2.1 Acquired resistance

Acquired resistance to TKIs manifests as resistance that emerges after an initial positive response to TKI treatment. Compared with primary resistance, acquired resistance is more frequently observed in TKI-targeted therapy. Acquired resistance can arise from various factors including secondary mutations in the targeted tyrosine kinase, overexpression or amplification of the target proteins/genes, loss of the original targeted mutations, activation of alternative signaling pathways, and tumor histological transformation (Figure 1).

2.1.1 Mutations in the target kinase domain (on-target)

Secondary mutations in the target kinase domain, which are also called on-target mutations, are the most commonly encountered causes of TKI-acquired resistance. According to the location and function of mutated genes in the target kinase domain, secondary on-target mutations can be categorized into five types (Figure 2): (1) gatekeeper mutations, (2) solvent-front mutations, (3) covalent binding site mutations, (4) compound mutations, and (5) other site mutations.

Gatekeeper mutation

Among all types of on-target mutations, gatekeeper mutation is the most frequently identified and extensively investigated TKI-resistance mechanism. Gatekeeper residues situate in the hinge region of tyrosine kinases’ ATP-binding pocket.⁴⁹ They play a central role in controlling the accessibility of TKIs to the ATP-binding pocket.⁴⁹ The mutation of gatekeeper residues can influence the interaction between the inhibitors and their targeting kinases, thereby reduce the efficacy of TKIs and lead to drug resistance.^{50, 51} A gatekeeper point mutation, involving the substitution of threonine with isoleucine at position 315 (T315I) in the BCR–ABL protein, was identified in CML patients resistant to imatinib in 2001.⁵² Molecularly, the hydroxyl group of threonine 315 can form a hydrogen bond with imatinib upon binding, while the side chain located at position 315 also regulates the steric interaction, governing the inhibitor's binding to hydrophobic regions neighboring the ATP-binding pocket.^{53, 54} The substitution of threonine with isoleucine leads to the elimination of the hydrogen bond between imatinib and targeted kinase, destabilizing the placement of imatinib.⁵⁴ Therefore, a point mutation of gatekeeper is sufficient to confer imatinib resistance in patients with CML.

Similarly, a gatekeeper mutation in the kinase domain of EGFR (T790M) was found in 50−60% of NSCLC patients resistant to the first- and second-generation EGFR-TKIs.^55-57 The substitution of threonine with methionine in the catalytic cleft of EGFR kinase domain, EGFR T790M, was first identified in a NSCLC patient with disease progression after gefitinib treatment in 2005.⁵⁷ The methionine substitution introduced a bulkier amino acid side chain to the ATP-kinase-binding pocket, which resulted in a steric hindrance interfering with the binding of gefitinib or erlotinib.⁵⁷ In addition, the presence of T790M mutation was also proven to enhance the affinity of ATP to the EGFR kinase domain.⁵⁸ As EGFR-TKIs, such as gefitinib, compete with ATP for binding to the kinase active site, the increased ATP affinity reduces the potency of the inhibitor.⁵⁸ Later on, other gatekeeper mutations (Table 2), such as ALK-L1196M, ROS1-L2026M, FGFR1-V561F/M, FGFR2-V564F/I, FGFR3-V555E, FGFR4-V550L/E/M, RET-V804L/M, FLT3-F691I/L, KIT -T670I, and PDGFRα-T674I, were identified in patients with IGST, NSCLC, or other cancers after corresponding TKI treatment.⁵⁹

TABLE 2. Representative and commonly reported gatekeeper mutations leading to TKI resistance.

PTKs	Mutations	Cancer types	Prior TKIs	Prevalence (%)	References
BCR–ABL	T315I	CML	Imatinib	7%	52, 60, 61
		ALL	Dasatinib	70%	62
EGFR	T790M	NSCLC	First- or second-generation EGFR-TKIs	50–60%	23, 57
ALK	L1196M/Q	NSCLC	Crizotinib	7%	63
			Alectinib	6%
		ALCL	Alectinib	Cases reported *	64
ROS1	L2026M	NSCLC	Crizotinib	8%	65
RET	V804L/M	NSCLC	Vandetanib	Cases reported *	66-68
		MTC	Multikinase inhibitors	Cases reported *
KIT	T670I	GIST	Imatinib	Cases reported *	69
PDGFRα	T674I	GIST	Imatinib	Cases reported *	70, 71
BTK	T474I	CLL	Pirtobrutinib	22%	72
FGFR1	V561M	SCLL	AZD4547, E3810 (lucitanib)	Cases reported *	73, 74
FGFR2	V564F	CCA	BGJ398	Cases reported *	75
FGFR3	V355M/L	Myeloma	AZ12908010	Cases reported *	74, 76
FGFR4	V550L/M	HCC, RMS	BLU-554	Cases reported *	77, 78
NTRK1	F589L	CCA, CRC	Larotrectinib	Cases reported *	79, 80
FLT3	F691I/L	AML	Quizartinib, gilteritinib	Cases reported *	81

• This table includes selected gatekeeper mutations and does not aim to represent the entirety of all reported gatekeeper mutations.
• “Cases reported *” indicates cases with corresponding mutations were reported in case reports, with no available prevalence rate.
• Abbreviations: ALCL, anaplastic large cell lymphoma; ALK, anaplastic lymphoma receptor tyrosine kinase; ALL, acute lymphoblastic leukemia; AML, acute myeloid leukemia; BCR–ABL, breakpoint cluster region-Abelson; BTK, Bruton tyrosine kinase; CCA, cholangiocarcinoma; CLL, chronic lymphocytic leukemia; CML, chronic myeloid leukemia; CRC, colorectal carcinoma; EGFR, epidermal growth factor receptor; FGFR, fibroblast growth factor receptor; FLT3, FMS-like tyrosine kinase 3; GIST, gastrointestinal stromal tumors; HCC, hepatocellular carcinoma; KIT, KIT proto-oncogene; MTC, medullary thyroid carcinoma; NSCLC, non-small cell lung cancer; NTRK, neurotrophic receptor tyrosine kinase; PDGFR, platelet-derived growth factor receptor; PTKs, protein tyrosine kinases; RET, RET proto-oncogene; RMS, rhabdomyosarcoma; ROS1, ROS proto-oncogene 1; SCLL, stem cell leukemia/lymphoma syndrome; TKI, tyrosine kinase inhibitor.

Solvent-front mutation

Another predominant on-target mutation mechanism leading to TKI-resistance is solvent-front mutation, which confers acquired resistance to ALK, ROS1, RET, NTK1, NTK2, and NTK3 fusion-targeted therapy.^{63, 82-86} Solvent-front mutations occur in amino acid residues that are located near the solvent-accessible surface of the ATP-binding pocket.⁸⁷ For example, solvent-front mutations ALK-G1202R and ALK-S1206Y which sit near the crizotinib-binding site are reported to cause acquired resistance to crizotinib in NSCLC patients.^{63, 87} Computational modeling suggests that the G1202R mutation results in a bulkier basic residue that would induce steric hindrance, potentially impeding the binding of crizotinib to targeted kinases. And the S1206Y mutation may disrupt the stability of the interaction between the side-chain hydroxyl group of Ser¹²⁰⁶ and the carboxylate group of D¹²⁰³.⁸⁷ Generally, solvent-front mutations lead to conformational changes in the target kinase domain, diminishing the affinity of crizotinib for the mutant ALK, and thus abrogating the efficacy of crizotinib to inhibit aberrantly activated ALK.⁸⁷ ROS1-G2032R, involving the substitution of glycine with arginine in the solvent front of ROS1 kinase domain, was found to mediate resistance to crizotinib and lorlatinib in ROS1 fusion-positive NSCLC patients.^{83, 88} Analogous solvent-front mutations have also been identified in other tyrosine kinases and confer to TKI-resistance, such as RET-G810 solvent mutations after selpercatinib treatment in NSCLC patients,^{84, 85} EGFR-G706S/R mutation following osimertinib treatment in NSCLC patients,⁸⁹ TRKA-G595R and TRKA-G667C mutations after entrectinib treatment in colorectal cancer patients.⁸⁶ The most frequently reported solvent-front mutations are summarized in Table 3.

TABLE 3. Representative and commonly reported solvent-front mutations leading to TKI resistance.

PTKs	Mutations	Cancer types	Prior TKIs	Prevalence (%)	References
EGFR	G796S/R L792X	NSCLC	Osimertinib	Cases reported *	89, 90
ALK	G1202R	NSCLC	Crizotinib	2%	63
			Alectinib	21–29%
			Ceritinib
			Brigatinib
	I1171T	NSCLC	Crizotinib	2%	63
			Alectinib	12%
RET	G810R/S/C	NSCLC	Selpercatinib	Cases reported *	85
ROS1	G2032R	NSCLC	Crizotinib	38%	88
			Lorlatinib	32%
	D2033N	NSCLC	Crizotinib	2.4%	88
	L2026M	NSCLC	Crizotinib	Cases reported *	88
TRKA	G595R	CRC	Entrectinib	Cases reported *	79
			Larotrectinib	Cases reported *
TRKC	G623R	MASC	Entrectinib	Cases reported *	79
			Larotrectinib	Cases reported *
BTK	C481S	CLL	Ibrutinib	50−75%	91, 92

• This table includes selected solvent-front mutations and does not aim to represent the entirety of all reported solvent-front mutations.
• “Cases reported *” indicates cases with corresponding mutations were reported in case reports, with no available prevalence rate.
• Abbreviations: ALK, anaplastic lymphoma receptor tyrosine kinase; BTK, Bruton tyrosine kinase; CLL, chronic lymphocytic leukemia; CRC, colorectal carcinoma; EGFR, epidermal growth factor receptor; NSCLC, non-small cell lung cancer; PTKs, protein tyrosine kinases; RET, RET proto-oncogene; ROS1, ROS proto-oncogene 1; TKIs, tyrosine kinase inhibitors.

Covalent binding site mutation

A less common yet formidable resistance mechanism in targeted therapy is the covalent binding site mutation. As discussed above, the EGFR-T790M mutation can arise in 50−60% of NSCLC patients after receiving treatment with first- and second-generation EGFR-TKIs.⁵⁵ Osimertinib, a third-generation EGFR-TKI with greater potency and selectivity, was designed specifically to overcome the EGFR-T790M resistance mutation through potent covalent-binding to the ATP-binding pocket.⁹³ In NSCLC patients with T790M acquired resistance mutation treated with first-generation EGFR-TKIs, osimertinib showed a superior overall response rate (ORR) (71 vs. 31%) and longer PFS (10.1 vs. 4.4 months) compared with conventional chemotherapy.⁹⁴ Following this, osimertinib was compared with standard first- and second-generation EGFR-TKIs in patients with advanced-stage EGFR-positive NSCLC in the pivotal FLAURA trial, revealing a prolonged median PFS (18.9 vs. 10.2 months).⁹⁵

While osimertinib effectively counteracts the EGFR-T790M resistance mutation, its covalent binding to cysteine 797 (C797) residue in the ATP-binding pocket makes C797 residue vulnerable to mutations.²³ Unsurprisingly, EGFR-C797S is reported as the most frequent resistance mutation to osimertinib, which has been identified in 10−33% patients after second-line osimertinib treatment, and 6% patients after first-line osimertinib treatment.^96-99 EGFR-C797S mutation occurs at the EGFR C797 codon in exon 20, which is located within the ATP-binding cleft.^{100, 101} The mutation leads to a substitution of cysteine with serine, resulting in the loss of the covalent bond between osimertinib and EGFR.^100-102 Nevertheless, it is noteworthy that the allelic context where C797S arises holds potential implications for treatment. In some circumstances, patients harboring the C797S mutation may remain responsive to quinazoline-based EGFR-TKIs.^103-105 When C797S emerges in trans with the T790M mutation, patients are responsive to both first-generation and third-generation EGFR-TKIs to address C797S- and T790M-positive alleles, respectively.^103-105 Conversely, when the mutations occur in cis, patients demonstrate resistance to all available EGFR-TKIs, either alone or in combination.^103-105 Similar to EGFR-C797S, HER2-C805S has also been reported to cause resistance to HER2-TKI therapy in HER2-mutated preclinical cell models.¹⁰⁶ However, the clinical existence and significance of HER2-C805S covalent mutation is yet to be determined.

Compound mutations

Compound mutations refer to the presence of two or more mutations within the same gene or across different genes in the same molecule.^{107, 108} Compound mutations have the potential to confer cross-resistance to multiple TKIs through steric hindrance mechanisms.^{107, 109} Compound mutations were first reported in CML patients following sequential therapy with different ABL kinase inhibitors.¹⁰⁸ In a study, 20 out of 48 CML patients receiving various ABL-TKI treatments were found to harbor compound mutations.¹⁰⁷ However, the genuineness of BCR–ABL compound mutations has been long debated due to their rarity of “low-level mutations,” which can lead to false-positive results in polymerase chain reaction amplification.^{110, 111} By applying next-generation sequencing (NGS), Deininger et al.¹¹¹ reported that no single or compound mutation was consistently identified to confer primary or acquired resistance to BAR–ABL inhibitor ponatinib in chronic-phase CML patients. In EGFR-mutant NSCLC, EGFR compound mutations are defined as double or multiple mutations occurring at the EGFR tyrosine kinase domain.^{112, 113} EGFR compound mutations are usually a combination of an EGFR typical mutation, such as ex19del, L858R, G719X, and atypical mutations.¹¹³ It is reported that patients with EGFR compound mutations have poorer clinical outcomes than patients with EGFR-TKI sensitive mutations (ex19del, L858R), but have a more favorable prognosis than patients with resistant mutations (T790M, ex20ins).^{112, 114} Similarly, compound ALK resistance mutations were identified in ALK-positive patients treated with ALK-TKIs.⁶³ Some studies show that most on-target ALK mutations conferring resistance to third-generation ALK-TKI lorlatinib are ALK compound mutations.^{23, 115-120}

Other site mutations

In addition to the classic secondary mutations discussed above, mutations in other sites of target kinase domains have also been reported to confer TKI resistance. Typically, other site mutations occur in the kinase residues with important functions. They usually are less common. Tyrosine kinase domains exhibit common structural elements, such as the ATP-binding pocket located between an N-terminal lobe containing the αC helix, and a C-terminal lobe containing an activation loop. Mutations in these structural components can cause resistance to inhibitors. For example, EGFR-T854A mutation in the ATP-binding site, EGFR-D761Y, EGFR-L747S mutations near the αC helix, and EGFR-L792F/H (1.2%) in the hinge region have been found to confer resistance to EGFR-TKIs.^121-124 D276G mutation, located on the β3-αC loop of BCR–ABL kinase domain, was reported to cause resistance to imatinib by inducing destabilization of the inactive conformation of the kinase.¹²⁵ Similarly, ATP-binding site mutation (ALK-G1269A) and near αC helix site mutations (ALK-C1156Y, ALK-L1152R, ALK-F1174C, and ALK-1151T insertion) were identified in ALK-rearranged cancers, conferring resistance by causing conformational changes.^{87, 126-128}

2.2 Primary or intrinsic resistance

Primary resistance, defined as the lack of initial responses to TKIs, has been observed in 20−30% of advanced EGFR-mutant NSCLC patients,⁹⁵ 20−40% advanced ALK-positive NSCLC patients,¹⁹⁴ 10% GIST patients,¹⁹⁵ and other cancer patients.¹⁹⁶ Primary resistance to TKIs is often associated with coexistence of additional genetic alterations together with TKI-sensitive mutations within the context of tumor heterogeneity.¹⁹⁷ Specifically, molecular mechanisms contributing to primary or intrinsic resistance include the presence of TKI-resistant mutations (e.g., existence of de novo EGFR-T790M and EGFR ex20ins are associated with primary resistance to early-generation EGFR-TKIs^{29, 198}), concurrent alterations in other oncogenes (e.g., de novo MET amplifications or KRAS mutation in EGFR-TKI resistant patients and KRAS mutation in imatinib resistance GIST patients^{195, 199-201}), mutations in downstream signaling pathways (e.g., PTEN-loss confers resistance to EGFR-TKIs²⁰²), and germline deletions of gene encoding BCL-2-like protein 11 (BIM).²⁰³

3.1 Development of next-generation TKIs

Acquired drug resistance mutations, including gatekeeper, solvent-front mutations, and aspartic–phenylalanine–glycine (DFG)-loop mutations, most commonly affect the binding of the drug to its target. Developing new TKIs or finding new sites of action is a promising solution for “on-target” resistance. For instance, resistance to first-generation EGFR-TKIs is commonly associated with the EGFR T790M secondary mutation.^{55, 204} Osimertinib (Tagrisso) is a third-generation TKI designed to target the EGFR T790M to overcome the resistance of first-generation EGFR-TKIs.²⁰⁵ The common strategies to overcome resistance mainly include drug structure optimization, designing covalent inhibitors, and designing allosteric inhibitors.

3.2 Combination therapies

The development of next-generation inhibitors to overcome drug resistance has become entrenched in current kinase drug discovery. However, as exemplified by osimertinib and lorlatinib, continual development of further next-generation agents that directly target the drug-resistant kinase mutant still acquired resistance.^{96, 213, 215, 221} Besides, resistance caused by “on-target” mutations only accounts for a minority of cases. The majority of patients develop resistance due to “off-target” pathway activation, while some resistance mechanisms remain unknown. Only a small proportion of patients can continue with monotherapy, while more patients require combination therapy to simultaneously target kinase protein and off-target mutations. Combination therapies have the potential to suppress the emergence of drug resistance by completely and simultaneously suppressing cooperating oncogenic signals or concurrently blocking bypass signaling pathways that may mediate drug resistance. In the following section, we highlight recent advances in rational combination strategies that are based on different principles. Some of these drug combinations are already in clinical practice, whereas others are in earlier stages of clinical development.

3.3 Novel dual-targeted bispecific antibodies or antibody–drug conjugates

Antibody–drug conjugates (ADCs) and dual-targeted bispecific antibodies represent innovative therapeutic approaches that are currently in development and hold promise for patients with TKI-resistant cancers. These novel agents are especially intriguing due to their ability to potentially address a wide range of resistance mechanisms. Amivantamab is a bispecific antibody targeting EGFR and MET that received accelerated approval for patients with NSCLCs harboring EGFR exon 20 insertion mutations in May 2021 based on the results of the CHRYSALIS trial.³³⁹ The combination of amivantamab with a third-generation EGFR-TKI lazertinib in patients with osimertinib-relapsed, chemotherapy-naïve, EGFR-mutated advanced NSCLC has shown preliminary efficacy, with an ORR of 36% and a median PFS of 4.9 months.²⁶⁴ Inspiringly, amivantamab plus carboplatin–pemetrexed chemotherapy with and without lazertinib also demonstrated antitumor activity in patients with osimertinib refractory EGFR-mutated advanced NSCLC. Objective response rate was significantly higher for amivantamab–chemotherapy and amivantamab–lazertinib–chemotherapy versus chemotherapy (64 and 63 vs. 36%, respectively).²⁶³ Of note, an exploratory analysis suggested that EGFR/MET immunohistochemistry (IHC) staining may predict for response to amivantamab plus lazertinib. A retrained signature of MET 3+ stating on 25% tumor cells was identified as predictive of response regardless of molecular resistance mechanism.^{340, 341}

As another example, patritumab deruxtecan is a specifically engineered potential first-in-class HER3-directed ADC. In HERTHENA-Lung01, patritumab deruxtecan was studied in 225 patients with EGFR-mutated locally advanced or metastatic NSCLC following disease progression with an EGFR-TKI and platinum-based chemotherapy, which demonstrated an ORR of 29.8%. The median duration of response was 6.4 months.³⁴² An analysis of 48 paired pretreatment and posttreatment tumor samples showed augmented HER3 expression in EGFR-mutant tumors with acquired resistance to EGFR-TKIs³⁴³ and that EGFR inhibition increased membrane expression of HER3 and led to enhanced efficacy of HER3-DXd,³⁴⁴ thus supporting the rational combination of osimertinib and HER3-DXd in EGFR-mutant NSCLC. The activity of patritumab deruxtecan in patients with previously treated EGFR-mutant NSCLCs is currently being investigated further, both as a monotherapy (NCT04619004) and in combination with osimertinib (NCT04676477). BL-B01D1, a bispecific topoisomerase inhibitor-based ADC that targets both EGFR and HER3, is currently being evaluated in a phase I study (BL-B01D1-LUNG101; NCT05194982) for safety and efficacy in individuals with metastatic or unresectable NSCLC. Data from earlier clinical studies of BL-B01D1 were presented in 2023 at the American Society of Clinical Oncology, European Society for Medical Oncology (ESMO) and the San Antonio Breast Cancer Symposium; these data demonstrated impressive antitumor activity in patients with EGFR-TKI resistance, with an ORR of 63.2%.³⁴⁵ Both EGFR and HER3 are highly expressed in most epithelial tumors; thus, BL-B01D1 may have promising antitumor activity in patients with a range of solid tumors.

Trastuzumab deruxtecan (DS-8201, known as dequdacimab), a HER2-targeted ADC, has shown significant clinical efficacy in patients with lung cancer who harbor HER2 gene alterations. The treatment of advanced HER2-mutant NSCLC has historically been guided by the approach for advanced NSCLC without driver gene mutations. HER2-TKIs and anti-HER2 antibodies have shown limited efficacy in this setting,^{346, 347} while ADC drug have achieved breakthrough progress in HER2-mutant NSCLC. The DESTINY-Lung01 and DESTINY-Lung02 studies have demonstrated the efficacy of DS8201 in HER2-mutant NSCLC who have progressed after standard treatment or targeted HER2-TKI therapy, with ORR of 54.9 and 53.8%, respectively, and a median duration of response of 8.2 months.^{348, 349} In the DESTINY-PanTumor02 trial, which included 267 patients covering various tumor types such as endometrial cancer, cervical cancer, ovarian cancer, bladder cancer, and bile duct cancer, DS-8201 demonstrated significant efficacy in multiple HER2-positive solid tumors. In IHC3+ HER2-positive patients, the ORR reached an impressive 61.3%, with a PFS of 11.9 months and a median duration of response (DOR) of 22.1 months.³⁵⁰ Based on this study, on April 5, 2024, the US FDA granted accelerated approval to trastuzumab deruxtecan for previously treated and no further treatment options available, unresectable or metastatic HER2-positive (IHC3+) solid tumors. This marks a significant breakthrough for DS8201, allowing for potential treatment across multiple cancer types.

The potential role of alternative ADCs targeting the human trophoblast cell-surface antigen 2 (TROP2) (NCT04152499) and MET^{351, 352} (NCT03859752) is also being evaluated in patients with lung cancers, potentially including NSCLCs with EGFR mutations or ALK rearrangements. TROP2 is highly expressed in various solid tumors and is associated with worse prognosis.³⁵³ It has emerged as a hot target in the field of oncology in recent years, and the development of ADCs targeting this antigen has garnered significant attention. SKB264 (MK-2870) is a novel anti-TROP2 ADC. In a phase II study, 22 cases of EGFR mutation patients were enrolled who experienced progression during or after TKI treatment, 50% of patients had experienced at least one chemotherapy failure. The ORR was 60%, with a median PFS of 11.1 months.³⁵⁴ The ongoing phase III MK-2870-004 study (NCT06074588) will evaluate the efficacy and safety of MK-2870 versus chemotherapy for previously TKI-treated NSCLC with EGFR mutations or other genomic alterations in ALK, ROS1, BRAF V600E, NTRK, MET exon 14 skipping, or RET and those with less common EGFR mutations (exon 20 S768I, exon 21 L861Q, and/or exon 18 G719X). ABBV-399 (also called Telisotuzumab-Vedotin) is currently the only MET-targeting ADC that has entered phase III clinical trials. In January 2022, ABBV-399 was granted breakthrough therapy designation by the US FDA for the treatment of advanced or metastatic MET overexpressing EGFR WT nonsquamous NSCLC patients who have progressed after platinum-based therapy. In the single-arm phase II LUMINOSITY trial (NCT03539536), ABBV-399 demonstrated promising results in the treatment of MET protein overexpressing, EGFR WT, advanced/metastatic non-squamous NSCLC. The ORR was 53.8% in the high MET expression group and 25.0% in the moderate MET expression group, with a median duration of response of 9 and 7.2 months, respectively.³⁵⁵ ABBV-399 is not the only MET ADC drug in development, others like MYTX-011(NCT05652868), SHR-A1403(NCT03856541), TR1801(NCT03859752), and more are also in clinical trial stages. Targeting MET with ADCs provides a new treatment option for patients who develop MET overexpression after developing resistance to previous therapies.

3.4 PROteolysis targeting chimeras

PROteolysis targeting chimeras (PROTAC) is a molecular tool designed for targeted protein degradation. It is a heterobifunctional compound typically composed of three main parts: a binding moiety to target the protein of interest, a binding moiety to recruit the cellular ubiquitin E3 ligase, and a linker.^{356, 357} The mechanism of PROTAC relies on the dual binding of the target protein and the E3 ligase. When PROTAC binds to both the target protein and the E3 ligase, it forms a ternary complex. The E3 ligase then ubiquitinates the target protein, attaching a chain of ubiquitin molecules to it. The ubiquitinated target protein is recognized and degraded by the cell's ubiquitin-proteasome system, ultimately leading to the degradation of the target protein.^{358, 359} Traditional small molecule drugs may require higher drug concentrations to effectively inhibit the activity of the target protein, whereas PROTAC molecules can be reused after the degradation of ubiquitinated proteins, resulting in lower effective drug concentrations needed for PROTAC action.^{360, 361} Additionally, compared with traditional reversible inhibition, targeted degradation provides a longer duration of continuous blockade of the target protein, as the degraded protein needs to be resynthesized.^{360, 361} These advantages make PROTAC particularly suitable for addressing resistance caused by acquired mutations or amplifications in cancer.

Targeting proteins for proteasomal degradation using the PROTAC approach is now widely pursued in clinical development, mainly targeting proteins in cancer (e.g., BCR–ABL,^{362, 363} EGFR,^{361, 364, 365} androgen receptor (AR),^366-368 estrogen receptor (ER),^{369, 370} bromodomain-containing protein 4 (BRD4),³⁷¹ or Bruton's kinase (BTK)^{372, 373}). Bavdegalutamide (ARV-110) is an orally active, specific AR PROTAC degrader, which is in development for the treatment of men with metastatic castration-resistant prostate cancer (mCRPC) who have progressed on abiraterone and/or enzalutamide.³⁷⁴ Phase I/II study of ARV-110 (NCT0388861) demonstrated clinical activity in heavily pretreated mCRPC, with the greatest PSA50 activity and RECIST responses in patients with AR T878 and/or H875 mutations.³⁶⁸ Another example that demonstrates the clinical impact of this strategy is the BTK degrader developed to tackle on-target resistance to BTK inhibitors. The most common resistance mechanism in patients whose disease progresses on covalent BTK inhibitors is a mutation in the BTK C481 residue, abrogating the covalent binding of irreversible BTK inhibitors such as ibrutinib.^{91, 72} PROTAC technology itself does not require a strong affinity between the binder and the target protein. In fact, it has been shown that PROTAC molecules with ibrutinib as the binder, although having weaker binding to BTK C481S, can induce degradation of the BTK protein.³⁷⁵ This indicates that protein kinase conferring resistance to anticancer drugs may be targeted using the PROTAC approach.

4.1 Methods for detecting and monitoring resistance during treatment

Early monitoring of drug resistance mutations can determine the best treatment regimen, particularly with the development of high-quality detection methods in recent years. The current approach to disease management revolves around tumor-specific, stage-based guidelines, primarily relying on pathology and imaging outcomes to optimize patient treatment plans.³⁷⁶ Imaging techniques such as CT scans, MRI, or PET scans are valuable for monitoring disease progression or relapse and assessing treatment response; however, interpreting radiological findings can be challenging, resulting in high rates of false positives and negatives.³⁷⁷ Serial monitoring of tumor markers, such as carcinoembryonic antigen, cancer antigen (CA)-125, CA19-9, PSA, and lactate dehydrogenase, offers a noninvasive means to assess treatment response. However, many of these established biomarkers are deemed unreliable due to potential elevation from noncancer-related conditions, leading to reduced sensitivity and specificity.^378-380 NGS allows for the simultaneous analysis of multiple genes and molecular alterations in tumor samples. This provides a comprehensive view of the genomic landscape of cancer, including driver mutations, copy number alterations, and gene fusions.³⁸¹ Currently, NGS testing is based on tissue biopsy at the time of diagnosis and, when possible, a rebiopsy at the time of progression remains the gold standard, providing direct analysis of genetic mutations or alterations indicative of resistance.³⁸² However, tissue samples are not always feasible, and the sample amount is often insufficient for molecular detection.³⁸³ Additionally, tissue samples are not often enough to systematically represent the mutational status of the entire tumor due to tumor heterogeneity and clonal dynamics. In contrast, liquid biopsy offers a minimally invasive alternative, analyzing circulating tumor cells or cell-free DNA to detect real-time genetic changes associated with resistance.³⁸⁴ Dynamic sequencing analysis of circulating tumor DNA (ctDNA) has multiple potential clinical applications, including screening, distinguishing early-stage diseases, detecting molecular residual disease after curative surgery, predicting recurrence, genotyping advanced tumors, early assessment of treatment efficacy, monitoring treatment response, and identifying mechanisms of resistance, serving as an effective alternative to tissue-based genetic testing.^385-388 For example, in the monitoring of advanced colorectal cancer, ctDNA enables the dynamic and repeated assessment of changes in multiple key molecules of the tumor, such as EGFR, ERBB2, PIK3CA, and MAP2K1, among others. This facilitates the identification of novel potential therapeutic targets and mechanisms of resistance to targeted therapies, including anti-EGFR, anti-BRAF, and anti-HER2 drugs. Studies have demonstrated a high concordance between ctDNA detection results and tissue biopsy outcomes.^386-388 Moreover, ctDNA is advantageous due to its ability to capture molecular heterogeneity within and between tumors, unrestricted by sampling location.^{389, 390} Dynamic sequencing of ctDNA was also able to decode the evolutionary response of NSCLC patients who received targeted therapy. This approach effectively guides further research directions and clinical treatment practices.^{391, 392} In metastatic breast cancer, ctDNA demonstrates higher accuracy compared with standard serum markers like CA15-3. Dynamic changes in ctDNA are associated with PFS during chemotherapy, endocrine therapy, and combination targeted therapy.^393-396 Additionally, sequential ctDNA analysis can also be employed to assess genomic mechanisms of resistance that arise before clinical progression. In colorectal cancer patients receiving anti-EGFR monoclonal antibody therapy, RAS and EGFR-extracellular domain mutations were detected in ctDNA 10 months before radiographic progression, and these mutations disappeared after discontinuation of anti-EGFR therapy.^{397, 398} For advanced breast cancer patients receiving aromatase inhibitor (AI) and CDK4/6 inhibitor therapy, monitoring the emergence of estrogen receptor 1 (ESR1) mutations holds potential clinical utility. In the PADA-1 trial, patients with detected ESR1 mutations were randomized to receive fulvestrant with continued CDK4/6 inhibitor therapy, when this group was compared with those continuing AI and CDK4/6 inhibitor therapy, an improvement in PFS was observed. The PFS benefit observed with early fulvestrant use after detection of ESR1 mutations may not be compensated for by later fulvestrant use after disease progression.³⁹⁹ These data suggest that detecting resistance-related mutations in ctDNA earlier rather than later (at radiographic progression) may have more decision-making significance. Although ctDNA holds immense potential as a minimally invasive biomarker for tumor monitoring, its clinical application still faces several limitations and challenges. The optimal timing for dynamic assessment of ctDNA and the most accurate efficacy prediction thresholds require further investigation. To monitor the emergence of resistance mutations before clinical progression, further research is needed to determine the frequency of monitoring. Randomized intervention studies are necessary to assess whether changing treatment based on dynamic ctDNA assessment can improve prognosis or even replace current standard clinical metrics.

4.2 Identification of novel targets

The future of kinase inhibitors partly depends on the identification of novel targets. Current targeted kinase inhibitor therapies focus mostly on known tyrosine kinases, yet there are upward of 500 total kinases in the human genome. During the past 5 years, the number of small molecule kinase inhibitors entering clinical development has increased by over 200%.²²⁹ Currently, investigational kinase inhibitors in clinical trials target approximately 110 new kinase targets, accounting for 20% of the human kinome,²²⁹ reflecting the active exploration of innovative targets in the kinase field.

Aurora kinase family (Aurora kinase A, Aurora kinase B and Aurora kinase C) has been the most targeted kinase family in oncology, with over 20 drugs entering clinical trials. Aurora kinases are crucial mitotic regulators required for genome stability, and they are often overexpressed in human cancers such as breast cancer, ovarian cancer, gastrointestinal cancer, and this abnormal expression is associated with poor prognosis in cancer patients.⁴⁰⁰ In recent years, increasing evidence suggests that Aurora inhibitors hold promise for cancer therapy and overcoming resistance to inhibitors targeting CDK4/6 and EGFR.^{401, 402} Even though Aurora inhibitors hold treatment potential in oncology, so far, no Aurora kinase inhibitor has been approved for clinical use because of the limited clinical efficacy and drug toxicity. Currently, second-generation, isoform-selective Aurora kinase inhibitors are being investigated as monotherapies or in combinations (etc., LY3295668 combined with osimertinib (NCT05017025), MLN8237 and bortezomib in treating multiple myeloma (NCT010345535), VIC-1911 combined with lenvatinib (NCT05718882)) for hematological malignancies and solid tumors.

The CDK family has been a therapeutic target for over 20 years because of its critical roles in controlling cell division, and cancers often show dysregulation of the cell cycle.⁴⁰³ From 2015 to 2017, selective inhibitors targeting CDK4 and CDK6 gained US FDA approval for treating breast cancer, including palbociclib, abemaciclib, and ribociclib. Members of the CDK family continue to be a hotspot for the development of small-molecule kinase inhibitors. Apart from CDK4 and CDK6, there are currently multiple small molecule inhibitors targeting CDK7, CDK8/19, CDK2, and CDK9. CDK2, a critical regulator of the cell cycle, is a promising drug target, especially in cancers with cyclin E1 protein overexpression, including HR+/HER2− breast cancer resistant to CDK4/6 inhibitors.⁴⁰⁴ Multiple companies, including AstraZeneca, Pfizer, and Blueprint Medicines, are actively developing CDK2 inhibitors. Currently, several CDK2 inhibitors have entered clinical trial stages, being tested both as monotherapy and in combination with CDK4/6 inhibitors.

Hematopoietic progenitor kinase 1 (HPK1) is a family member of mitogen-activated protein kinase kinase kinase kinase, which is a kinases that catalyze the phosphorylation of serine or threonine residues in proteins.^{405, 406} HPK1 is an intracellular negative regulator of T/B-cell proliferation and signaling. HPK1 kinase knockdown mice demonstrate increased CD8+ T-cell killing function and robust antitumor immune responses even in a cold tumor environment, making HPK1 a high-priority target in immuno-oncology.^{407, 408} Several lines of preclinical evidence have been collected over the past years that support the development of compounds targeting HPK1, especially in combination with anti-PD-1/PD-L1 ICIs. In 2020, Treadwell Therapeutics' HPK1 inhibitor CFI-402411 became the first to enter phase I clinical trials (NCT04521413). Subsequently, orally highly selective HPK1 inhibitors such as Pfizer's PF-07265028 (NCT05233436), BeiGene's BGB-15025 (NCT04649385), and Zhuhai Yufan Biotechnologies’ PRJ1-3024 (NCT05315167) have entered phase I/II clinical trials. These inhibitors are primarily being investigated as monotherapy or in combination with PD-1/PD-L1 inhibitors for the treatment of advanced solid tumors.

SHP2 is a phosphatase that dephosphorylates substrate proteins. It is a “star target” in the field of cancer therapy. In the RTK pathway, SHP2 effects are upstream of RAS, in mediating tumor proliferation. Virtually all RTKs initiate the RAS signaling pathway by activating SHP2. Thus, inhibitors of SHP2 can block the RTK/RAS/MAPK signaling pathways and inhibit the growth and proliferation of tumor cells driven by RTK or with KRAS, BRAF Class 3 and NF1 loss of function mutations.^{245-248, 409} SHP2 effects are also downstream of the immune checkpoint regulator, PD-1, in inhibiting the antitumor effect of T cells.^410-412 Given the clinical success of anti-PD-1/PD-L1 immunotherapy, SHP2 inhibitors serve as small molecule tumor immunotherapeutic agents and are considered important complements to PD-1/PD-L1 inhibitors, holding great promise in clinical practice. SHP2 also functions downstream of colony stimulating factor 1 receptor, in promoting TAM activity. Thus, targeting SHP2 has the potential for multiple antitumor effects. Currently, there are no drugs targeting the SHP2 target approved for market, but several SHP2 inhibitors have entered clinical trials. These include Novartis’ TNO155 (NCT03114319), Sanofi's RMC-4630 (NCT04916236), Etern BioPharma's ET0038 (NCT05354843) and Jacobio Pharma's JAB-3312 (NCT04720976). The main indications for these inhibitors include combination therapy with KRAS G12C inhibitors, EGFR inhibitors, BRAF inhibitors, MEK inhibitors, ERK inhibitors, and anti-PD-1/PD-L1 antibodies. Among them, the most rapidly advancing is JAB-3312, which has become the world's first SHP2 inhibitor to be approved for phase III pivotal clinical trials in combination with KRAS G12C inhibitors, leading the field of SHP2 inhibitors. Data presented at the 2023 ESMO Congress revealed that in the dose group combining the KRAS G12C inhibitor sotorasib with the SHP2 inhibitor JAB-3312, the objective response rate was 50%, with a disease control rate of 100%.⁴¹³

At least 70% of the kinome is still unexplored in drug development. New large-scale screening methods such as large-scale quantitative proteomic screens and scRNA-seq may help identify transcript levels of kinases that drive clonal evolution or transcriptional reprogramming, holding promise for the development of novel targeted kinase inhibitor therapies.

4.3 Dual-acting kinase inhibitors

As mentioned above, combination therapy theoretically may have additive or synergistic effects; however, it often leads to unpredictable side effects, such as increased toxicity. As an alternative strategy to combination therapy, dual-target or multitarget drugs are considered to have lower risks of drug–drug interactions and better pharmacokinetics (PK) and safety profiles. Compared with combination therapy, these drugs mitigate issues such as reduced bioavailability, metabolism, and antagonistic effects. The advantage of this approach lies in its ability to simultaneously target multiple points, thereby providing a more comprehensive therapeutic effect against complex disease biology processes. By reducing interactions between drugs, dual-target or multitarget drugs may help reduce the risk of side effects and improve the safety and tolerability of treatment. Currently, many dual-target drugs have been investigated in cancer treatment. For example, alectinib is a dual-target small molecule inhibitor that simultaneously targets ALK and RET. Next-generation RET-TKI TPX-0047 simultaneously targets RET and SRC.⁴¹⁴ The dual-target CHK1/FLT3 inhibitor TLX83 can overcome adaptive and acquired resistance to FLT3 inhibitors, offering a new strategy for patients with FLT3-ITD-mutant acute myeloid leukemia (AML).⁴¹⁵ The development of EGFR dual-target inhibitors is a promising strategy to decrease the potential drug resistance of EGFR-TKIs induced by aberrant alternative signaling pathways. Preclinical studies have shown that the combination of erlotinib and HDAC inhibitors promotes apoptosis in TKI-resistant NSCLC cells.⁴¹⁶ CUDC-101, a multitarget HDAC/EGFR/HER2 inhibitor, has shown effective inhibitory activity against advanced solid tumors in vivo.⁴¹⁷ Phase I study (NCT00728793) demonstrated CUDC-101 was generally well tolerated with preliminary evidence of antitumor activity.⁴¹⁸ A new MEK inhibitor currently in clinical trials for metastatic melanoma (NCT04720417) simultaneously inhibits RAF feedback activity. So far, various design strategies for dual-target inhibitors targeting different tumor types and pathways have been reported.

As mentioned above, the future development of kinase inhibitors faces challenges beyond the discovery of new targets, particularly in developing compounds with high clinical efficacy and low risk of off-target toxic side effects and resistance. In this regard, developing inhibitors with different binding modes, such as allosteric and covalent inhibitors, as well as developing dual-acting kinase inhibitors, or targeted protein degraders, may play a crucial role in kinase drug development in the next 20 years.

4.4 Nanoparticle-based drug delivery systems

The combination of nanotechnology and targeted therapy has emerged as a promising field in cancer research. Nanoparticle-based drug delivery systems, including polymeric, lipid, carbon nanotubes, dendrimer, and micellar nanoparticles have been extensively designed and investigated in preclinical cancer treatment.^419-421 Compared with the conventional delivery routes, nanoparticle-based drug delivery systems possess some unique advantages with improved PKs, increased efficacy, and reduced off-target cytotoxicity. For example, nanoparticle-based drug delivery systems can be utilized to codeliver two different drugs or reagents concurrently. A codelivery system of afatinib and paclitaxel was designed for the treatment of EGFR-TKI-resistant NSCLC.⁴²² The results of this study showed that afatinib and paclitaxel from the system can sustain high concentrations in the lung for up to 96 h, while maintaining lower concentrations in other tissues in Sprague–Dawley rats.⁴²² Besides, as mentioned above, hypoxia in solid tumors can mediate acquired resistance to TKI therapy. Li et al.⁴²³ codelivered oxygen-releasing perfluorooctylbromide with erlotinib in liposomal nanoparticles to address hypoxia-induced drug resistance. They found the codelivery system could significantly improve the efficacy of erlotinib when compared with nanoparticles loaded with erlotinib alone.⁴²³

However, nearly all US FDA-approved nanomedicines are formulated for intravenous (i.v.) use, while the majority of TKIs are designed for oral. More advanced nanocarriers capable of overcoming the challenges of oral administration will be needed. In addition, most of these studies are limited to preclinical settings. Nano-applications of TKIs are still in their infancy stages of development, and yet to be validated in clinical practice. Despite their immense potential, they still have a considerable journey ahead of them.

4.5 Targeting drug-tolerant persistent cells (DTP cells)

In recent years, DTP cells have gained considerable attention as a promising area in cancer research, particularly regarding their role as nongenetic mechanisms of therapeutic resistance.⁴²⁴ DTP cells are a subpopulation of cancer cells that survive exposure to targeted therapy, despite the treatment being initially effective against the bulk of tumor cells.⁴²⁵ These cells can enter a quiescent or slow-proliferative state, allowing them to evade the cytotoxic effects of the drug.^{425, 426} While DTP cells may initially respond to treatment, they can eventually lead to disease recurrence or the development of drug resistance.^{425, 426} Understanding and targeting DTP cells is crucial for improving the long-term efficacy of cancer-targeted therapies.

In cancer, DTP cells were first found and described in the EGFR-mutant NSCLC cell line PC9 by Sharma and colleagues in 2010.⁴²⁷ They observed that these cells exhibited over 100-fold lower drug sensitivity and could revert to their original state upon drug withdrawal.⁴²⁷ An epigenetic regulator, histone demethylase KDM5A, was found to be required for the establishment of DTP cells, and histone deacetylase (HDAC) inhibitors could ablate the generation of DTP cells.⁴²⁷ Subsequently, DTP cells have been identified in various experimental models after targeted therapy, including additional EGFR-mutant NSCLC cell lines, MET-amplified gastric cancer, BRAF-mutant melanoma, prostate cancer, colorectal cancer, and HER2-positive breast cancer.^428-433 Two common phenotypic features of DTP cells, including slow proliferation and reversible biological capacity, have been well documented in different studies.⁴³⁴ However, the mechanisms responsible for regulating DTP state are context-specific across various cancer models and therapeutic approaches, which include diverse epigenetic reprogramming, transcriptional regulation, and metabolic remodeling.^{425, 434}

Despite accumulating evidence regarding the existence of DTP cells and the putative mechanisms underlying the regulation of DTP state, universally accepted biomarkers for DTP cells remain elusive. The definition of cancer persistent cell state is yet to be established. Besides, most studies characterized DTP cells by using in vitro cell line experimental models,^{427, 428} with few studies employing in vivo patient-derived xenograft mouse models.^{431, 435} Data on patients are still limited. Our clinical understanding of their roles in cancer treatment resistance remains unclear, and therapeutic strategies targeting DTP cells have yet to be explored. Based on the concept of DTP cells, a potential therapeutic strategy of withdrawing targeted therapy and enabling tumor repopulation with drug-sensitive cells has been discussed. However, tumor evolution is unpredictable. What's more, clinical experience over decades has demonstrated attempting to reintroduce the same therapy to a patient is consistently futile.⁴²⁴ Additionally, DTP cells have been proven to be highly heterogeneous and context-specific. With the advancement of single-cell multiomics technologies, including spatial transcriptomics, epigenomics, proteomics, and metabolomics, a more comprehensive understanding of DTP cells may be established and elucidated.

4.6 Investigating resistance mechanisms at single-cell resolution

Most resistance mechanisms discussed above were uncovered through the analysis of bulk tumor specimens. Considering that tumors are highly heterogeneous, bulk-tumor analysis may not completely reveal the entire spectrum of resistance mechanisms.⁴³⁶ Single-cell analyses of matched tumor specimens or cell line models before and after treatment are informative in identifying potential mechanisms of TKI-therapy resistance in various studies.^{432, 437-440} For example, by performing single-cell RNA sequencing (scRNA-seq) and single-cell ATAC-sequencing analysis on both TKI-resistant cells and clinical specimens, Kashima and colleagues identified CD74 as a novel key factor in regulating tumor drug-tolerant state.⁴³⁷ Although single-cell RNA-seq holds promise in deciphering the transcriptomic profiles of tumor subclones, it lacks coverage of critical mutation information in cells. As we discussed above, most well-reported resistance mechanisms are based on secondary mutations. An innovative genomic technology for single-cell mutational analysis and concurrent transcriptomic sequencing will allow for further understanding of molecular mechanisms underlying treatment resistance, by providing insights into dysregulated pathways in both mutant and nonmutant tumor cells. Alba and colleagues developed such a method for the detection of hotspot mutations, in parallel with unbiased RNA-seq analysis within single cells.⁴⁴¹ However, this method has not yet been applied to comprehensively explore resistance mechanisms of cancer treatment.

Multiomics techniques, particularly NGS and scRNA-seq, have revolutionized our understanding of tumor biology by uncovering new biomarkers and molecular regulators linked to oncogenesis, metastasis, and drug resistance.^{438-440, 442} Nonetheless, these methods have limitations as they do not capture the spatial organization of cells within the TME, thus providing an incomplete picture of tumor biology. There is an increasing demand for image-based methodologies with single-cell resolution to visualize intratumor heterogeneity, cell–cell interactions, and protein expression information within the TME. Emerging technologies that utilize multiplexed fluorescence, RNA, DNA, and isotope labeling now enable the deep investigation of cancer biology within their spatial context.^{443, 444} By conducting single-cell multiomic assays including nucleus RNA, open-chromatin, spatial transcriptomics, spatial proteomics and exome sequencing, on 86 matched primary-recurrent tumor specimens, Lin and colleagues comprehensively described a single-cell atlas of glioblastoma evolution under treatment.⁴⁴⁵

While multiomics techniques offer immense potential in cancer research, several challenges hinder their widespread application. One major obstacle is the resistance mechanisms identified by multiomics techniques are usually individual-specific and context-specific, making it difficult to develop universal therapeutic approaches applicable to all patients. Moreover, the cost and time required for multiomic analyses are substantial, limiting their accessibility and scalability. Besides, tumors exhibit high levels of heterogeneity, both within and between tumors. Thus, single-cell resolution is crucial for capturing this heterogeneity, but it also introduces complexity to data analysis, further complicating the interpretation of results. Overcoming these challenges will require advancements in technology, computational methods, and collaborative efforts within the scientific community to fully leverage the potential of single-cell multiomic technologies in cancer research.