Zhejiang University Review Maps Where CDK Drugging Goes After the CDK4/6 Inhibitors

Three PROTACs and a glue now in patients

NKT3964 is enrolling advanced and metastatic solid tumors with CCNE1 amplification and degrades CDK2. NKT5097 removes both CDK2 and CDK4 in a Phase 1 study weighted toward breast cancer and CCNE1-amplified disease. BTX-9341 degrades CDK4 and CDK6 and is in Phase 1 in hormone receptor-positive, HER2-negative advanced breast cancer. Beside them the authors put CT7439, a molecule that inhibits CDK12 and CDK13 and also glues cyclin K to the CRL4-DDB1 ligase, in Phase 1/2 testing in solid malignancies.

Compound 26 depleted CDK4 at a 26 nM DC50

Zheng and colleagues sort CDK degradation into PROTACs, molecular glues, HSP90-mediated targeting chimeras, hydrophobic tagging and autophagy-tethering compounds. Li and colleagues' compound 26, the prototype HEMTAC, depleted CDK4 with a DC50 of 26 nM and 88% maximum degradation and CDK6 at 19 nM and 92% after 24 hours in murine melanoma cells, and slowed growth in zebrafish and mouse xenografts. Hydrophobic tagging produced LPM3770277, an adamantyl CDK4/6 degrader, and an SNS-032-derived CDK9 series ending in AZ-9, which recruits the autophagy protein ATG101 in place of any ligase. The single ATTEC entry, degrader 10, drags CDK9 and cyclin T1 into the autophagosome through LC3B. On the glue side, HQ461, CR8, dCeMM2/3/4, NCT02 and SR-4835 all destroy cyclin K, and CDK2-directed glues plus the cyclin E1 glue MRT-55811 remain preclinical.

CRBN, DDB1 and UBE2G1 loss carries cereblon-side resistance

A functional-genomics screen by Shirasaki and colleagues, cited at length here, found resistance to cereblon- and VHL-based degraders converging on disruption of the ligase complex itself. Loss of function in CRBN, DDB1, UBE2G1 and COP9 signalosome genes (COPS2, COPS3, COPS7B, COPS8) accounted for the cereblon side; VHL, CUL2, RBX1, ELOB, ELOC and UBE2R2 for the other. Glue resistance instead clusters as hotspot mutations in the substrate receptors of the hijacked ligase. Separately, the efflux pumps ABCB1 and ABCC1 carry both intrinsic and acquired PROTAC resistance. The authors draw one testing-relevant line from that evidence: "E3 ligase expression and the functionality of the E3 ligase complex serve as straightforward biomarkers for degrader efficacy." Their proposed counters are BH3 mimetics, mTOR inhibitors, efflux inhibitors, or a switch to a degrader that recruits a different ligase or uses the lysosome.

Patient selection is the thinnest part

Where the review does address who gets treated, it works almost entirely through cyclin E. Selective CDK2 inhibitors, the authors write, "have shown early antitumor activity in cases with cyclin E overexpression or CCNE1/2 amplification," and seven appear in the trial table: INCB123667, tagtociclib, AVZO-021, ECI830, AZD8421, BG-68501 and INX-315, several enrolling on CCNE1 amplification or on hormone receptor-positive, HER2-negative status. Acquired resistance to CDK4/6 inhibitors, by their account, converges on CDK2 activation driven by CCNE1/2 amplification or cyclin E upregulation, with induced binding between CDK6 and p18INK4C as a second mechanism. RB1 surfaces only in a genomic tally, that nearly half of patients carry loss-of-function lesions in TP53, RB1, CDKN2A or CDKN2B. No immunohistochemistry cut-off, specimen standard or companion diagnostic is proposed anywhere in the article.

Kinase relocation as a separate idea

CDK-TCIP1, also called BAK-04-021, keeps the kinase alive and moves it. It links SNS-032 to BI3812, a ligand for the BTB domain of the transcriptional repressor BCL6, so CDK9 arrives at BCL6-occupied promoters with its catalytic activity intact. Released there, it phosphorylates RNA polymerase II and switches silenced proapoptotic genes including PMAIP1, BBC3 and BIK back on, killing BCL6-dependent diffuse large B-cell lymphoma cells. CDK-TCIP3 (BAK-04-338) does the same for CDK12 and CDK13.

Why This Matters to the APO|APE Reader

BTX-9341's Phase 1 population, hormone receptor-positive and HER2-negative advanced breast cancer, is one of the most heavily genotyped groups the laboratory sees, and a CDK4/6 degrader entering it adds no new predictive assay to the order set. The new requests would come from the CDK2 agents and from the degrader class itself: CCNE1 copy number, which sits on most large panels but is rarely reported for treatment selection, and cereblon or VHL pathway integrity, which sits on none of them. Zheng and co-authors advance E3 ligase status as a degrader biomarker with no validated assay behind it, and nothing in their May 2026 trial table requires one for enrollment.