Targeting CDK2 for cancer therapy

SUMMARY

Targeting cell-cycle regulatory processes by inhibiting cyclin-dependent kinases (CDKs) has long been considered a significant therapeutic strategy for oncology. Recent studies have highlighted the complexity of targeting CDK2 for cancer therapy. Unlike CDK4/6 inhibitors, CDK2 inhibitors can impact different phases of the cell cycle by modulating distinct effector pathways, and the response to CDK2 inhibitors is controlled by the genetic and epigenetic makeup of the tumor. Biomarkers have emerged that can inform the effective use of these drugs and include cyclin E and p16INK4A. Work across several different tumor types indicates that CDK2 inhibitors can be combined effectively with various drug classes. However, more investigation is needed to understand the potential limitations and drug toxicities of existing CDK2 inhibitors and those in development.

INTRODUCTION TO CDK-CYCLINS AND CDK INHIBITORS

The drivers of eukaryotic cell division have been known for over 25 years to encompass specialized cyclin-dependent kinases (CDKs) that act with their cognate cyclins to drive cell-cycle transitions. Initially defined in model organisms, CDK-cyclins are critical for the division of all mammalian cells.^1,2 Since cancer is associated with deregulation of proliferation, targeting CDK-cyclins has been a long-term goal in oncology.^3,4 Most oncogenic signaling deregulates G1/S control through inactivation of the retinoblastoma (RB) tumor suppressor protein, and targeting this transition specifically has been most effective with CDK inhibitors.^5–7

In mammalian cells, different CDK-cyclins have been thought to be critical for different transitions through the cell cycle. Historically, CDK4 or CDK6 in complex with D-type cyclins (cyclins D1, D2, and D3) were considered the focal point of coupling mitogenic signaling pathways to the initiation of the cell cycle in early G1⁸ (Figure 1). These complexes have a limited substrate repertoire and are important for initiating the phosphorylation of RB and related proteins, p107 and p130.^9,10 A principal action for RB, p107, and p130 in the cell cycle is mediating the repression of the E2F family of transcription factors.¹¹ There is a broad spectrum of genes coordinated by this transcriptional regulation, including CDKs, cyclins, and required genes for DNA replication and mitotic progression.^12–15 CDK2 and its regulation were considered largely downstream of RB, because the expression of CDK2 and key associated cyclins (cyclin E1, cyclin E2, and cyclin A) are at least partly regulated via RB-E2F^7,16–18 (Figure 1A). CDK2 phosphorylates many targets beyond the RB family and has been considered critical for G1/S transition, DNA replication, and G2/M progression^19,20 (Figure 1). However, this stepwise view of the cell cycle has been challenged by findings in multiple tumor models indicating that dominant individual CDK activities, either CDK4, CDK6, or CDK2, may be largely responsible for traversing G1/S in certain contexts.⁵

Figure 1. Distinct cell cycles under different conditions.

(A) The conventional depiction of the cell cycle with initiation occurring with coordinated phosphorylation of RB and related proteins by CDK4/6 and CDK2 complexes. This event releases E2F activity to promote further activation of CDK2 and progression through G1/S and additional phases of the cell cycle.

(B) In the context of resistance to CDK4/6 inhibitors, the activity of CDK2 either in complex with cyclin E or cyclin D1 can bypass the requirement for CDK4 or CDK6 activity.

(C) In the context of tumors with high expression of p16INK4A, the transition through G1/S is more dependent on CDK2. This dependence can be apparent in both RB-proficient and RB-deficient tumor models.

While CDK inhibitors were first deployed clinically ~25 years ago, toxicity of early broad-acting inhibitors such as flavopiridol or dinaciclib limited drug development.^21,22 Selective CDK4/6 inhibitors harbor less toxicity and have been found to be particularly effective in HR+/HER2− breast cancer.^4,21,23 This clinical activity led to FDA approval and wide utilization of CDK4/6 inhibitors for the treatment of HR+/HER2− breast cancer as a standard of care. While these drugs are effective at essentially doubling the progression-free survival of patients with metastatic HR+/HER2− breast cancer,^24,25 veritably all patients ultimately progress on therapy. Additionally, in multiple disease settings beyond HR+/HER2− breast cancer, to date CDK4/6 inhibitors have not demonstrated sufficient clinical activity to justify their adoption.^4,26 This lack of efficacy has been found in molecular focused trials (e.g., the NCI-MATCH trial) as well as with combinations that were highly effective in preclinical models.^4,27 The success and challenges of CDK4/6 inhibitors have led to renewed efforts to more broadly target the cell cycle with other selective CDK inhibitors.

Complex rationale for targeting CDK2 in cancer therapy

Our understanding of the biology of CDK2 and its stock as a therapeutic target for cancer has varied dramatically over the last 20 years. Initially, it was suspected that CDK2 is universally required for mammalian cells to divide. This supposition was based on key roles of CDK2 in the inactivation of RB by hyperphosphorylation, initiation and progression of DNA replication, and features of the G2/M transition.^5,20,28,29 However, genetic deletion of CDK2 in mice is surprisingly well tolerated, and many cancer cells were found to continue to divide in the absence of CDK2.^30–32 Analyses of large datasets with genetic deletion of CDK2 illustrated that >90% of cancer cell lines can continue to proliferate in the absence of CDK2.^33,34 Such findings, coupled with the challenge of selectively inhibiting CDK2 without affecting CDK1, diminished overall enthusiasm for pursuing CDK2 as a therapeutic target. Interestingly, CDK2-deficient mice are infertile due to defects in meiosis,³⁰ which has supported developing CDK2 inhibition in the context of male contraception.^35,36

Several recent findings expanded the significance of CDK2 as a cancer target. First, deregulation of CDK2 is considered one of the key mechanisms through which cells acquire resistance to CDK4/6 inhibitors.^4,37–39 Notably, cyclin E overexpression is associated with a short duration of response to CDK4/6-inhibitor-based therapies in HR+/HER2− breast cancer patients,³⁸ and cyclin E overexpression in preclinical models functionally drives resistance.⁴⁰ Additionally, many of the other genetic drivers of resistance impinge on CDK2 to enable bypass of CDK4/6 inhibitors.^5,39,41 For example, AMBRA1 deletion was shown to increase the abundance of CDK2-cyclin D1 complexes that enable escape from CDK4/6 inhibition.^42,43 Importantly, depletion of cyclin E or CDK2 also appeared to enhance the response to CDK4/6 inhibitors^33,37,41 (Figure 1B). Second, CCNE1 amplification has been progressively recognized as an oncogenic driver in a variety of cancers.^3,44 Particularly in gynecological malignancies, the frequency of CCNE1 amplification is quite high, and amplification occurs sporadically in many solid tumors.^45,46 In these tumors, the depletion of CDK2 or CCNE1 can serve to inhibit proliferation, which suggests that, at least in the context of CCNE1-amplified tumors, targeting CDK2 could be very effective in treating these aggressive tumors with particularly poor prognosis (Figure 1C). Thus, several indications have been identified for which clinical inhibition of CDK2 could be a successful strategy.

Chemical strategies for CDK2 inhibition

The most widely pursued strategy for CDK2 inhibition has been the development of type I competitive kinase inhibitors that target the ATP-binding site of the active CDK2-cyclin E heterodimeric enzyme.⁴⁷ One approach has been to use CDK4/6 inhibitors as a starting point and optimize potency and selectivity for CDK2.^48,49 For example, the Incyclix CDK2 inhibitor INX-315 was derived from the CDK4/6 inhibitor trilaciclib; key modifications to the molecule enhance interactions with sidechains in and near the kinase hinge region that are specific for CDK2.⁴⁸ The CDK2/4/6 inhibitor from Pfizer was derived from the CDK4/6 inhibitor palbociclib by similarly beginning with replacement of the hinge-interacting piperazine-pyridine group⁴⁹ (Figure 2). In contrast, other compounds, for example, BLU-222 from Blueprint Medicines and a pyrazolyl-aminopyrimidine molecule from Incyte, were identified from a kinase inhibitor library, with a bias toward pursuing screening hits that showed CDK2 selectivity over other CDK family members.^50–52 Other CDK2 inhibitors have been recently discovered through in silico screening and artificial intelligence approaches.^53,54 Overall, a remarkable number of scaffolds have been identified,⁴⁷ including molecules containing pyrazolyl-pyrimidine⁵¹ or fused bicyclic heterocycle cores,⁴⁹ fused tricycles,⁴⁸ and macrocycles.⁵⁴

Figure 2. Structural comparison of CDKs bound to type I ATP-competitive inhibitors.

(Left) Crystal structure of CDK2 bound to PF-06873600 (PDB: 7kjs). (Middle) Structural model of palbociclib bound to CDK4 created by overlaying the position of the drug in the palbociclib-CDK6 crystal structure (PDB: 5l2i) with the position of CDK4 in the abemaciclib-CDK4-cyclin D3 crystal structure (PDB: 7sj3). One approach to engineering CDK2 potency has been replacement of the palbociclib piperazine-pyridine ring system with a group that contacts CDK2-specific residues near the hinge region of the kinase. For example, the aminopiperidine sulfonamide group in PF-06873600 makes a hydrophobic contact with F82 in the CDK2 hinge, which corresponds to H95 in CDK4, and forms a hydrogen bond with K89 in the CDK2 helix proximal to the hinge (T102 in CDK4). (Right) Structural overlay of PF-06873600 and CDK1 in the CDK1-cyclin B-Cks2 crystal structure (PDB: 4yc3). Notably, all the sidechains around the ATP binding pocket are identical in CDK1 and CDK2, so the origins of the modest selectivity of this compound and other CDK2 specific inhibitors remain uncertain.

The underlying challenge to developing type I inhibitors has been the high homology between CDK2 and CDK1. The two enzymes are ~65% identical and contain the same set of amino acids forming the ATP binding site (Figure 2). Since CDK1 inhibition has been associated with toxicity in preclinical models and clinical trials, selectivity for CDK2 over CDK1 has been prioritized along with overall potency. Some of the compounds used in the clinic have showed modest selectivity; INX-315, BLU-222, and the Pfizer CDK2 inhibitor PF-07104091 all demonstrate ~50 to 100-fold higher potency inhibiting CDK2-cyclin E compared to CDK1-cyclin B in enzyme assays and show similar selectivity in cells.^48–50 Unfortunately, due to a lack of published structural information characterizing these molecules bound to CDKs, the origin of the selectivity is unclear. It has been observed that selectivity of several ATP-competitive inhibitors, including dinaciclib, shows better CDK2 selectivity over CDK1 when assaying binding to the inactive monomer compared to the active heterodimers.⁵⁵ At the same time, several compounds have been discovered that bind preferentially to the inactive kinase conformation like type II inhibitors; interestingly, some of these compounds inhibit association of activating cyclins.^56–58 These results suggest the potential benefits of an alternative strategy of targeting the inactive CDK2 monomer and preventing formation of the active complex.

Two other chemical strategies toward CDK2 inhibition include the use of allosteric inhibitors and proteolysis-targeting chimeras (PROTACs). An allosteric pocket in CDK2, which was originally identified through interaction of CDK2 with the dye molecule and promiscuous protein ligand ANS, sits adjacent to the orthosteric ATP site in the N-lobe of the kinase domain.^36,59 Molecules that bind this pocket have improved CDK2 specificity and negatively impact cyclin binding due to their effects on the orientation of the kinase C-helix. Another allosteric site, which sits near the CDK2-cyclin E interface, was identified from a library of covalent electrophilic probes.⁶⁰ Outside of the kinase domain, there have been some efforts to target the substrate recruitment site in the cyclin domain.^61,62 Macrocycles that bind to this so-called RxL site in cyclin A have been developed and shown to induce apoptosis in part through preventing proper CDK2 interaction with and inactivation of activator E2Fs.⁶² A few promising CDK2 PROTAC degraders have been reported, which covalently link ATP-competitive CDK2 inhibitors to molecules that bind the E3 ligase cereblon (CRBN).^47,63,64 Interestingly, in two cases, the resulting PROTAC molecules had much better CDK2 selectivity over CDK1 (~1,000-fold) compared to the original ATP-competitive inhibitors from which they were built.^63,64 This improved specificity likely results from CDK2 interactions with the linker or from additional interactions between the CDK2-cyclin E complex and CRBN within the quaternary drug-protein complex. In addition to enhanced CDK2 selectivity, degraders and cyclin association inhibitors may confer distinct advantages in that they can change the concentrations and distributions of CDK-cyclin complexes in the cell. It remains to be seen what preclinical and clinical contexts will benefit most from such alternative strategies.

CDK2 catalytic inhibitor mechanism of action and distinction from genetic dependencies

Relatively few cancer cells harbor a genetic dependency for CDK2 or CCNE1, while CCNA2 is largely required.³³Studies performed in panels of cell lines have demonstrated that cell lines that are highly dependent on CDK2 or CCNE1 are potently arrested by the CDK2 catalytic inhibitors INX-315 and BLU-222.^48,50,52,65 Multiple additional inhibitors have been presented in published abstracts, including NKT3694, BG-68501, and AZD8421, which are in clinical trials. In the highly sensitive models that overexpress cyclin E and p16INK4A, the CDK2 inhibitors generally lead to a G1-arrest, which is accompanied by the dephosphorylation of RB and repression of E2F target genes (Figure 3A). Additionally, in such models, suppression of CDK2 activity can be observed through a lack of phosphorylation of nucleolin (NCL) and the CDK-substrate DHB sensor.⁶⁶Thus, in RB-proficient cancer cells overexpressing cyclin E and p16INK4A, CDK2 inhibition functions similarly to CDK4/6 inhibition in CCND1-driven tumors.^48,50,52,65 These CDK2-sensitive CCNE1-amplified models are typically resistant to CDK4/6 inhibitors, reflecting distinct cell-cycle regulatory mechanisms governing S-phase progression.^5,48,50 Notably, CDK2 inhibitor monotherapy also potently suppresses proliferation in vivo in these models.^{5,48,50,52,65} The CDK2 inhibitor-mediated G1-arrest in CCNE1-amplified, p16INK4A-overex-pressing tumors is critically dependent on RB. In contrast, in RB-deficient tumors, including small cell lung cancer, CDK2 inhibitor activity has been linked to G2/M checkpoint activation.^50,62

Figure 3. Distinct mechanisms through which CDK2 catalytic inhibitors can suppress cell cycle progression.

(A) (Left) In tumors that express high levels of p16INK4A and cyclin E, the catalytic inhibitors prevent the phosphorylation of substrates including RB and nucleolin and suppress the phosphorylation of sensors for CDK activity. These molecular events lead to the repression of E2F and suppression of cyclin A, which are associated with a G1 arrest. (Right) In tumors that do not express p16INK4A, the treatment with catalytic CDK2 inhibitors does not inhibit RB phosphorylation or other molecular events. In this condition, the catalytic inhibitor induces an enrichment of cells with a 4N DNA content. The inclusion of a CDK4/6 inhibitor with the CDK2 catalytic inhibitor can phenocopy features of p16INK4A expression and lead to RB dephosphorylation and other molecular effects.

(B) Comparison between genetic depletion and catalytic inhibition of CDK2. While many cells tolerate CDK2 depletion by relying on cyclin A/CDK1 for G2/M progression, catalytic inhibitors maintain inactive CDK2-cyclin complexes, resulting in the accumulation of cells with 4N DNA content. The exact nature of this arrest remains under investigation and could be related to multiple functions of CDK2 in cell biology.

Surprisingly, many cell models that are not particularly dependent on CDK2 can still be arrested with CDK2 catalytic inhibitors, albeit requiring a higher drug dose.^48,65 In these contexts, CDK2 inhibition is dispensable for RB inactivation, as CDK4/6 activity compensates for CDK2-inhibitor-induced attenuation of RB phosphorylation.⁶⁷ This compensation from other CDK activity can also be observed in persistent CDK substrate sensor and nucleolin phosphorylation.⁶⁷ However, CDK2 inhibition induces a G2 enrichment that is not related to RB dephosphorylation or suppression of E2F target genes (Figure 3A). This alternative arrest is observed in models of HR+/HER2− breast cancer and other tumor types that do not harbor CCNE1 amplification.^48,65 The higher inhibitor concentrations needed to induce the G2 enrichment may suggest that the effect is from “off-target” inhibition of CDK1. However, the depletion or deletion of CDK2 can overcome the G2 arrest induced by INX-315 and PF-4091.⁶⁵ Furthermore, in unbiased CRISPR screens, CDK2 was the top driver of resistance to the catalytic inhibitor.⁶⁵ These findings indicate that the G2 arrest represents an “on-target” effect of CDK2 catalytic inhibitors. This feature of the catalytic CDK2 inhibitors is perhaps related to the biology of dominant negative CDKs that can inhibit the cell cycle.⁶⁸ Since these same tumor models are not affected by CDK2 deletion, these findings highlight a key distinction between depletion of CDK2 and catalytic inhibition (Figure 3B).

A key question remains: what is the exact nature of the G2 inhibition (Figure 3B)? One possibility is that the catalytically inhibited CDK2 complex sequesters cyclins or other co-factors, for example cyclin A, which is important for mitotic activation of CDK1.⁶⁹ Such a supposition is supported by the finding that this form of cell-cycle inhibition is similar to that observed with CCNA2 depletion.^50,65 An alternative possibility is that the inhibited CDK2 complexes cannot phosphorylate substrates required for the progression through G2/M (e.g., completion of DNA replication). Similarly, CDK2 is important for the synthesis of histones, and catalytic inhibition could compromise histone pools, which is known to induce a G2 arrest.^70,71 These possibilities are supported by data that CDK2 inhibitors induce some features of a G2/M checkpoint that reflect incomplete DNA replication or DNA damage.⁶⁵ Lastly, the CDK2 inhibitors could be preventing activation of mitotic transcription factors such as B-MYB and FOXM1, which are known CDK substrates that require phosphorylation to upregulate mitotic gene expression.^12,72

Irrespective of the precise mechanism, it seems that networks that control G2/M represent key mediators of response to CDK2 inhibitors in the majority of cancer cells, as observed through CRISPR screens in which deletion of G2/M drivers is associated with improved response to catalytic CDK2 inhibitors.⁶⁵ Key unanswered questions are whether CDK2 degraders and complex disruptors will mediate this same effect and whether they will need to be deployed solely against those tumors that are genetically dependent on CDK2 or deployed in combinations as discussed below.

Biomarkers related to different stages of cell-cycle arrest with CDK2 inhibitors

The dramatic differences in response to CDK2 inhibitors underscore the importance of biomarkers to direct their use. Initially, CCNE1 amplification emerged as an obvious biomarker and was used to define the clinical landscape for most clinical trials (Table 1). The role of CCNE1 amplification and expression in sensitivity to CDK2 depletion is also apparent in DepMap.³³ Since cyclin E can become overexpressed through multiple mechanisms, for example, deregulated E2F activity¹⁴ or loss of FBXW7,⁷³ the extent to which the overexpression of cyclin E protein via other mechanisms is a marker for sensitivity to CDK2 inhibitors is largely unknown.

Table 1.

Summary of trials with CDK2-selective inhibitors

Study	Summary from Clinicaltrials.gov	Reported clinical findings
Study of AVZO-021 in Patients With Advanced Solid Tumors (NCT05867251)	phase 1/2 study of AVZO-021 for advanced/metastatic cancers that have progressed on standard therapy including HR+/HER2− breast cancer and CCNE1-altered malignancies. Monotherapy of AVZO-021 is being interrogated as well as combinations with endocrine therapy (fulvestrant and letrozole), CDK4/6 inhibitors (ribociclib and abemaciclib), the antibody-drug conjugate sacituzumab govitecan-hziy, and carboplatin. Estimated participants: 430	no results available
Study to Evaluate the Safety, Tolerability, PK, and Efficacy of INX-315 in Patients With Advanced Cancer (NCT05735080)	phase 1/2 study of INX-315 for advanced/metastatic cancers that have progressed on standard therapy, including ovarian cancer with CCNE1 amplification and HR+/HER2− breast cancer. Monotherapy of INX-315 is being tested as well as combinations with fulvestrant and the CDK4/6 inhibitor abemaciclib. Estimated participants: 150	SABCS, 2024a N = 27 patients reported treatment-related grade ≥3 adverse events included fatigue (7%) and decreased white blood cell count (7%), as well as diarrhea, anemia, and decreased neutrophil count (4%) two patients had partial responses, and 14 patients (63.6%) had stable disease
PF-07104091 as a Single Agent and in Combination Therapy (NCT04553133)	phase 1/2 study of PF-07104091 for advanced/metastatic cancers that have progressed on standard therapy including small cell lung cancer, ovarian cancer, triple-negative breast cancer, HR+/HER2− breast cancer, and non-small cell lung cancer. Monotherapy of PF-07104091 as well as combinations with endocrine therapy (letrozole and fulvestrant) and the CDK4/6 inhibitor palbociclib are being tested. Estimated participants: 154	ASCO 2023b N = 35 patients reported treatment-related grade ≥3 adverse events included nausea (14.3%), diarrhea (8.6%), fatigue (20.0%), and anemia (8.6% G3) among 16 metastatic breast cancer patients, partial responses were observed in 3 (18.8%) patients, and 6 (35.7%) patients had stable disease
(VELA) Study of BLU-222 in Advanced Solid Tumors (NCT05252416)	phase 1/2 study of BLU-222 for advanced/metastatic cancers that have progressed on standard therapy including HR+/HER2− breast cancer and CCNE1-amplified tumors (ovarian, uterine, and gastric). Monotherapy of BLU-222 as well as combination therapies with carboplatin, fulvestrant, and the CDK4/6 inhibitor ribociclib are being tested. Estimated participants: 366	ASCO 2024c N = 64 patients reported the most common treatment-related adverse events included gastrointestinal events (nausea, diarrhea, and vomiting), fatigue, anemia, photophobia, and hypokalemia
A Study with NKT3964 for Adults with Advanced/Metastatic Solid Tumors (NCT06586957)	phase 1/2 of study of NKT3964 for advanced/metastatic cancers that have progressed on standard therapy including ovarian cancer, endometrial cancer, gastric cancer, gastroesophageal junction esophageal adenocarcinoma, and other solid tumors with CCNE1 amplification; small cell lung cancer; triple negative breast cancer; and HR+/HER2− breast cancer. Monotherapy of the NKT3964 is being tested. Estimated participants: 90	no results available
A Study with NKT3447 for Adults with Advanced/Metastatic Solid Tumors (NCT06264921)	phase 1/2 study of NKT3447 for advanced/metastatic cancers that have progressed on standard therapy including ovarian cancer, endometrial cancer, gastric cancer, gastroesophageal junction cancer, or other solid tumors with CCNE1 amplification; small cell lung cancer; triple negative breast cancer; HR+/HER2− breast cancer. Monotherapy of NKT3447 is being tested. Estimated participants: 90	no results available
A Study to Examine the Safety of Different Doses of BG-68501 Given to Participants With Advanced-Stage Tumors (NCT06257264)	phase 1/2 study of BG-68501 for advanced/metastatic HR+/HER2− breast cancer that has progressed on standard therapy. The study assesses BG-68501 as monotherapy and in combination with fulvestrant and the CDK4-selective inhibitor BG-43395. Estimated participants: 138	ASCO 2025d N = 41 patients reported treatment-related adverse events include nausea, vomiting, and fatigue of 24 metastatic breast cancer patients, partial response was observed in 1 patient (4.2%), and 10 patients (41.7%) had stable disease
Study to Evaluate the Safety, and Tolerability of AZD8421 Alone or in Combination in Participants With Selected Advanced or Metastatic Solid Tumors (NCT06188520)	phase 1/2 study of AZD8421 for advanced/metastatic cancers that have progressed on standard therapy including ovarian cancer and HR+/HER2− breast cancer. Monotherapy of AZD8421 as well as combinations with endocrine therapy (camizestrant) and CDK4/6 inhibitors (abemaciclib, ribociclib, or palbociclib) are being tested. Estimated participants: 204	no results available
Study of ECI830 Single Agent or in Combination in Patients With Advanced HR+/HER2− Breast Cancer and Other Advanced Solid Tumors (NCT06726148)	phase 1/2 study of ECI830 for advanced/metastatic cancers that have progressed on standard therapy including HR+/HER2− breast cancer and solid tumors harboring CCNE1 amplification. Monotherapy of ECI830 as well as combinations with ribociclib and fulvestrant are being tested. Estimated participants: 280	no results available
Study of INCB123667 in Participants With Platinum-Resistant Ovarian Cancer With Cyclin E1 Overexpression (NCT07023627)	phase 1/2 study for platinum-resistant ovarian cancer with CCNE1 overexpression. Monotherapy of INCB123667 is being tested. Estimated participants: 160	ASCO 2025e N = 90 patients reported treatment-related grade ≥3 adverse events included intestinal obstruction (8.9%), anemia (6.7%), neutropenia (5.6%), and thrombocytopenia (5.6%) among 90 patients, 4.4% had complete response, 10.0% had partial response, and 51.2% achieved stable disease
Study PF-07220060 in Combination With PF-07104091 in Participants With Breast Cancer and Solid Tumors (NCT05262400)	phase 1/2 study for advanced/metastatic cancers that have progressed on standard including HER2−, HR+/HER2+, and HR +/HER2− breast cancer. Combination of PF-07220060 and PF-07104091 with endocrine therapy (letrozole or fulvestrant). Estimated participants: 192	ESMO 2024f N = 33 patients reported treatment-related grade ≥3 adverse events included nausea (3.0%), neutropenia (33.3%), and diarrhea (3.0%) of 18 metastatic breast cancer patients, partial response was observed in 5 patients (27.8%), and 5 patient (27.8%) had stable disease. For metastatic breast cancer patients (n = 26) median progression-free survival was 8.3 months

Summary of active clinical trials with CDK2-selective inhibitors on clinicaltrials.gov as of July 2025. Descriptions of the clinical studies are adapted from clinicaltrials.gov. Reported clinical results are summarized from the published abstracts.

^aINX-315, SABCS, 2024: https://doi.org/10.1158/1557-3265.SABCS24-P4-10-16

^bPF-07104091, ASCO 2023: https://doi.org/10.1200/JCO.2023.41.16_suppl.3010

^cBLU-222, ASCO 2024: https://doi.org/10.1200/JCO.2024.42.16_suppl.1056

^dBG-68501, ASCO 2025: https://doi.org/10.1200/JCO.2025.43.16_suppl.3115

^eINCB123667, ASCO 2025: https://doi.org/10.1200/JCO.2025.43.16_suppl.5514

^fPF-07220060 with PF-07104091, ESMO 2024: https://doi.org/10.1016/j.annonc.2024.08.685

A key complementary biomarker that emerged from preclinical studies and assessment of DepMap data is the CDK4/6 inhibitor p16INK4A, encoded by CDKN2A.^{33,50,52,65,66} Functionally, p16INK4A serves as an endogenous CDK4/6 inhibitor.^74,75 Thus, high expression of p16INK4A resulting in suppression of CDK4/6 activity would enforce greater or possibly even sole dependence on CDK2 for the G1/S transition. Importantly, this would create a tumor-selective vulnerability to CDK2 inhibition (Figure 2C). While the deregulation of CDKN2A/p16INK4A is generally associated with senescence and profound suppression of cell growth, p16INK4A overexpression is also a feature of tumors lacking function of the RB-tumor suppressor.^75,76 p16INK4A overexpression is found in cancers harboring the selective mutation of RB as well as viral oncoproteins. High levels of p16INK4A are also observed in tumors driven by cyclin E, and there is generally a positive correlation between the expression of CCNE1 and CDKN2A in The Cancer Genome Atlas (TCGA) or other tumor datasets. Importantly, p16INK4A is a commonly used diagnostic stain, which could be rapidly implemented clinically. Furthermore, using multiplex immunofluorescence, it is possible to illustrate that tumors with cyclin E^high, p16INK4A^high, and RB⁺ biomarker configurations can be readily detected in ovarian and triple-negative breast cancers^50,65,66 in addition to preclinical models. Presumably, such tumor cells would be particularly sensitive to CDK2 inhibitors. In addition, high levels of p16INK4A are associated with resistance to CDK4/6 inhibitors,^77,78 which is likely due to CDK2 being the key driver of proliferation in this context.

The role of RB as a biomarker for response to CDK2 inhibitors is complex. In DepMap, RB1 mutation/deletion is associated with increased sensitivity to CDK2 deletion. This finding is complicated because most RB-deficient cells express high levels of p16INK4A, thereby creating greater dependence on CDK2 activity. However, in the context of tumors with high cyclin E and p16INK4A, RB is functionally important for a G1 arrest. Consequently, the depletion of RB is observed as a driver of resistance in unbiased CRISPR screens,⁵² and in functional studies, RB modulation alters the IC₅₀ to CDK2 inhibitors.^50,52 Such a finding is reminiscent of the response to CDK4/6 inhibitors, wherein RB1 loss enables the expression of E2F target genes and S-phase progression.^50,65 In contrast, RB1 loss does not overcome the G2-arrest phenotypes observed with CDK2 catalytic inhibitors.⁵⁰ Similarly, CDK2 inhibition can have robust activity in models of small cell lung cancer that have lost RB1 and express high levels of p16INK4A.⁶² Importantly, when CDK2 inhibitors arrest cells in G2, RB is hyperphosphorylated and E2F target genes are activated. These findings have been used to support the use of CDK2 inhibitors in clinical trials to target RB-deficient tumors, as the role of RB in that context is secondary. Overall, these data support that cyclin E and p16INK4A are dominant drivers of sensitivity, while RB serves to tune that response.

Combinatorial strategies

One of the features of CDK4/6 inhibitors is their broad ability to combine with other agents, most notably endocrine therapies and specific pathway inhibitors, which have spawned multiple clinical trials.^26,41,79 Combination therapies involving CDK2 inhibitors are emerging as promising strategies to enhance anti-tumor efficacy, particularly in models with elevated CCNE1 expression in CDK4/6 inhibitor-resistant HR+/HER2− breast cancer. Preclinical studies have shown that co-treatment with the new selective CDK4 inhibitor atirmociclib and the CDK2 inhibitor PF-07104091 induces synergistic tumor growth inhibition and profound tumor regression in HR+/HER2− breast cancer xenograft models, including those resistant to palbociclib.⁸⁰ This combination achieved superior efficacy compared to palbociclib treatment and was mechanistically linked to complete suppression of RB phosphorylation, depletion of FOXM1, and reduced expression of E2F targets such as cyclin A2.⁸⁰ Similar benefits were observed in CCNE1-high, RB1 wild-type models with low p16INK4A expression following treatment with the CDK2 inhibitor BLU-222 and the CDK4/6 inhibitor ribociclib. Notably, p16INK4A levels were essential for the synergistic activity of CDK2 and CDK4/6 inhibitors, as high p16INK4A expression diminished the benefit of adding CDK4/6 inhibition.⁵²

Several studies support the role of dual CDK2 and CDK4/6 inhibition in delaying resistance and enforcing sustained cell-cycle exit in HR+/HER2− breast cancer.^67,81,82 Moreover, the addition of estrogen receptor (ER) inhibition to this combination further enhances anti-proliferative effects, effectively suppressing growth in cell lines resistant to CDK4/6 inhibitors.^83,84 Notably, in p53-deficient cells, co-inhibition of CDK2 and CDK4/6 activates the DREAM complex via suppression of p130 phosphorylation, resulting in durable growth arrest.⁸² Further, the CDK2 inhibitor INX-315 combined with palbociclib not only delayed resistance but also promoted therapy-induced senescence and deeper suppression of E2F targets in breast cancer models.⁸¹ The efficacy of this dual inhibition has been demonstrated beyond breast cancer in pancreatic cancer xenografts.⁶⁵ While CDK2 inhibitors synergize effectively with CDK4/6 inhibitors, antagonism was observed in combinations with WEE1 or CHK1 inhibitors, suggesting that CDK2 inhibition may counteract the mitotic entry that these agents are designed to promote.⁵⁰

Beyond combinations with other selective CDK inhibitors, CDK2 inhibitors have shown potential when combined with chemotherapy and DNA damage response inhibitors. For example, BLU-222 acts as a chemosensitizer to carboplatin and paclitaxel in CCNE1-amplified tumors; however, this synergistic effect appears to extend beyond CCNE1 overexpression, as CCNE1, RB1, and CDKN2A expression levels did not predict response.⁵² In CCNE1-amplified ovarian cancer, CDK2 activity was critical for efficient homologous recombination, as it facilitated the repair of collapsed replication forks. Consequently, inhibition of CDK2 sensitizes these cells to the PARP inhibitor olaparib, resulting in enhanced anti-tumor efficacy compared to either agent alone.⁸⁵ Interestingly, CDK2 inhibition can induce features of chromosome segregation dysfunction that could further contribute to sensitivity to other forms of chemotherapy.⁸⁶ Together, these findings underscore the therapeutic promise of CDK2 inhibitor-based combination treatments in targeting cell-cycle vulnerabilities and enhancing responses across a spectrum of oncogenic contexts.

Active clinical trials and preliminary outcomes

Several drugs that have the capacity to inhibit CDK2 have been interrogated clinically. Older clinical studies with flavopiridol, roscovitine, AT-7519, and dinaciclib were conducted in the absence of a biomarker strategy and generally had modest efficacy relative to their toxicity profiles.^87–89 These agents inhibited multiple CDK activities, including those that regulated features of transcription (e.g., CDK9) as well as CDK1. As such, most of these agents have the capacity to potently inhibit proliferation and/or kill tumor cells in preclinical models. Fadraciclib (Cyc065), which has more specificity for CDK2 and CDK9, remains in clinical development but generally appears distinct from current CDK2-selective inhibitors due to its effect on anti-apoptotic pathways and other targets through CDK9 inhibition.^90–92

As discussed above, one mechanism to enhance the durability of CDK4/6 inhibitors would be to utilize molecules that coordinately inhibit CDK2, as exemplified by the CDK2/4/6 inhibitor PF-06873600. Other drugs in this class that entered the clinic include NUV-422. The clinical data on PF-06873600 that were recently published indicated substantial toxicity with relatively modest clinical benefit,⁹³ which likely led to the discontinuation of further clinical development at present. Similarly, the agent NUV-442 was recently discontinued. Whether these limitations are pharmacophore selective, for example these specific molecules still lack sufficient CDK2 selectivity over CDK1, or a general feature of trying to dually target CDK4/6 and CDK2 remains a critical unknown.

Multiple CDK2-selective inhibitors are being tested in clinical trials (summarized in Table 1). In general, these trials have similar approaches for testing drug efficacy. Initial dose escalation studies are performed to define a recommended dose for subsequent study. The subsequent study arms include one focused on patients with CCNE1 amplification, typically as a monotherapy, which is consistent with the findings from the preclinical evaluation of these agents. Another arm is often in the context of HR +/HER2− breast cancers that have progressed on CDK4/6 inhibitor-based therapy. Since HR+/HER2− breast cancer is invariably treated with endocrine therapy, these expansion cohorts generally include fulvestrant for second-line treatment as a combination. With the supposition that CDK2 inhibitors will ameliorate resistance to CDK4/6 inhibitors, some of these studies also include the triplet of endocrine therapy with an approved CDK4/6 inhibitor in addition to the CDK2 inhibitor being tested. Lastly, as discussed above, there are studies of CDK2 inhibitors in small cell lung cancer, which is often characterized by loss of RB. CCNE1 amplification is also observed in small cell lung cancer and is linked to poor prognosis and chemotherapy resistance,⁹⁴ supporting the rationale for CDK2 targeting in this context.

The reported clinical data with CDK2-selective inhibitors to this point are relatively limited and solely present in abstract form as summarized in Table 1. While it is hard to extrapolate from the relatively small numbers of patients, the CDK2 catalytic inhibitors seem to be well tolerated. Principle toxicities shared across all of the reported drugs are gastrointestinal (e.g., nausea and diarrhea) or hematological in nature (e.g., neutropenia and anemia) (Table 1). However, select toxicities have also been reported. For example, BLU-222 exposure was associated with photophobia and INCB123667 with intestinal obstruction, although these data are based on very small numbers of events and could be at least partially related to underlying causes. Clinical responses have been provided for single agent treatment with INCB123667, PF-07104091, and INX-315. In all cases, there have been partial responses and disease stabilization, and INCB123667 achieved complete responses in a subset of platinum-resistant ovarian cancer with CCNE1 overexpression. The one study reporting combinatorial data is for the CDK2 inhibitor PF-07104091 in combination with the CDK4-selective inhibitor PF-07220060 and endocrine therapy in patients with metastatic breast cancer that have progressed on prior therapy (Table 1). This combination appears to have manageable toxicity and yielded partial responses and stable disease with a median progression-free survival of 8.3 months. Thus, while still very early and with limited clinical outcomes, the existing data support continued evaluation of CDK2 inhibition as a therapeutic strategy.

Concluding remarks

Over the last several years, there has been immense progress in targeting CDK2 for cancer therapy. Pharmaceutical strategies have emerged very rapidly that include conventional catalytic inhibitors as well as mechanisms of inhibiting or depleting CDK2. Preclinical studies have illustrated complex mechanisms through which CDK2 inhibitors act in multiple phases of the cell cycle. Biomarkers for exceptional responses to CDK2 inhibitors have been described and are partially incorporated into current clinical design. Early stage clinical trials are ongoing that should address monotherapy and combinatorial efficacy of CDK2 inhibition in multiple indications. Despite this progress, key questions remain.

Will monotherapy with CDK2 inhibitors be effective?

While CDK2 inhibition can have robust monotherapy in CCNE1-amplified and p16INK4A-high tumors, the durability of this response and acquisition of resistance is scantly understood at this point. While CDK2 inhibition can limit the growth of many tumor models that are arresting in G2, it remains unknown whether this response represents a significant therapeutic strategy.

Are there more determinants of response to CDK2 inhibitors?

While it appears CCNE1 and p16INK4A are key determinants of response to CDK2 inhibitors, the questions still remain of whether RB-deficient tumors will be responsive or whether other biomarkers may be crucial in pinpointing those cancers most sensitive to CDK2 inhibition.

Will CDK2 + CDK4/6 inhibition combination be effective or hampered by toxicity?

The development of highly selective CDK2 inhibitors has reduced toxicity compared to earlier compounds, enabling their clinical evaluation in combination with CDK4/6- or CDK4-selective inhibitors. It remains to be determined how tolerable these combinations are and how efficacious this approach will be, particularly in the context of patients who have already progressed on a CDK4/6 inhibitor-based therapy for metastatic HR+/HER2− breast cancer.

Are there ways to achieve better CDK2 selectivity over CDK1?

It may be that the shortcomings of recent clinical trials have been due to CDK1 toxicity, which results from insufficient CDK2 selectivity of the current type 1 ATP competitive inhibitors. Additional publicly available structural information may inspire more medicinal chemistry campaigns. More exploration of the selectivity, efficacy, and biological effects of allosteric inhibitors and degraders is needed.

Other combination therapies?

Currently, the clinical evaluation of CDK2 inhibitor-based combination therapies, particularly with CDK4/6- or CDK4-selective inhibitors, remains limited. However, the diverse molecular mechanisms of CDK2 inhibition across tumor types and genetic backgrounds provide opportunities for additional combinations. For instance, in RB-deficient tumors in which the G1/S checkpoint is compromised, targeting cyclin A/CDK2 activity may synergize with agents that impair DNA replication or repair, such as PARP inhibitors or chemotherapy. Moreover, CDK2 inhibition may enhance the efficacy of macrocyclic cyclin A/B inhibitors, which drive E2F hyperactivation and apoptosis in small cell lung cancer.

As these questions and others are addressed in preclinical models and clinical trials, it is likely that CDK2-selective inhibitors will join the growing armamentarium of targeted cancer drugs.