Non-small cell lung cancer

Non-small cell lung cancer (NSCLC) includes adenocarcinoma, squamous cell carcinoma (SCC), and large-cell carcinoma (LCC) and accounts for approximately 85% of all lung cancer cases [6]. The majority of diagnosed NSCLC cases present as locally advanced or metastatic diseases [7]. Twenty of the 22 therapeutic drugs are approved for NSCLC as the initial indication, and most of them are classified as epidermal growth factor receptor (EGFR) and anaplastic lymphoma kinase (ALK) inhibitors. Therefore, EGFR mutation and ALK rearrangement tests are recommended for NSCLC before EGFR- or ALK-directed therapies [8, 9] (Fig. 2a and Table ​Table11).

Open in a separate window

Fig. 2

FDA-approved therapeutic drugs for lung cancers. a Distribution of therapeutic drugs for lung cancers during the past 31 years (adapted from [126]). b Microtubule inhibitor. c EGFR inhibitors and EGFR-directed mAb. d EGFR- and MET-bispecific antibody and EGFR inhibitor for NSCLC with EGFR^ex20ins mutations. e ALK inhibitors. f MET inhibitors. g RET inhibitors. h Multitarget TKI. i KRAS^G12C-targeted small-molecule inhibitor. j DNA topoisomerase inhibitor and DNA alkylating agent for SCLC

Vinorelbine is recommended as an ingredient of systemic therapy regimens for neoadjuvant and adjuvant therapy of NSCLC. As a derivative of vinca alkaloid, it binds to tubulin in a complex with the RB3 protein stathmin-like domain (RB3-SLD), heavily overlapping the binding site of vinblastine [10, 11], thereby destabilizing α/β-tubulin heterodimers and leading to mitotic arrest and cell death [12] (Fig. 2b).

EGFR mutations occur in approximately 50% of Asian patients and 11~16% of patients in European countries with NSCLC [13–15]. Exon 19 deletion and exon 21 L858R point mutation make up the majority (>90%) of all EGFR mutation-positive NSCLC [16, 17], which frequently leads to lung tumorigenesis and sensitivity to EGFR-targeted therapies [18]. The FDA has approved six EGFR tyrosine kinase inhibitors (TKIs) which have been the first-line standard of care for patients with NSCLC harboring EGFR mutations [19]. These TKIs include the first-generation reversible EGFR inhibitors (gefitinib [20] and erlotinib [21]), the second-generation irreversible EGFR inhibitors (afatinib [22] and dacomitinib [23]), and the third-generation irreversible EGFR inhibitor (osimertinib [24]). First-generation EGFR inhibitors exert their clinical efficacy by targeting the ATP-binding pocket of the kinase domain [20, 21]. However, despite the initial response, patients almost invariably develop primary resistance to gefitinib and erlotinib and relapse after several months [25, 26]. The most common resistance mechanism is associated with the T790M ‘gatekeeper’ mutation at exon 20 of EGFR [27], which blocks reversible ATP competitive inhibitors from binding and, in turn, increases ATP binding [28]. Second-generation irreversible EGFR TKIs are highly active against the T790M point mutation of EGFR [26, 29] and exert their effect by irreversibly alkylating Cys797 and forming a covalent bond with Cys797 at the ATP-binding pocket [30], thus avoiding the increased ATP affinity conferred by the T790M gatekeeper mutation. However, EGFR T790M shares a similar ATP affinity with wild-type (WT)-EGFR, which limits the ability to achieve plasma concentrations sufficient to inhibit EGFR^T790M and results in skin rash and diarrhea in patients, thereby failing to overcome T790M-mediated resistance [31]. The third-generation irreversible EGFR inhibitor osimertinib shares a similar binding mechanism with second-generation irreversible EGFR inhibitors but exhibits lower activity against EGFR^WT, thereby overcoming the T790M-mediated TKI resistance [32]. As expected, osimertinib significantly prolongs median progression-free survival (PFS) by almost nine months compared with first-generation EGFR inhibitors [24]. However, acquired EGFR^C797S point mutation-induced impairment in the covalent binding between EGFR^Cys797 and osimertinib and acquired MET amplification induced activation of the bypass pathway [33] lead to resistance to osimertinib [34]. Additionally, necitumumab is a fully human anti-EGFR IgG1κ that binds specifically to EGFR domain III, which overlaps with the EGF binding site, thereby preventing EGF ligands from binding to EGFR [35]. Thus, necitumumab was approved for first-line treatment (in combination with gemcitabine and cisplatin) for patients with metastatic squamous NSCLC [36] (Fig. 2c). Notably, necitumumab binds to most cetuximab- and panitumumab-resistant EGFR variants, such as EGFR^S440L and EGFR^S468R [37].

EGFR exon 20 insertion (EGFR^ex20ins) is clustered between codons 762–775, such as A767_V769dup (V769_D770insASV) and S768_D770dup (D770_N771insSVD) [38, 39]; it represents approximately 6~12% of EGFR mutations in NSCLC cases [40–43] and frequently leads to the constitutive activation of EGFR [38]. Most EGFR^ex20ins driver mutations in NSCLC are insensitive to first- and second-generation EGFR inhibitors [44–46], except osimertinib, which exhibits partial activity against some EGFR^ex20ins driver mutations in preclinical studies [39, 45, 47]. However, the clinical trials of osimertinib are inadequate and yield contradictory results [48, 49].

Amivantamab is a bispecific IgG1 that targets both EGFR and MET produced from the two purified bivalent parental antibodies by controlled Fab-arm exchange, each containing single matched point mutations in the CH3 domains (K409R and F405L) [50, 51]. The amivantamab EGFR H-arm shares an epitope identical to that of zalutumumab and binds to EGFR domain III, which overlaps with the EGF binding site, while the MET arm of amivantamab binds to the MET Sema region, which overlaps with the hepatocyte growth factor (HGF) binding site [52]. Amivantamab exhibits antitumor efficiency through the Fc-dependent antibody-dependent cellular cytotoxicity (ADCC) mechanism, Fc-independent EGFR/MET inactivation/degradation and blockade of downstream signaling transduction, and increased interferon-γ (IFNγ) secretion [44, 53, 54]. It yielded robust and durable responses with tolerable safety in patients with EGFR^ex20ins mutations who progressed on or after platinum-based chemotherapy [55].

Designing a novel EGFR inhibitor is another strategy to address EGFR^ex20ins mutations. However, the conformation of EGFR^ex20ins mutants largely resembles that of EGFR^WT proteins because there are no amino acid substitutions in the binding site [39, 56]. Mobocertinib is an irreversible EGFR inhibitor that is structurally similar to osimertinib. It targets potential structural nuances between the EGFR^ex20ins and EGFR^WT proteins in the vicinity of the α C-helix to gain selectivity by binding to the portions of the binding site that are not exploited by osimertinib [39]. Mobocertinib demonstrates greater activity against EGFR^ex20ins mutants than EGFR^WT and more potent efficacy than erlotinib, gefitinib, afatinib, or osimertinib against EGFR^ex20ins mutants, except EGFR^C797S-containing triple mutants [39, 57]. In subsequent clinical trials, mobocertinib exhibited potent activity with manageable toxicity in patients with advanced previously treated EGFR^ex20ins NSCLC [58, 59] (Fig. 2d).

Both aberrant ALK expression caused by ALK rearrangements [60] and ALK amplification are oncogenic driving factors of NSCLC [61]; for example, gene fusion of EMAP-like protein 4 (EML4) and ALK induced by ALK rearrangements encodes a cytoplasmic chimeric protein with constitutive kinase activity, which accounts for 3~13% of NSCLC [62, 63]. The FDA has approved five ALK inhibitors, which have been the first-line standard of care for patients with NSCLC harboring ALK rearrangements [17], including the first-generation ALK inhibitor (crizotinib [64, 65]), the second-generation ALK inhibitors (ceritinib [66], alectinib [67], and brigatinib [68]), and the third-generation ALK inhibitor (lorlatinib [69]). As with EGFR inhibitors, acquired drug resistance inevitably occurs in most patients after treatment with ALK inhibitors [70, 71]. The mechanisms of ALK inhibitor resistance also involve on-target mechanisms (e.g., ALK mutations and amplification) and off-target mechanisms and are even more complicated [72]. Approximately 20~36% of crizotinib-resistant NSCLCs harbor ALK mutations, including 1151Tins, L1152R, C1156Y, I1171T/N/S, L1196M, G1202R, S1206C/Y, E1210K, and G1269A mutations [17, 70–74]. Regarding second-generation ALK inhibitors, ALK mutations account for more than half of the instances of resistance [72]. Specifically, 1151Tins, L1152P, C1156Y, F1174C/L/V, and G1202R mutations confer resistance to ceritinib [17, 72, 74], while I1171T/N/S, V1180L, L1196M, and G1202R mutations confer resistance to alectinib [72]. In addition, G1202R, D1203N, S1206Y/C, and E1210K mutations are associated with resistance to brigatinib [17, 72]. Thus, the G1202R mutation is the most common mechanism of first- and second-generation ALK inhibitor resistance. Fortunately, G1202R mutation-induced resistance can be overcome by the third-generation ALK inhibitor lorlatinib [75], which is active against the EML4-ALK^G1202R mutation [76]. Intriguingly, acquired C1156Y and L1198F mutations after lorlatinib treatment resensitize the tumor to crizotinib [69]. However, the off-target mechanisms of ALK inhibitor resistance are still under exploration [77] (Fig. 2e).

Mesenchymal–epithelial transition gene (MET) exon 14 skipping mutations and MET amplification occur in approximately 3~4% [78–80] and 1~6% [81–83] of patients with NSCLC, respectively [84]. MET exon 14 skipping mutations produce a truncated MET with a missing regulatory domain that disrupts ubiquitin-mediated degradation, resulting in increased MET levels, sustained MET activation, and oncogenesis [85]. Thus, MET exon 14 skipping mutations and MET amplification act as oncogenic-driven factors and confer EGFR inhibitor resistance to various cancers, including NSCLC, making it a promising therapeutic target [86]. Capmatinib is a highly selective, reversible type Ib MET inhibitor that targets MET and its mutants (M1250T and Y1235D) [87, 88]. It is more potent than other MET inhibitors (approximately 30 and five times more potent than crizotinib and tepotinib in vitro, respectively) [89]. Capmatinib directly binds to the phenol moiety of the MET^Y1230 residue, while MET^D1228 forms a salt bridge with MET^K1110 to support the Y1230–capmatinib interaction, similar to crizotinib (Type Ia MET inhibitor) [88]. Capmatinib occupies the ATP-binding site of MET, blocks MET phosphorylation, and inhibits MET-mediated downstream signaling activation [88]. Capmatinib exhibits substantial antitumor activity in patients with advanced NSCLC harboring MET exon 14 skipping mutations and MET amplification [84, 85]. Additionally, capmatinib reverses MET-dependent EGFR inhibitor resistance and blocks the signaling pathway activation mediated by EGFR and HER3 [87]. Significant resistance was observed in cells and clinical NSCLC cases bearing MET^D1228 and MET^Y1230 mutations due to the structural model of the MET–capmatinib interaction [88, 90]. Tepotinib is another selective, reversible type Ib MET inhibitor for a similar clinical setting to capmatinib. Tepotinib shares a similar mechanism with capmatinib in blocking MET [91]. Thus, they achieved equivalent clinical outcomes and adverse events [84, 92, 93]. In vitro, tepotinib overlaps the most MET mutation-induced resistance with capmatinib, especially MET^Y1230 mutations, suggesting that tepotinib may not overcome capmatinib resistance [90]. Compared with standard chemotherapy, tepotinib plus gefitinib exhibits improved antitumor activity in patients with EGFR-mutant NSCLC with MET overexpression or MET amplification [94] (Fig. 2f).

The rearranged during transfection (RET) gene rearrangements occur in approximately 1~2% of patients with NSCLC [95], which is frequently associated with brain metastases [96]. Two selective RET inhibitors (selpercatinib and pralsetinib) were approved as first-line treatments for patients with NSCLC harboring RET rearrangements [97, 98]. Selpercatinib and pralsetinib are designed to penetrate the central nervous system (CNS), thereby achieving poor CNS concentrations sufficient to maintain antitumor activity [99]. Both selpercatinib and pralsetinib exhibit activity against acquired RET^V804M/L gatekeeper resistance mutations [100, 101]. However, RET^G810C/S solvent front mutations (on-target) and MET amplification (off-target) were observed in selpercatinib- and pralsetinib-resistant cases [102–105]. Selpercatinib and pralsetinib bind to the RET kinase in a similar mode that occupies both front and back pockets in the active site clefts without passing through the gate between V804 and K758 into the BP-I pocket [106]. This novel binding mode avoids gatekeeper V804M/L mutation-induced resistance but fails to overcome RET mutations in G810 and V738 [106]. New-generation RET inhibitors are needed for this clinical dilemma. Fortunately, selpercatinib plus crizotinib therapy may be an available strategy to overcome selpercatinib resistance in RET fusion-positive NSCLC with MET amplification [104] (Fig. 2g).

ROS proto-oncogene 1 (ROS1) rearrangements occur in approximately 1% of patients with NSCLC [107, 108]. Crizotinib has been the first-line therapy for patients with metastatic ROS1 fusion-positive NSCLC since 2016 [108, 109]. However, 47% of patients with ROS1-positive NSCLC develop brain metastases upon crizotinib treatment because of crizotinib’s poor CNS penetration due to P-glycoprotein-mediated efflux [110–112]. In addition, the ROS1^G2032R mutation is frequently observed in NSCLC with acquired resistance to crizotinib [113]. Entrectinib is a multitarget TKI that targets ROS1, tropomyosin receptor kinases (TRKs) (encoded by neurotrophic tyrosine receptor kinase (NTRK) genes), and ALK [114]. Compared with crizotinib, entrectinib is a weak substrate of P-glycoprotein that is 30 times more potent against ROS1, thereby overcoming P-glycoprotein-mediated efflux and achieving high CNS concentrations [112, 114, 115]. However, ROS1^G2032R and ROS1^F2004C/I mutations are also found in NSCLC with acquired resistance to entrectinib [116]. In addition, more mechanisms of entrectinib resistance are being identified; these include NTRK1^G595R and NTRK1^G667C mutations in colorectal cancer [117], NTRK3^G623R mutation in mammary analog secretory carcinoma (MASC) [118], and insulin-like growth factor-1 receptor (IGF1R) activation and increased P75 expression in neuroblastoma [119]. These findings present new clinical challenges (Fig. 2h).

KRAS, one of the most frequently mutated oncogenes in various cancers, was once considered an undruggable protein due to its small size, relatively smooth surface, and rapid and tight binding properties to GTP in its active state [120]. KRAS^G12C is an oncogenic driver mutation that occurs in approximately 13% of patients with NSCLC [121]. Sotorasib is the first and only KRAS^G12C inhibitor that binds to KRAS^G12C via the cysteine residue mutated from the glycine residue, locking KRAS in an inactive state [120, 122, 123]. Sotorasib provides durable clinical benefits in previously treated patients with NSCLC, making it a milestone in cancer therapy [121]. Nevertheless, acquired resistance to sotorasib inevitably occurs via both on- and off-target mechanisms in most patients [124, 125]. G12C/R68S and G12C/Y96C/A double mutants and the G12D mutant of KRAS confer on-target resistance to sotorasib [124]. MET amplification is detected in sotorasib-resistant subclonal NSCLC cells with KRAS^G12C mutation in vitro. Thus, sotorasib plus crizotinib therapy may be a potential strategy to combat off-target resistance [125] (Fig. 2i).

Cytotoxic drugs for breast cancer

Cytotoxic drugs are still widely used in clinical practice, especially in systemic chemotherapy for recurrent unresectable (local or regional) human epidermal growth factor receptor 2 (HER2)-negative breast cancer and other malignancies [133]. Among these cytotoxic drugs, docetaxel represents one of the most notable microtubule-stabilizing agents. Docetaxel shares the same taxane binding site of β-tubulin with its analog paclitaxel [136] but shows more potent antitumor activity [137]. It exerts its activity by binding to free β-tubulin and inducing microtubule polymerization, resulting in cell cycle arrest and death [138, 139]. Ixabepilone is a β-lactam analog of epothilone B and is also classified as a microtubule-stabilizing agent. It binds tubulin in a similar but not identical manner to that of paclitaxel and exhibits potent cytotoxic activity in paclitaxel-resistant cells harboring P-glycoprotein expression or mutant tubulin [140]. In contrast, eribulin is a microtubule-destabilizing agent that terminates protofilament elongation by binding predominantly to the vinca domain on β-tubulin, resulting in microtubule catastrophes [141] (Fig. 3b). Capecitabine, a prodrug of 5-fluorouracil (5-FU), is first metabolized to 5′-deoxy-5-fluorouridine (5′DFUR) by carboxylesterase and cytidine deaminase in the liver. 5′DFUR is converted to 5-FU by thymidine phosphorylase (TP) and/or uridine phosphorylase (UP). Given the significantly higher concentrations of both TP and UP in tumor tissues than in normal tissues [142–144], the formation of 5-FU and the subsequent production of active metabolites, including fluorodeoxyuridine monophosphate (FdUMP), fluorouridine triphosphate (FUTP), and fluorodeoxyuridine triphosphate (FdUTP), preferentially occur in tumor tissues [145]. These metabolites finally lead to cell injury by attenuating thymidylate synthase activity (by FdUMP) and incorporating fraudulent bases into RNA (via FUTP) and DNA (via FdUTP) [146] (Fig. 3c). Epirubicin is a 4′-epimer of anthracycline antibiotic doxorubicin that exhibits at least equipotent cytotoxicity but is less myelotoxic than doxorubicin [147]. It binds to topoisomerase IIα (TOP2A), which interferes with helicase activity and TOP2A-DNA cleavable complex formation, resulting in irreversible DNA double-stranded breaks (DSBs) and gene transcription inhibition [148, 149] (Fig. 3d).

ER- or HR-positive breast cancer

Hormone receptor (HR)-positive breast cancers, including estrogen receptor (ER)- and/or progesterone receptor (PR)-positive breast cancers, account for more than 70% of all breast cancer cases [150, 151] and lead to approximately 50% of breast cancer-induced deaths [152]. Selective ER modulators (SERMs), such as tamoxifen (brand name: Nolvadex, approved on Nov. 30, 1977, by the FDA), have been the standard of care for patients with ER-positive breast cancer for over 40 years. At present, aromatase inhibitors/inactivators, SERMs, selective ER degrader/down-regulator (SERD), and cyclin-dependent kinases 4/6 (CDK4/6) inhibitors are the first-line standard of care for patients with HR-positive and HER2-negative breast cancers [153].

In premenopausal women, estrogens are mainly synthesized in the ovaries. In postmenopausal women, however, estrogens are synthesized in adipose tissue, breast, and skin, and this process is mediated by aromatase [154]. As a member of the P450 superfamily, aromatase (encoded by CYP19) is expressed at extragonadal sites, such as adipose tissue, breast, vascular tissue, bone, brain, and skin, in postmenopausal women [154, 155]. It converts androstenedione and testosterone released from ovaries and adrenal glands to estrone (E1) and E2, respectively [156]. Based on this principle, three third-generation aromatase inhibitors have been developed and approved for postmenopausal women with ER-positive breast cancer [153]. The reversible nonsteroidal aromatase inhibitors anastrozole and letrozole are triazole derivatives that exert clinical efficacy by binding to the heme prosthetic group of aromatase [157, 158]. In contrast, the irreversible aromatase inactivator exemestane binds to the substrate-binding pocket of aromatase, leading to its degradation [159–161]. Among the third-generation aromatase inhibitors, letrozole exhibits the most potent inhibitory effect on aromatase enzyme activity in vivo [159, 162, 163]. It is consistently 10~30 times more potent than anastrozole in inhibiting intracellular aromatase [164]. Nevertheless, conflicting results exist in various independent studies on clinical efficacy [165–167]. These contradictory results are potentially correlated with the mutation status of GATA binding protein 3 (GATA3) [168] or the saturation effect (all third-generation aromatase inhibitors reproducibly cause~98% aromatase inhibition in humans) [161] (Fig. 3e).

Toremifene is a SERM structurally similar to tamoxifen, differing only by a single chlorine atom [169]. Like tamoxifen, toremifene exerts pharmacological activity by competitively inhibiting estradiol (E2) binding to the ER. It thus cannot be used as second-line therapy after tamoxifen failure due to similar pharmacological mechanisms [170]. In contrast, fulvestrant is a full ER antagonist approved as a SERD that overcomes the agonistic effects of tamoxifen and toremifene [171, 172]. However, because of its poor physicochemical features, fulvestrant must be administered monthly intramuscular injections, limiting its clinical application [173]. Mechanistically, it has recently been proven to exert its properties by markedly impairing the intranuclear mobility of the ER [152] (Fig. 3f).

The formation of the cyclin D–cyclin-dependent kinases 4/6 (CDK4/6) complex (also known as G1-CDK) and CDK4/6-induced retinoblastoma (RB) phosphorylation are core events of the G1-S transition in the cell cycle [174]. Inhibition of CDK4/6 induces RB hypophosphorylation and reactivation, resulting in stable cell cycle arrest in the G1 phase [175]. Three CDK4/6 inhibitors (palbociclib, ribociclib, and abemaciclib) have been approved for the first-line therapy of patients with HR-positive and HER2-negative breast cancers in combination with nonsteroidal aromatase inhibitors [176–179] or SERD (fulvestrant) [180, 181] (Fig. 3g), thereby delaying or overcoming endocrine resistance [182]. Although three CDK4/6 inhibitors share multiple similarities, unique characteristics exist in each of them [182]. Palbociclib primarily targets CDK4 monomers instead of endogenous CDK4 trimer complexes or CDK6 but promotes the formation of inactive CDK2 complexes [183]. Palbociclib and ribociclib are more selective for CDK4/6 than abemaciclib, probably due to the greater lipophilicity and larger binding site side chains than abemaciclib, which may reduce the probability of interaction with off-target kinase ATP-binding pockets [184, 185]. Ribociclib is less potent than palbociclib and abemaciclib in inhibiting RB phosphorylation [184]. In contrast, abemaciclib binds to the ATP cleft more readily and forms a hydrogen bond with the conserved catalytic residue (Lys43) of CDKs, which decreases its selectivity [184, 185].

HER2-positive breast cancer

HER2-positive breast cancer (including some luminal B subtype cancers) accounts for 13~15% of all breast cancer cases [186] and is associated with aggressive and metastatic behavior [187]. As the first mAb to be approved to treat solid tumors, trastuzumab is a landmark in tailored therapies [133]. Trastuzumab binds to the extracellular region of HER2 on the C-terminal portion of domain IV and exerts its function via several mechanisms, including ADCC, inhibition of HER2 shedding, and disruption of ligand-independent downstream cascades [188–191]. However, trastuzumab is insufficient to block ligand-induced HER2/HER3 dimerization [191]. In contrast, pertuzumab binds to the extracellular domain II of HER2 and blocks both ligand-dependent and ligand-independent HER2/HER3 dimerization and activation [191–193]. The addition of pertuzumab to the combination of trastuzumab plus docetaxel significantly improves median PFS, and overall survival (OS) compared to that with a pertuzumab-free regimen [194, 195]. Margetuximab, as the latest approved HER2 mAb, improves the ADCC effect in HER2-low tumors with enhanced targeting activity and overcomes trastuzumab resistance [196] (Fig. 3h). Compared with trastuzumab plus chemotherapy, margetuximab plus chemotherapy significantly improves PFS in HER2-positive patients who have received two or more prior anti-HER2 therapies [197].

Trastuzumab significantly improves the clinical outcomes of patients with HER2-positive breast cancer [198]. In the metastatic setting, however, resistance to trastuzumab and disease progression occurs in most patients treated with trastuzumab within one year [199, 200]. The general mechanisms of trastuzumab resistance refer to obstacles for trastuzumab–HER2 interaction, reactivation of HER2 downstream signaling pathways, initiation of bypass signaling pathways, and failure to trigger immune-mediated mechanisms [201]. For this reason, two HER2-based ADCs (ado-trastuzumab emtansine and trastuzumab deruxtecan) were introduced (Fig. 3i). Ado-trastuzumab emtansine is composed of trastuzumab and DM1, linked with a non-cleavable thioether linker, N-succinimidyl-4-(N-maleimidomethyl) cyclohexane-1-carboxylate (SMCC, designated MCC after conjugation) [202, 203]. DM1 is a derivative of maytansine isolated from various Maytenus species [204] that exerts antitumor activity by destabilizing microtubules [205]. Ado-trastuzumab emtansine retains all the antitumor efficiency of trastuzumab and is active against lapatinib-resistant breast cancer cells and lapatinib-insensitive tumors [203]. Ado-trastuzumab emtansine shows significant clinical advantages over lapatinib plus capecitabine [206] and trastuzumab plus docetaxel [207]. Despite these therapeutic advances, most patients treated with ado-trastuzumab emtansine eventually experience disease progression [208, 209]. The resistance mechanisms of ado-trastuzumab emtansine partially overlap with those of trastuzumab but also include P-glycoprotein overexpression and receptor-mediated endocytosis defects [210, 211]. Trastuzumab deruxtecan is composed of trastuzumab and TOP1 inhibitor payload (Dxd, an exatecan derivative) linked with a protease-cleavable maleimide tetrapeptide linker [212]. Trastuzumab deruxtecan exhibits durable antitumor activity in patients previously treated with ado-trastuzumab emtansine [213]. Recently, the phase 3 DESTINY-Breast03 trial (NCT03529110) demonstrated that trastuzumab deruxtecan exhibits superiority over trastuzumab emtansine in patients previously treated with the trastuzumab plus taxane regimen [214].

Moreover, three HER2 inhibitors (lapatinib, neratinib, and tucatinib) were approved as third-line regimens for the treatment of HER2-positive breast cancer in combination with trastuzumab and/or capecitabine [215] (Fig. 3j). In contrast to HER2-directed mAbs, HER2 inhibitors bind to the cytoplasmic tyrosine kinase domain instead of the extracellular region of HER2. Lapatinib is a potent dual inhibitor of both EGFR and HER2 [216, 217] that exerts antitumor activity by reversibly binding to the cytoplasmic ATP-binding sites of EGFR and HER2, leading to the impediment of tyrosine kinase phosphorylation, which dampens or abrogates the activation of HER2-mediated downstream pathways [218]. Intriguingly, lapatinib also reverses P-glycoprotein- and ABCG2-mediated multidrug resistance (MDR) by directly attenuating their transport activity [219]. However, HER2^T798M/I gatekeeper mutations and bypass signaling pathway initiation inevitably confer resistance to lapatinib [220–222]. Neratinib is an irreversible inhibitor of EGFR, HER2, and HER4 that binds to the conserved Cys773 of EGFR and Cys805 of HER2, which forms a covalent bond with the HER family at the cleft of the ATP-binding site [223, 224]. Neratinib exhibited substantial clinical activity in patients with and without prior trastuzumab treatment [225], while the neratinib plus paclitaxel regimen yielded higher complete pathological response rates than the trastuzumab plus paclitaxel regimen in patients with HER2-positive, HR-negative breast cancer [226]. However, the HER2^T798I gatekeeper mutation also confers resistance to neratinib [227], suggesting that neratinib may not overcome lapatinib resistance, although it displays nanomolar antiproliferative activity against this mutant in vitro [228]. Tucatinib is another reversible HER2 inhibitor that shares a similar binding mechanism with lapatinib but exhibits the highest selectivity to HER2 among these HER2 inhibitors [229]. Compared to placebo, tucatinib’s addition to the trastuzumab plus capecitabine regimen exhibited acceptable toxicity [230], improved survival outcomes, improved objective response rate (ORR), and reduced the risk of death [231, 232]. Among these HER2 inhibitors, neratinib exhibits the most potent activity against HER2 kinase, followed by tucatinib and lapatinib [228].

BRCA-mutated breast cancer

Germline mutations of BRCA1 and/or BRCA2 are observed in more than 5% of all breast cancer cases and approximately 13% of basal-like breast cancer (BLBC) cases [233]. BRCA1/2 mutations frequently indicate a deficiency in repairing DNA DSBs by homologous recombination [234] and predispose patients to breast, ovarian, and other cancers [235–237]. Poly (ADP-ribose) polymerases (PARPs) are essential for DNA single-strand break (SSB) repair by base excision repair (BER) [238]. The N-terminal zinc finger motifs of PARPs bind to damaged DNA, which activates its catalytic C-terminal to hydrolyze nicotinamide adenine dinucleotide (NAD⁺) and produce ADP-ribose units, thereby yielding linear and branched poly (ADP-ribose) (PAR) for the resealing of DNA SSBs during BER [239, 240]. PARP inhibitors are designed to inhibit auto-PARylation by competitively binding to PARPs at the NAD⁺ binding site [241, 242], leading to cell death in BRCA1/2-mutated cancer cells through a synthetic lethality mechanism [243]. Breast cancers harboring germline mutations in either BRCA1 or BRCA2 are highly sensitive to PARP inhibitors [244, 245], and thus, inhibiting PARPs has become a therapeutic strategy for targeting BRCA1/2-mutated cancer cells [246]. Talazoparib is the fourth (also the latest) PARP inhibitor approved by the FDA (after olaparib, rucaparib, and niraparib) [247]. Through hydrogen-bonding and π-stacking interactions, including those mediated by active site water molecules, talazoparib is anchored to the nicotinamide-binding pocket [248], leading to a noticeable displacement of the bound ligand within the NAD⁺ site [249]. Compared with olaparib, rucaparib, and niraparib (IC₅₀ values 1.94, 1.98, and 3.8 nM, respectively, for the inhibition of PARP1), talazoparib is three times more potent, with an IC₅₀ of 0.57 nM [250, 251]. Therefore, talazoparib exhibits superiority over olaparib or rucaparib in trapping PARP–DNA at the site of DNA damage [239]. Talazoparib monotherapy demonstrates a tolerable safety profile and preliminary clinical activity in patients with sporadic cancers harboring germline BRCA1/2 mutations [252]. It also exhibits a significant benefit over standard chemotherapy (capecitabine, eribulin, gemcitabine, or vinorelbine) among patients with germline BRCA1/2-mutated breast cancer [253] (Fig. 3k).

PIK3CA-altered breast cancer

Phosphatidylinositol 3-kinase catalytic subunit A (PIK3CA) gene mutation is observed in approximately 40% of HR-positive and HER2-negative breast cancers [233, 254, 255]. PIK3CA mutation induces phosphatidylinositol 3-kinase (PI3K) activation, leading to cell proliferation and apoptosis evasion [233]. Alpelisib is a PI3Kα inhibitor that binds to PI3Kα and forms multiple hydrogen bonds with PI3Kα at the ATP-binding pocket, thereby inhibiting the enzymatic activity of PI3Kα and PI3Kα-mediated downstream pathways [256, 257]. Alpelisib demonstrates tolerable safety and favorable clinical efficiency in patients with PIK3CA-altered, HR-positive, HER2-negative breast cancer in combination with fulvestrant [255, 258–260] or letrozole [261] (Fig. 3l).

Ovarian cancer

Ovarian cancer is the third most common gynecologic cancer, accounting for 3.4% of all female malignancies and 4.7% of cancer-related deaths in females worldwide in 2020 [2]. FDA has granted four new therapeutic drug approvals for ovarian cancer. Paclitaxel is undoubtedly a milestone in the history of cancer drugs. It was isolated by Wall and Wani from the bark of Taxus brevifolia in 1971 [279] (Fig. 4b). Currently, paclitaxel (including nab-paclitaxel albumin-bound) and its analog docetaxel are widely used to treat various malignancies [280]. It covalently binds to β-tubulin at amino acid residues 1–31 [281], 217–233 [282], and Arg282 [283] and enhances microtubule polymerization, thereby suppressing microtubule dynamics and blocking cell mitosis [284].

Germline mutations of BRCA1 and/or BRCA/2 are present in approximately 14.1% of all ovarian cancer cases [285]. Based on the same principle described in the breast cancer section above, three PARP inhibitors (olaparib [286], rucaparib [287], and niraparib [250]) were approved as maintenance therapies for BRCA1/2-mutated ovarian cancer [288–290] (Fig. 4c). These PARP inhibitors are designed to competitively bind to the NAD⁺ binding site of the PARP enzyme [237, 238]. Platinum and PARP inhibitor sensitivity commonly coexist in BRCA1/2-mutated ovarian cancer due to homologous recombination deficiency (HRD) [291]; however, nucleotide excision repair (NER) alterations confer enhanced platinum sensitivity but not PARP inhibitor sensitivity [292]. There is no significant efficacy difference between these PARP inhibitors as maintenance therapies in patients with BRCA-mutated, platinum-sensitive relapsed ovarian cancer [293]. Additionally, PARP inhibitors yield similar response and survival rates in patients harboring either somatic or germline BRCA mutations [294]. Of note, olaparib represents the most cost-effective [295] PARP inhibitor, and the olaparib plus bevacizumab regimen achieved a dramatic improvement in PFS in ovarian cancer patients with BRCA mutations (37.2 months) and without BRCA mutations (28.1 months) compared to that with placebo plus bevacizumab (17.7 and 16.6 months, respectively) [296]. Thus, the olaparib plus bevacizumab regimen was approved for first‐line maintenance treatment of HRD-positive advanced ovarian cancer [297]. Clinical trials of PARP inhibitors (rucaparib and niraparib) combined with bevacizumab for ovarian cancer maintenance therapy are still ongoing [298, 299].

Endometrial cancer

Endometrial cancer is the second most common gynecologic cancer and originates in the inner epithelial lining of the uterus [300]. It accounted for 4.5% of all malignancies among females and 2.2% of cancer-related deaths in females worldwide in 2020 [2]. Mismatch repair deficiency (dMMR) is a consequence of germline mutations or epigenetic silencing in MMR genes, resulting in the accumulation of errors introduced during DNA replication [301]. Therefore, dMMR leads to genome-wide instability, especially in regions of simple repetitive DNA sequences (known as microsatellite instability—high (MSI-H)), resulting in tumorigenesis [302]. MSI-H/dMMR is observed in 18~28% of endometrial cancer cases [303, 304] and confers sensitivity to PD1 blockade [304]. A higher number of CD3⁺ and CD8⁺ TILs and increased PD1 expression (but not PDL1) are observed in the hypermutated subgroups (POLE mutations or MSI-H/dMMR) of endometrial cancer than in the hypomutated microsatellite-stable subgroup [305], explaining why MSI-H/dMMR-positive endometrial cancer is sensitive to PD1 blockade. Dostarlimab is the fourth (also the latest) FDA-approved PD1-directed mAb after pembrolizumab, nivolumab, and cemiplimab [306]. It exhibits a high affinity for both human and cynomolgus monkey PD1, preventing PDL1 and PDL2 from interacting with PD1 [307]. Dostarlimab demonstrates a manageable safety profile equivalent to that of other PD1-directed mAbs and robust clinical activity in previously treated patients with recurrent or advanced MSI-H/dMMR or MMR proficient/stable (MMRp/MSS) endometrial cancer [308, 309] (Fig. 4d). Of note, dostarlimab achieved a complete response in 100% of patients with dMMR-positive locally advanced rectal cancer [310].

Esophageal cancer

Esophageal cancer accounted for 3.1% of cancer cases and 5.5% of cancer-related mortalities worldwide in 2020 [2]. Porfimer sodium was approved by the FDA as a photosensitizer for photodynamic therapy of obstructing esophageal cancer [319] (Fig. 5b). In the presence of oxygen, this approach utilizes light to activate the porfimer sodium, which is relatively selectively concentrated in cancer cells, leading to cell death [320]. However, photodynamic therapy with porfimer sodium as an endoscopic therapy for esophageal cancer is losing popularity due to the potential for long-term complications [321]. Fluoropyrimidine plus platinum-based chemotherapies are frequently used as first-line therapy for advanced esophageal cancer [322]. Compared to chemotherapy alone, pembrolizumab plus 5-FU and cisplatin (chemotherapy) significantly improve clinical outcomes in the first-line treatment of advanced esophageal cancer [323].

Gastric cancer

Gastric cancer accounted for 5.6% of cancer cases and 7.7% of cancer-related mortalities worldwide in 2020 [2]. Vascular endothelial growth factor receptor 2 (VEGFR2), the principal receptor of VEGF-induced angiogenesis, is expressed in most solid tumors, including gastric cancer [324]. Ramucirumab is a VEGFR2-directed mAb [325] that binds selectively to the g-like extracellular domain III of VEGFR2, which prevents VEGF ligands from binding to VEGFR2 [326], thereby inhibiting VEGF ligand-induced cell proliferation, migration, and angiogenesis [327] (Fig. 5c). Ramucirumab monotherapy exhibits significant survival benefits in patients with advanced gastric or gastroesophageal junction adenocarcinoma who have disease progression after first-line chemotherapy compared to placebo [328]. The ramucirumab plus paclitaxel regimen also demonstrated superiority over placebo plus paclitaxel therapy in the same clinical setting; thus, ramucirumab plus paclitaxel could be regarded as a new second-line treatment for advanced gastric cancer [329].

Gastrointestinal stromal tumors

Gastrointestinal stromal tumors (GISTs) constitute the largest subset of mesenchymal tumors that arise from precursors of the connective tissue cells of the gastrointestinal tract [330, 331]. They occur predominantly (60%) in the stomach, with 30% of cases in the small intestine and 10% of cases in other sites of the gastrointestinal tract; 10~30% are malignant and exhibit intra-abdominal spread or liver metastases [332]. RTKs, such as VEGFR2, platelet-derived growth factor receptor α/β (PDGFRα/β), and KIT, are frequently overexpressed or mutated in GISTs, leading to constitutive activation of these kinases [333, 334]. Approximately 75~80% of GISTs harbor KIT mutations, and 5~8% of GISTs harbor PDGFRA mutations [334]. Therefore, the FDA-approved three multitarget TKIs for GIST treatment (Fig. 5d).

Imatinib (Additional file 1: Table S1, page 13; Table S2, page 44) is still the first-line treatment for advanced GISTs [335, 336]. However, approximately 50% of patients develop resistance within two years [336, 337]. Sunitinib is a potent inhibitor of multiple RTKs, including PDGFRα/β, VEGFR2, and KIT [338], and has been approved as second-line therapy for imatinib-resistant GISTs [339, 340]. The ATP-binding-pocket mutants KIT^V654A, KIT^T670I, and PDGFRα^D842V are the most common in imatinib-resistant GISTs, whereas certain mutant-induced resistance can be overcome by sunitinib, except PDGFRα^D842V [337, 340, 341]. Given the failures in overcoming the PDGFRA^D842V-induced resistance, avapritinib was approved as a first-line regimen for GISTs harboring PDGFRA exon 18 (including D842V) mutation [342]. Avapritinib is a potent TKI that targets KIT exon 17 (including D816V) and PDGFRA exon 18 (including D842V) mutations. In contrast, imatinib, sunitinib, and regorafenib exhibit weak potency in blocking mutation-induced constitutive kinase activity [343, 344]. Given the heterogeneity of KIT and PDGFRA mutants in GISTs, broader spectrum drugs are needed to overcome the multiple mutations of KIT and PDGFRA, as well as other RTKs. Ripretinib was designed to overcome the drug resistance of GISTs harboring broad KIT and PDGFRA mutations [345]. As a ‘switch control’ kinase inhibitor, ripretinib forces the activation loop of KIT or PDGFRα into an inactive conformation through a switch control mechanism that prevents switches from adopting a type I active state and stabilizes switches in type II inactive state [345, 346]. Therefore, the FDA-approved ripretinib for the fourth-line treatment of patients with advanced GIST who have received prior treatment with three or more TKIs [345]. Notably, the common PDGFRα^D842V mutant is sensitive to avapritinib and crenolanib but resistant to ripretinib, and secondary resistance mutations after imatinib or avapritinib treatment, such as the triple mutant PDGFRα^{D842V/V658A/G652E}, can be overcome by the heat shock protein 90 (HSP90) inhibitor tanespimycin [347].

Gastroenteropancreatic neuroendocrine tumors

Gastroenteropancreatic neuroendocrine tumors (GEP-NETs) account for more than 60% of NETs that arise from neuroendocrine cells of the digestive tract [348]. Regarding prevalence, GEP-NETs have been the second most common gastrointestinal cancer [349]. Somatostatin receptors (SSTRs) are G-protein-coupled receptors frequently expressed in GEP-NETs [350]. Somatostatin is the ligand of SSTRs that inhibits the release of pituitary and gastrointestinal hormones [351]. Octreotide (brand name: Sandostatin, approved by the FDA on Oct. 21, 1988), a synthetic octapeptide (D-Phe-c[Cys-Phe-D-Trp-Lys-Thr-Cys]-Thr-ol), is a somatostatin analog with long-acting pharmacologic properties mimicking natural somatostatin [352]. Therefore, it has been approved for metastatic carcinoid and vasoactive intestinal peptide-secreting tumors [353]. While it does not affect tumor progression, it can improve symptoms. On the other hand, Lutetium-177 (¹⁷⁷Lu) is a medium-energy β- and low-energy γ-emitting radionuclide with a maximal tissue penetration of 2 mm [354] and a half-life of 160 h [355], allowing detection by scintigraphy and subsequent dosimetry. Combining the properties of ¹⁷⁷Lu and octreotate (differs from octreotide only in that the C-terminal threoninol is replaced with threonine but exhibits a higher affinity for SSTR2 than octreotide [356]), [¹⁷⁷Lu-DOTA⁰,Tyr³]-octreotate (Lutetium Lu-177 dotatate) was approved by the FDA for peptide receptor radionuclide therapy (PRRT) of SSTR-positive advanced GEP-NETs [349] (Fig. 5e).

Cholangiocarcinoma

Hepatocellular carcinoma (HCC, comprising 75%~85% of liver cancer cases) and intrahepatic cholangiocarcinoma (ICC, comprising 10~15% of liver cancer cases) are the most frequent types of primary liver cancer, which accounted for 4.7% of cancer cases and 8.3% of cancer-related mortalities worldwide in 2020 [2]. Compared to HCC, ICC has a poorer prognosis in terms of both mOS (HCC 71.7 months vs. ICC 21.5 months) and disease-free survival (DFS) (HCC 68.2 months vs. ICC 15.5 months) [357]. Genomic alterations (including mutation, fusion, and rearrangement) that activate fibroblast growth factor receptor 2 (FGFR2) are almost exclusively found in patients with ICC, making it a promising therapeutic target [358, 359].

Pemigatinib is a potent, selective inhibitor of FGFR1-3 that binds the ATP-binding pocket of FGFR at the hinge region, thereby inhibiting FGFR-mediated cell proliferation, differentiation, and angiogenesis [360, 361]. Pemigatinib exhibits a manageable safety profile and durable antitumor activity in previously treated patients with cholangiocarcinoma harboring FGFR2 fusions/rearrangements [362]. A phase 3 FIGHT-302 clinical trial of first-line pemigatinib vs. gemcitabine plus cisplatin for advanced cholangiocarcinoma harboring FGFR2 fusions/rearrangements is still ongoing [363]. Similar to all other TKIs, acquired resistance mutations in FGFR2 (N549K/H, E565A, L617V, K641R, and K659M) are observed in patients with progressive disease and may confer resistance to pemigatinib [359]. Infigratinib is another FGFR1-3 inhibitor [364] that binds to FGFR at a hinge region, similar to pemigatinib [365]. It shows manageable toxicity and meaningful clinical activity against chemotherapy-refractory cholangiocarcinoma FGFR2 fusions/rearrangements [366, 367]. Strikingly, 5 of 6 FGFR2 mutations observed in infigratinib-resistant patients completely overlapped with the five FGFR2 mutations observed in pemigatinib-resistant cases, except for the FGFR2^V564F gatekeeper resistance mutation, which exclusively exists in infigratinib-resistant patients [359, 368, 369]. Thus, pemigatinib may theoretically overcome FGFR2^V564F mutation-induced resistance to infigratinib (Fig. 5f).

Pancreatic cancer

Pancreatic cancer has the highest mortality-to-incidence ratio (1.808) among all malignancies, accounting for 2.6% of cancer cases and 4.7% of cancer-related mortalities worldwide in 2020 [2]. Although genomic and microenvironment alterations of pancreatic cancer have been elucidated [370, 371], most alterations (e.g., KRAS and TP53 mutations) are not druggable. Given the lack of effective targets, systemic chemotherapy is still the first-line regimen. Gemcitabine is an analog of deoxycytidine that acts as a DNA synthesis inhibitor. It is phosphorylated by deoxycytidine kinase to form its active products (including gemcitabine diphosphate and gemcitabine triphosphate), which are incorporated into the DNA, leading to the inhibition of the DNA synthesis process [372] (Fig. 5g). Gemcitabine exhibits significant superiority over 5-FU in patients with advanced pancreatic cancer [373]. Currently, systemic chemotherapy combinations, including FOLFIRINOX (5-FU, leucovorin, irinotecan, and oxaliplatin) [374] and gemcitabine plus nab-paclitaxel [375], have become the first-line treatment for patients with advanced pancreatic cancer [376]. There was no significant difference in the treatment efficacy between the FOLFIRINOX and gemcitabine plus nab-paclitaxel regimens [377].

Renal cell carcinoma

Kidney cancer accounted for 2.2% of cancer cases and 1.8% of cancer-related mortalities worldwide in 2020 [2]. Renal cell carcinoma (RCC) is the most common subtype (~70%) of kidney cancer [492]. One recombinant human interleukin-2 (IL-2, aldesleukin), four TKIs (sorafenib, pazopanib, axitinib, and tivozanib), and two mammalian targets of rapamycin (mTOR) inhibitors (temsirolimus and everolimus) have been approved for the treatment of RCC in the past three decades (Fig. 7b-d).

IL-2 was first discovered as a T cell growth factor in 1976 [493] and cloned in 1983 [494]. Over the ensuing years, IL-2 was proven to be a pivotal cytokine produced primarily by CD4⁺ T cells [495, 496]. As a pleiotropic mediator within the immune system, it interacts with IL-2 receptors (IL-2Rα, IL-2Rβ, and IL-2Rγ) and induces the proliferation and differentiation of immune cells, thereby regulating a range of diseases involving infection, autoimmune disease, and cancer [496]. Aldesleukin is a nonglycosylated, modified form of human endogenous IL-2 that exerts its antitumor activity by enhancing the cytotoxicity of T lymphocytes and the activity of natural killer and lymphokine-activated killer (LAK) cells [497] (Fig. 7b). Aldesleukin monotherapy achieves an ORR of 14~25% and exhibits durable antitumor activity in patients with metastatic RCC [498, 499].

Sorafenib is an oral first-generation multitarget TKI that targets several RTKs, including RAF1, BRAF^WT, BRAF^V600E, VEGFRs, PDGFR-β, FGFR1, FMS-like tyrosine kinase 3 (FLT3), KIT, and RET [500, 501]. It occupies the ATP adenine binding pocket of these RTKs with its distal 4-pyridyl ring, which blocks the autophosphorylation of these RTKs, thereby attenuating the mitogen-activated protein kinase (MAPK)/extracellular signal-regulated kinase (ERK) signaling pathway [500, 502]. Given its potent antitumor effects, sorafenib is applied to various malignancies in addition to RCC [501]. Sorafenib resistance inevitably occurs and mainly involves mutations in RTKs and activation of the bypass pathway [503]. Regorafenib, as a fluoro‐sorafenib, provides a nearly 3-month improvement in OS in HCC patients progressing on sorafenib treatment [504]. However, regorafenib cannot overcome sorafenib resistance because they share a similar structure [505]. Sunitinib is another first-generation multitarget TKI with a similar target profile to sorafenib. Intriguingly, sequential sorafenib-sunitinib and vice versa provide similar clinical benefits in metastatic RCC [506]. In contrast, pazopanib is an oral second-generation multitarget TKI that preferentially targets VEGFRs, PDGFRα/β, and KIT [507–509]. It competes with ATP for binding to the cytoplasmic domain of these RTKs and prevents ATP-induced activation [510]. Pazopanib retains clinical activity in patients with advanced clear-cell RCC after failure of sunitinib or bevacizumab [511]. Thus, pazopanib is non-inferior to sunitinib as first-line therapy in clinical efficacy and exhibits advantages in the safety profile [512]. Axitinib and tivozanib are selective second-generation VEGFR inhibitors that exhibit greater selectivity for VEGFRs than other TKIs (e.g., sorafenib, sunitinib, pazopanib) [513, 514]. Axitinib is a substituted indazole derivative produced from a structure-based drug design [515]. It exhibits antitumor activity and a manageable safety profile in sorafenib-refractory metastatic RCC [516] but has no significant superiority over sorafenib as first-line therapy [517]. Compared with sorafenib, axitinib significantly prolongs the median PFS by two months and can be an option for second-line therapy [518, 519]. Nevertheless, axitinib plus avelumab or pembrolizumab therapies exhibit significant clinical benefits in advanced RCC as first-line treatment compared to the standard of care of sunitinib [520, 521]. Tivozanib is a quinoline urea derivative that interacts with the ATP-binding site and the allosteric-binding site consisting of the DFG motif within the activation loop of VEGFR, similar to sorafenib [522]. It inhibits VEGF-induced VEGFR phosphorylation and blocks VEGF-dependent but not VEGF-independent MAPK activation [523]. Tivozanib improves PFS and is better tolerated as third- or fourth-line therapy than sorafenib [524] (Fig. 7c).

mTOR is a serine/threonine-protein kinase that governs a diverse set of biological events by joining with other components to form two distinct complexes known as mTOR complex 1 (mTORC1) and mTOR complex 2 (mTORC2) [525]. mTORC1 is composed of three core components: mTOR, mammalian lethal with SEC13 protein 8 (mLST8, also known as GβL) [526], and its unique defining subunit, regulatory-associated protein of mTOR (RAPTOR) [527]. In lieu of RAPTOR, mTORC2 contains rapamycin-insensitive companion of mTOR (RICTOR) [528, 529]. mTORC1 governs glucose metabolism, cell cycle progression, cell survival, and the biosynthesis of proteins, lipids, and nucleotides, while mTORC2 governs cytoskeletal rearrangement and prosurvival pathways [525]. Temsirolimus and everolimus are derivatives of sirolimus (also known as rapamycin) (Additional file 1: Table S1, page 11), a compound extracted from a Streptomyces hygroscopicus soil bacterium [530]. Mechanistically, rapamycin, temsirolimus, and everolimus are inhibitors of mTORC1. Similar to rapamycin, both temsirolimus and everolimus bind to FK506-binding protein 12 (FKBP12) and form a gain-of-function complex, which subsequently prohibits the activation of mTOR, resulting in cell cycle arrest and suppression of hypoxia-inducible factor-1α (HIF1α) and VEGF expression [531, 532].

Temsirolimus is a water-soluble ester of rapamycin with improved pharmaceutical properties, including stability and solubility, making it suitable for intravenous administration [533]. Intriguingly, temsirolimus is hydrolyzed by CYP3A4 to its major metabolite rapamycin in vivo [534]. Compared with interferon-α (IFNα) monotherapy, temsirolimus improves OS among patients with metastatic RCC [535]. Everolimus is a hydroxyethyl ether derivative of rapamycin with superior pharmaceutical characteristics, making it suitable for oral administration. Unlike temsirolimus, everolimus is not converted to rapamycin in vivo [532]. Compared with placebo, everolimus prolongs PFS in patients with metastatic RCC previously treated with sunitinib and/or sorafenib [536]. Compared to temsirolimus, everolimus exhibits superior clinical efficacy in metastatic RCC in terms of prolonging OS and PFS and decreasing the risk of death [537, 538] (Fig. 7d).

RCC is a highly vascularized tumor prone to distant metastasis [503]. Mechanistically, clear-cell RCC accounts for approximately 80~85% of metastatic RCC cases [539], whereas 60% of clear-cell RCC harbors loss-of-function of the von Hippel–Lindau (VHL) tumor suppressor gene, which leads to the accumulation of HIF1α and activation of its target genes, including VEGF and PDGF [540]. It explains why all these TKIs target VEGFRs and/or PDGFRs, although they have different target profiles. In addition, two mTOR inhibitors can reduce the expression of HIF1α and VEGF. Before 2005, nonspecific immune cytokines, such as IL-2 and IFNα, were previously the mainstays of therapy for advanced RCC [540]. Currently, PD1-directed mAbs, such as pembrolizumab and nivolumab, plus TKIs, such as axitinib [521], lenvatinib [541], and cabozantinib [542], have become the first-line regimens for patients with advanced or metastatic RCC.

Basal cell carcinoma

Basal cell carcinoma constitutes approximately 80% of all NMSCs, and more than five million new cases are diagnosed each year worldwide [633, 634]. However, the absolute incidence and mortality are difficult to determine since basal cell carcinoma is usually excluded from cancer registry statistics [633]. In part, basal cell carcinoma is the most frequent human cancer subtype [632]. Loss-of-function mutations of tumor suppressor gene patched homolog 1 (PTCH1) occur in 30~40% of basal cell carcinomas [635]. Dysfunctional PTCH1 causes constitutive activation of smoothened (SMO), resulting in continuous activation of hedgehog signaling and its target genes in basal cell carcinoma [174, 634], making SMO a promising target. Therefore, two SMO inhibitors (vismodegib and sonidegib) were approved for the treatment of basal cell carcinoma [636, 637] (Fig. 8f).

Vismodegib is the first SMO inhibitor that occupies the transmembrane domain core and forms hydrophobic interactions with SMO by a network of hydrogen bonds [638, 639], thereby inhibiting SMO activity and downstream signaling, regardless of PTCH1 [640]. However, approximately 21% of patients develop resistance within a year while undergoing continuous vismodegib treatment [641]. Various mutations in SMO are located in the drug-binding pocket of SMO and confer resistance to vismodegib by abrogating or impairing vismodegib binding to SMO, such as D473H, D473G, W281C, V321M, I408V, C469Y, and Q477E [642–646]. Sonidegib is another SMO inhibitor that binds to a drug-binding pocket of SMO similar to vismodegib. It exerts antitumor effects by inhibiting the transcriptional activity of glioma-associated oncogene (GLI) and inducing the expression of caspase-3 and the cleavage of PARP, resulting in cell cycle arrest and apoptosis [647]. However, sonidegib has an SMO binding pattern similar to that of vismodegib. As expected, sonidegib cannot overcome the vismodegib resistance induced by SMO mutations [648].

Merkel cell carcinoma

Merkel cell carcinoma is a rare but highly aggressive NMSC with neuroendocrine features [649] frequently associated with Merkel cell polyomavirus infection and accumulation of ultraviolet-induced mutations [650]. Approximately 50% of tumor cells and 55% of tumor-infiltrating immune cells (TIICs) express PDL1 in Merkel cell carcinoma [651]. Avelumab is a PDL1-directed mAb that binds to the CC’ loop of PDL1 [547, 652], thereby blocking the interaction between PDL1 and PD1, which reactivates the antitumor immunity of T lymphocytes, similar to atezolizumab and durvalumab (Fig. 8g). Avelumab is well tolerated with durable responses [653] and has become the standard-of-care treatment for metastatic and advanced Merkel cell carcinoma [654], similar to other anti-PD1 immunotherapies, including pembrolizumab [655, 656] and nivolumab [657].

Kaposi’s sarcoma

Kaposi’s sarcoma accounted for 0.2% of cancer cases and 0.2% of cancer-related mortalities worldwide in 2020 [2]. It is a relatively rare cancer caused by Kaposi’s sarcoma-associated herpesvirus (KSHV, also known as human herpesvirus 8 (HHV8)) infection [697]. The skin and superficial mucosae are the most common sites of Kaposi’s sarcoma lesions [698]. Retinoid X receptor α (RXRα) and retinoic acid receptor γ (RARγ) control cell differentiation, proliferation, and apoptosis and are predominantly expressed in the skin [699, 700]. Alitretinoin is a 9-cis-retinoic acid that acts as a pan-agonist of RARs and RXRs [700]. It modulates cell differentiation and apoptosis in a variety of sarcomas by potentially inducing the formation of a homodimer of RXRs [701–703] (Fig. 9d). Alitretinoin gel demonstrates durable responses with tolerable safety in patients with acquired immunodeficiency syndrome (AIDS)-related Kaposi’s sarcoma [704, 705].

Liposarcoma

Liposarcoma is a rare malignant tumor of adipocytic differentiation that accounts for 15~20% of STS cases [706]. It is characterized by recurrent amplifications within chromosome 12, which leads to the overexpression of disease-driving genes [706]. Leiomyosarcoma is a malignant mesenchymal tumor that accounts for 10~20% of STS cases [707]. Leiomyosarcoma also exhibits complex genomic alterations involving DNA copy number changes and gene mutations [708]. Trabectedin is a tetrahydroisoquinoline alkaloid derived from the Caribbean marine tunicate Ecteinascidia turbinata [709] (Fig. 9e). It binds to the minor groove of DNA that bends DNA toward the major groove by forming trabectedin-DNA adducts, which block the G2/M phase transition of the cell cycle and inhibit cell proliferation [709–711]. The FDA-approved Trabectedin for patients with unresectable or metastatic liposarcoma or leiomyosarcoma who received a prior anthracycline-containing regimen [712].

Soft tissue sarcoma

PDGFRα expression in STSs is sevenfold higher than in normal tissues [713]. Olaratumab is a PDGFRα-directed mAb that selectively binds to the extracellular domain of PDGFRα, which prevents PDGF-AA, PDGF-BB, and PDGF-CC ligands from binding to PDGFRα [714, 715] (Fig. 9f). It inhibits the ligand-induced autophosphorylation of PDGFRα and downstream signaling, thereby blocking PDGFRα-mediated cell mitogenesis [716]. Olaratumab plus doxorubicin combination therapy significantly improves the PFS and OS compared to doxorubicin alone in patients with advanced STSs [717].

Tenosynovial giant cell tumor

Tenosynovial giant cell tumor (TGCT) is a rare, locally aggressive neoplasm mainly characterized by colony-stimulating factor-1 (CSF1) translocations and CSF1 receptor (CSF1R) overexpression [718, 719]. CSF1 translocations result in local overexpression of CSF1, which attracts histiocytoid and CSF1R-expressing inflammatory cells [718–720]. Moreover, CSF1 promotes the differentiation of monocytes into tumor-associated macrophages (TAMs), which in turn facilitates tumor survival, growth, and metastases with their immunosuppressive effects [721–723]. Thus, the CSF1/CSF1R axis is critical for the tumorigenesis and progression of TGCTs. Pexidartinib is an inhibitor of CSF1R, KIT, and FLT3 that accesses the autoinhibited state of CSF1R through direct interactions with juxtamembrane residues embedded in the ATP-binding pocket, thereby blocking the CSF1/CSF1R axis [724] (Fig. 9g). Notably, pexidartinib retains activity against the quizartinib-resistant FLT3 gatekeeper F691L mutation [725]. As the first systemic therapy of TGCT, pexidartinib exhibits a robust tumor response with improved clinical outcomes [718, 723].

Epithelioid sarcoma

Epithelioid sarcoma is an ultrarare high-grade soft tissue sarcoma with clinicopathological complexities that predisposes patients to locoregional recurrence [726], accounting for 1.2~1.5% of STS cases [727]. INI1 (encoded by the SMARCB1 tumor suppressor gene) is a core subunit of the switch/sucrose nonfermentable (SWI/SNF) chromatin remodeling complex frequently inactivated in epithelioid sarcomas [728]. The SWI/SNF complex is a crucial regulator of nucleosome positioning, frequently located at sites marked by acetylated histone H3 lysine 27 (H3K27ac), which establishes an open chromatin state with other transcription factors for transcriptional activation [729]. Enhancer of zeste homolog 2 (EZH2) is an enzymatic subunit of polycomb repressor complex 2 (PRC2) that negatively regulates the activity of the SWI/SNF complex by placing the repressive trimethylated histone H3 lysine 27 (H3K27me3) mark [729]. EZH2 is expressed in approximately one-third of epithelioid sarcomas [730], making EZH2 a promising target. Tazemetostat is an inhibitor of EZH2 that blocks the lysine methyltransferase activity of EZH2 by selectively binding to the S-adenosyl methionine (SAM) binding site of EZH2 [731] (Fig. 9h). Tazemetostat exhibits clinical activity with favorable safety in patients with advanced epithelioid sarcoma harboring INI1 loss [732, 733].

Brain and other CNS tumors accounted for 1.6% of cancer cases and 2.5% of cancer-related mortalities worldwide in 2020 [2]. These tumors comprise over 100 histologically distinct subtypes with varying clinical characteristics, treatments, and outcomes, mainly including tumors of neuroepithelial tissue, cranial and spinal nerves, meninges, etc. [734].

Glioblastoma

Glioblastoma multiforme (GBM) is the most frequent and lethal subtype of brain cancer that originates in the CNS [735]. Compared with surrounding healthy tissue, brain cancers possess a more alkaline pH [736]. Temozolomide is a DNA alkylating prodrug stable at acidic pH values but labile at alkaline pH values [737] (Fig. 9i). The alkaline microenvironment within brain cancer preferentially facilitates the activation of temozolomide [736, 737]. It adds a methyl group to the O⁶ position of guanine (G), resulting in a methyl-guanine (meG)-to-thymine (T) mismatch during DNA replication instead of a G-to-cytosine (C) match [738], which leads to DNA damage and ultimately cell apoptosis [739, 740]. Temozolomide exhibits an acceptable safety profile and improves PFS compared with procarbazine in patients with GBM at first relapse [741]. Currently, temozolomide-containing regimens are still the first-line therapy for GBM [742, 743]. However, O⁶-meG methyltransferase (MGMT) removes alkyl groups from the O⁶ position of G, conferring resistance to temozolomide [744, 745].

Neuroblastoma

Neuroblastoma is a malignant embryonal tumor derived from primitive cells of the sympathetic nervous system [746, 747]. It is the most frequent and lethal solid tumor in children and is commonly associated with a poor overall prognosis [747]. Disialoganglioside GD2 is expressed almost uniformly on the surface of neuroblastoma cells and induces cell proliferation, invasion, and motility by activating RTK-mediated signal transduction [746, 748, 749], making it an effective and tractable target of neuroblastoma [750, 751]. Dinutuximab is a human/mouse chimeric GD2-directed mAb that recognizes and binds to the sugar moiety of GD2 exposed to the extracellular milieu (similar to 14G2a antibody [752]), thereby inducing cell lysis through ADCC and CDC [751–753]. Compared with standard therapy (six cycles of 13-cis-retinoic acid), dinutuximab significantly improved clinical outcomes in combination with alternating granulocyte-macrophage colony-stimulating factor (GM-CSF) and IL-2 after standard therapy in patients with high-risk neuroblastoma [754]. However, anti-drug antibodies, including human anti-mouse or -chimeric antibodies, may cause treatment delays, terminations, or even abrogate the antitumor effects [746]. Naxitamab is designed to reduce the effects of anti-drug antibodies but enhance ADCC through humanized IgG1-Fc and retain complement-mediated cytotoxicity potency through its high affinity for GD2 [746, 755]. As expected, naxitamab exhibits modest toxic effects, low immunogenicity, and substantial anti-neuroblastoma activity in combination with GM-CSF in patients with relapsed or refractory high-risk neuroblastoma [756–758] (Fig. 9j).

Malignant pleural mesothelioma

Malignant pleural mesothelioma (MPM) is a rare but highly aggressive and lethal cancer that originates in the serosal outer linings of the lungs (pleurae), heart, abdomen, and testes, with a 5-year OS rate of~5% [759, 760]. Folate receptors (FRα, FRβ, and FRγ) are cysteine-rich cell surface glycoproteins that mediate the cellular uptake of folate, commonly expressed at low levels in most normal tissues [761]. Folate-dependent one-carbon metabolism is required for the de novo synthesis of purines, thymidylate, and S-adenosyl methionine and is thus critical to DNA synthesis [762]. Nevertheless, FRα is highly activated and overexpressed in MPM tissues compared with normal adjacent tissues, making the folate–FRα delivery and metabolism system an attractive target for MPM treatment [763]. Pemetrexed is a multitargeted anti-folate agent that inhibits at least three enzymes (thymidylate synthase, dihydrofolate reductase, and glycinamide ribonucleotide formyltransferase) involved in folate metabolism and DNA synthesis [764] (Fig. 9k). Compared to cisplatin monotherapy, pemetrexed plus cisplatin therapy improved the mOS (12.1 months vs. 9.3 months) and was thus approved by the FDA for unresectable MPM treatment [765]. Most recently, compared with cisplatin-pemetrexed chemotherapy, durvalumab plus platinum-pemetrexed chemotherapy therapy significantly improved the mOS (20.4 months vs. 12.1 months) in patients with unresectable MPM [766].