSA, used as control. 94 Increased ac-H3 and ac-H4 protein expression was also observed for the VPA/Cig combination. 94 Combination of Quis with decitabine (Dec) synergistically reduced proliferation of MM cells and increased apoptosis, with increased cleaved Mcl-1. 96 Administration of this combination to a syngeneic mouse model showed enhanced survival, decreased tumor load in the BM and decreased serum M-spike 96 (Table 2).
ENT combined with AZD6244, Benda, BI-D1870, cladribine (Clad), sirolimus (SRL) and Ven showed synergistic antiproliferative activity in MM cell lines and patient cells, 80,97-100 while decreased viability observed with the ENT/ BI-D1870 combination was not synergistic 98 (Table 2). Decreased proliferation was also seen for MM cells in coculture with stromal cells, while no cytotoxicity was seen in normal PBMCs for the combination of ENT/SRL. 100 ENT/Benda and ENT/Clad combinations led to cell cycle arrest at S and G1 phases, respectively, with the second combination also causing mitotic catastrophe. 97,99 All combinations (with the exception of ENT/BI-D1870, but also including ENT/Bir) were reported to increase apoptosis. 80,82,97,99,100 Effects of ENT/Benda and ENT/Clad on proliferation and apoptosis were associated with increased p21, p-CHK2 and p-H2AX protein expression and decreased cyclin D1 and E2F-1 protein expression. 97,99 Effects of ENT/Bir were associated with decreased p52 and TRAF2 and increased TRAF3 protein expression. 82 Effects of ENT/SRL were associated with decreased Bcl-x L and p-histone H3 protein expression. 100 ENT/SRL positive effects were also observed in two xenograft mice models, in which tumor growth was decreased. Analysis of tumors showed increased ac-H3, and decreased Bcl-x L and p-Erk1/ 2 protein expression. 100 Nanatinostat (NAN), a class I HDACi, was studied in combination with TOS (  90 Combinations of RIC with Afu, FK506, JQ1, and S63845 increased apoptosis, even in co-culture with stromal cells, 78 which was associated with downregulation of Bcl-2, Bcl-x L , c-Myc, Mcl-1, pGSK-3 α/β, and PPP3CA protein expression and increased BAK activation. 35,77,78,90 Administration of RIC/JQ1 in mice injected with RPMI 8226 cells decreased tumor growth, increased cleaved caspase 3 and decreased Bcl-2, c-Myc and PCNA protein expression. 90 Finally, tubacin in combination with AZD6244 decreased cell proliferation in MM cells and in cells sensitive to common treatment, albeit nonsynergistically 80 ( Table 2).
Among the diverse combinations described so far, only PAN/MLP and SAHA/Len were tested in clinical trials.
Results will be presented in a later section.

| COMBINATIONS OF HDAC INHIBITORS AND PIs IN CLINICAL TRIALS
HDACi in combination with PIs demonstrated synergistic antiproliferative and proapoptotic activity in MM preclinical models leading to the initiation of various clinical trials ( months, while overall survival (OS) was not reached. Response rates appeared to be higher in patients whose last BTZ-containing regimen was not their last line of therapy before entering the study. Serious AEs were observed in 67% of patients. 103 In the second phase II clinical trial, 24 patients heavily pretreated with PIs and IMiDs were selected for the evaluation of the efficacy and tolerability of the drug combination. 22 Patients were treated following 2 different treatment schedules as in the previous study. 22 In the total cohort, PFS and OS were 3.5 and 9.8 months, respectively. Moreover, CBR was 52% and ORR was 33%, while 81% developed stable disease (SD).
Patients sensitive to BTZ had an ORR of 64% and significantly higher median PFS compared to BTZ-refractory (6.3 vs. 2.3 months). 22 The purpose of the PANORAMA 3 clinical trial was to investigate the activity and safety of three different dosing regimens of oral PAN in the PAN/BTZ/Dex indication. 106    IMiDs, such as Len and Thal, were also added to the PAN/BTZ/Dex combination [107][108][109] (Table 4) OS were not reached. 107 The same quadruplet drug combination was also evaluated in a phase Ib study involving 20 patients with a history of MM (with a median of 4 previous regimens including Len and BTZ). 108 The MTD of PAN was set at 10 mg based on the escalation phase. 108 Ten patients were then enrolled in the expansion phases for the evaluation of the safety and efficacy profile of the therapy. 108 ORR was 44%, CBR 61%, median PFS 7.4 months, while OS was not reached. 108 In a phase I/II trial named MUK-six, Thal was added to PAN/BTZ/Dex for improving the safety profile. 109 109 Overall, the treatment was well tolerated and only two patients withdrew consent because of toxicity. 109 Second and third generation PIs were also considered in combination with PAN in phase I and phase I/II clinical trials 110,111 (Table 4). In a phase I study, 32 patients with a median of 4 prior lines of therapy (including BTZ and Len) who had undergone autologous transplantation, were elected. 110 The major aim of the dose-escalation phase was to assess the MTD of PAN and CFZ in combination, which was set at 20 mg and 36 mg/m 2 , respectively. In the dose-expansion phase, ORR and CBR were 57% and 70%, respectively. Median PFS and OS were 8.0 and 23.0 months, respectively. No differences in terms of ORR, median PFS and OS were observed between BTZ-sensitive and refractory patients. 110 Another phase I/II clinical trial was designed to assess the safety and efficacy of PAN/ CFZ in 44 eligible patients, who had a median number of 5 prior therapy lines (including IMiDs, PIs and stem cell transplantation). 111 Thirteen patients were enrolled in the phase I and 31 in the phase II. 111 The main objective of phase I was to determine the MTD and of phase II to define the efficacy of the drug combination. MTD was not reached among the doses used. The doses for the expansion phase were established at 30 mg PAN and 20 mg/m 2 CFZ on days 1 and 2 of cycle 1 that were increased to 45 mg/m 2 for the rest of the treatment. ORR among all patients was 67% and CBR was 79%, with 6% of patients showing SD. Median PFS was 7.7 months, whereas OS was not reached. Grade 3/4 AEs in both studies included thrombocytopenia and fatigue. 110,111 Promising in vitro and in vivo results of SAHA in combination with standard MM agents led to the initiation of several clinical trials [112][113][114][115][116][117][118][119][120] (Table 4). A phase I trial evaluated 23 patients with RRMM and a history of 7 prior treatments including autologous transplantation, BTZ and IMiDs. 112 The MTD was reached at SAHA 400 mg and BTZ 1.3 mg/m 2 . The most common 3/4 AEs were myelosuppression, fatigue and diarrhea. There were no drugrelated deaths. 112 Thirty-four patients went through a phase I clinical trial after having received a median of 4 prior regimens comprising autologous transplantation, BTZ and IMiDs. 113 Some patients discontinued treatment due to FERRO ET AL.
| 2217 disease progression, lack of efficacy and toxicity. Therefore, MTD was not achieved. 113 SAHA/BTZ was tested in a phase I study evaluating 9 Japanese, 2 of them being BTZ refractory. 114 Three patients developed serious AEs and 6 patients discontinued therapy due to treatment-related clinical AEs. 114 Vantage 088 was a phase III study of SAHA or placebo (Pbo) in combination with BTZ in which 317 patients were allocated to the SAHA group and 320 to the Pbo group. 115 Median PFS was slightly longer in the SAHA group compared to the Pbo (7.6 vs. 6.8 months). 115 VANTAGE 095 was a phase IIb trial in which 143 patients, who underwent at least 2 prior drug regimens (including BTZ and IMiDs), were treated with SAHA/BTZ/Dex. ORR was 11%, CBR was 19%, PFS 3.1 months and OS was 11.2 months. SD occurred in 61% of patients. 116 Sixteen patients, with 1 median number of prior line of therapy with cyclophosphamide, IMiDs or Dex, participated in the phase II MUK four trial. 117 120 All patients included in the studies described previously, experienced treatment-related AEs. [114][115][116][117][118][119][120] The most common grade 3 or higher AEs were thrombocytopenia, anemia and neutropenia.
Twenty-five patients with RRMM treated with a median of 1 prior line of therapy (including autologous stem cells transplantation, BTZ, and/or IMiDs) participated in a clinical trial evaluating the safety and efficacy of ROM in combination with BTZ and Dex. 121 Six patients entered the dose-escalation phase after which the MTD of ROM was established at 10 mg/m 2 . Then, a total of 19 patients were treated at the MTD. The ORR among the 12 patients previously treated with IMiDs was 33%, and among the 13 patients not previously exposed to IMiDs, it was 77%.
The median OS was estimated at >3 years. 121 A total of 18 patients who had received prior anti-MM therapy and evidenced relapse or disease progression were enrolled in a dose-escalation phase Ib clinical trial of Quis in combination with BTZ/Dex. 122 Quis MTD was set at 10 mg. ORR was 88% and PFS was 8.2 months. 122 Thrombocytopenia was the most common grade 3 or higher hematologic toxicity followed by peripheral neuropathy. 121 RIC was combined with BTZ/Dex in a phase I/II study including 57 patients who had received at least 2 prior lines of therapy involving autologous stem cell transplantation, IMiDs or PIs 23 (  toxicities, data suggested that continuous treatment with PAN for 3 weeks may not be favorable, and suggested that 2 weeks on, 1 week off might be more suitable. 123 In a phase I/II dose-escalating combination study evaluating PAN/MLP, 40 patients were enrolled, 37 during phase I and 3 during phase II (with 4 median prior lines of therapy and 2 prior BTZ-based regimens). 124 Four treatment schedules, some with more than one cohorts, were applied.
Once MTD was determined at 20 mg PAN and 0.05 mg/kg MLP, three additional patients were enrolled as part of the expansion phase. 58% of patients showed SD, while 35% PD. Combination therapy was associated with a CBR and an ORR of 7.5% for both, which can be considered as a rather unsatisfactory efficacy. Median OS was not reached 124 (Table 5).
In a phase II study of PAN with Len and Dex, from the 27 relapsed MM patients (with a median of 3 prior lines of therapy) enrolled, six showed SD and one PD. Combination therapy was associated with an ORR and a CBR of 41% and 74%, respectively. Median PFS was 7.1 months, whereas median OS was not reached 125 ( Table 5).
The combination of SAHA and Len was reported in three clinical studies, with Dex co-administered in two of these ( Table 5). Thirty one patients (with a median of 4 prior lines of therapy) were elected to participate in a phase I study evaluating SAHA and Len combined with Dex. 127 The major aim of the dose-escalation phase (consisting of 5 treatment schedules) was to assess the MTD of the combination. no patient discontinuation was due to toxicity issues. 128 Based on the retrospective study, a phase IIb trial was conducted. Twenty-five Len-refractory patients (with a median of 5 prior regimens) were enrolled to further evaluate the efficacy of the response to orally combined SAHA/Len/Dex. ORR was 24%, SD was achieved in 52% of patients and CBR was 80%. 129 Median PFS was estimated to be 5.3 months, while median OS was not reached. It appeared that the presence of SAHA in the combination was rather effective for the Len-refractory patients.
Interestingly, the PAN/Len/Dex combination mentioned previously, 125 had similar CBR, but slightly higher ORR and PFS.
In a phase Ib study evaluating the selective HDAC6i RIC, 38 patients were randomized in a conventional doseescalation schedule (3 treatment schedules in total were used,  (Table 4). Neutropenia and thrombocytopenia were amongst the most common grade 3/4 AEs reported in the majority of the trials reviewed herein, [123][124][125][126]128,129 with patients also experiencing serious AEs. 124,127,129 Only a small number of clinical trials between HDACi and non-PI agents have been conducted so far. It is crucial that more of the successful in vitro/in vivo studies are channeled to clinical trials. It is very important to stress that the number of patients in the aforementioned studies was rather limited. In addition, PFS and OS were not always established. Taking these into consideration, RIC in combination with Len and Dex showed the longest median PFS, despite being a phase I study, 130 while the combination of SAHA with Len and Dex, as analyzed in a retrospective study, achieved the longest OS. 128

| IN VITRO AND IN VIVO STUDIES OF DUAL-TARGET INHIBITORS AND BIFUNCTIONAL AGENTS
During the 20th century the dominating doctrine for the development of therapeutics was "one diseaseone targetone drug." However, it is nowadays evident that for multifactorial diseases such as cancer, targeting only a single factor is not always efficient. Thus, the development of novel therapeutics, able to modulate multiple targets, is of increasing interest. An emerging approach is drug hybridization, which involves the covalent linking of two or more pharmacophores in one molecular entity, ultimately exhibiting synergistic and/or additive effects. Dual-target drugs can potentially delay or overcome drug resistance and show possible advantages compared to single-target therapies or combinations, such as better pharmacokinetic profiles, higher selectivity and efficacy, ease of use implying improved patient compliance, as well as reduced therapeutic doses and side effects.  induced apoptosis. Interestingly, in the BTZ-resistant cell line, ZY-2-induced apoptosis was higher than that caused by SAHA, ENT, BTZ and the combination of ENT and BTZ. 32 Ky-2, a class I HDAC inhibitor is the result of the combination between two HDACi, namely TSA and chlamydocin ( Figure 4). 131 Ky-2 has been shown to decrease mouse myeloma cell proliferation, while it had no effect on normal mouse cells. Furthermore, it arrested cell cycle at G1 through increased p21 and phosphorylation of Rb, as well as decreased Cdk6. Increased apoptosis was shown with cleaved caspase 3 and 9, and was reversed in cells over-expressing Bcl-2. 131  (situated in the solvent exposed region) with a variable spacer and the zing binding group (ZBG) of SAHA, two sets of candidate molecules were made, based on approaches A and B ( Figure 5).
Dual HDAC/JAK2 inhibitors 45h from "approach A" and 69c from "approach B" were selected among a series of compounds. Molecular docking studies of all three enzymes (HDAC1, HDAC6, and JAK2) with 69c could explain its activity; HDAC1/HDAC6 selectivity depended on the specific design and it could be controlled, while maintaining JAK2 activity. 132 Both hybrids inhibited HDAC6, as well as HDAC1 in the nM range. 132 Compound 69c had better JAK2 activity compared to 45h and exhibited 17-fold selectivity compared to JAK1 and 26-fold compared to JAK3 and TYK2. Furthermore, compound 45h inhibited KMS-12-BM cell proliferation.
The Bcl-2 protein family regulates the intrinsic apoptotic pathway through alterations in the expression of proand antiapoptotic proteins. HDAC/Bcl-2 dual target inhibitors were also designed using a common structural element to RIC and SAHA, as well as Ven, a Bcl-2 selective inhibitor. Indeed, earlier studies of the combinations of PAN and SBHA with ABT-737 showed encouraging synergistic effects, 79,93 as did the following studies of PAN, ENT, RIC, and ROM with Ven. 78,80 Docking studies of Ven with Bcl-2 showed that the tetrahydropyranyl methyl F I G U R E 4 Hybridization rationale for Ky-2, replacing the epoxyketone moiety of chlamydocin by the zinc binding group of TSA. [Color figure can be viewed at wileyonlinelibrary.com] group was exposed to the solvent and had less effect in the binding. Taking into consideration this result and the fact that in HDACi the CAP-part of the molecule can be tolerant to changes, Zhou et al. 133 designed a series of HDAC-Bcl-2 hybrids by replacing the tetrahydropyranyl methyl group with the spacer (bearing 1-7 carbon atoms) and the ZBG of the HDACi ( Figure 6).
This strategy was confirmed by the experimental results, indicating that the above replacement had little effect on Bcl-2 binding. The hybrid compounds with a linker containing 5-7 carbons (7e, 7f, 7g) showed selectivity against HDAC6 over HDAC1 (selectivity index >10), which is comparable to the one of RIC. 133 The hybrid molecule 7g increased more markedly ac-H3 expression, while all 3 compounds (7e-7g) increased ac-tubulin expression in RPMI 8226 cells (Table 6). All compounds showed antiproliferative activity (in the nM range for RPMI 8226 cells and better compared to RIC, SAHA, and Ven, while in the μM range for U266, Table 6).
P1, a hybrid of HDAC/aminopeptidase N (APN) inhibitors, more specifically of SAHA/bestatin, was initially designed as a potent HDAC degrader 134 ; degraders and proteolysis-targeting chimeras (PROTAC) are emerging strategies, but are beyond the scope of this review. P1 was more potent against HDAC 1, 6, and 8 compared to SAHA, and against APN compared to bestatin (Figure 7). P1 increased ac-H3 and ac-tubulin protein expression and decreased intracellular HDAC 1, 6, and 8 protein expression in RPMI 8226 cells, in a proteasome independent manner, hence suggesting it was not due to a PROTAC effect of the hybrid. 134 It was previously presented that SAHA, NAN, and VPA in combinations with an aminopeptidase inhibitor, TOS, had synergistic anti-MM effects. 92 Both studies strengthened that dual inhibition of both targets could display better activity.
Qian et al. 89  CUDC-907, a potent candidate among a number of HDAC/PI3K dual-target inhibitors of the library, inhibited class I (similarly to PAN and higher than SAHA) and class II HDAC activity, as well as class I PI3K activity (similar to the PI3K inhibitor GDC-0941). 89 Its antiproliferative activity against a number of MM cell lines was higher than SAHA alone or SAHA in combination with GDC-0941. It induced apoptosis (albeit comparable to that induced by PAN) and decreased p-Src and p-STAT3 protein expression in RPMI 8226 cells. 89  mitochondrial membrane potential. 137 CUCD-907 was also combined with the mAbs daratumumab and elotuzumab, leading to enhanced antibody-dependent cellular cytotoxicity. 77 Another dual-target agent in clinical trials is EDO-S101, a first-in-class alkylating agent and HDACi fusion molecule. It structurally combines SAHA with the strong DNA damaging effect of Benda ( Figure 9). It is intended to simultaneously induce DNA damage, while inhibiting DNA repair activity. 135 EDO-S101 had similar or slightly lower IC 50 compared to SAHA against classes I and II HDACs (Table 6). It also increased ac-H3 and downregulated the expression of DNA repair proteins p-ATM, p-ATR, and p-CHK2 in MM.1S cells. 135 These examples indicate that the development of HDACi-based dual-target agents requires, among others, a combination with a relevant second target, a balanced activity between both targets, maintenance of druggability, and avoidance of off-target interactions. However, dual acting HDACis are very promising compounds for the treatment of cancer and especially in cases where the single-target strategies are proven to face serious limitations, mainly due to drug resistance. Nowadays, the spotlight is on the dual inhibitor technology, which combines in a single molecule the merit of inhibiting two targets. Four dual targeting molecules have already entered clinical trials for different malignancies.
Two of these trials were conducted in MM patients and involved the hybrids CUDC-907 and EDO-S101, which have been described herein. To guide the design of future strategies, it is imperative to further synthesize and channel into the clinic more selective hybrids intended to minimize side effects, therapeutic doses and to address MM multidrug resistance. Beyond cancer, the hybrid approach can also benefit other pathological conditions such as neurological and age-related disorders, immune and cardiovascular diseases.

ACKNOWLEDGMENTS
Open access funding provided by Universite de Geneve.

DATA AVAILABILITY STATEMENT
Not applicable.