Cyclin-Dependent Kinase 12, Immunity, and Prostate Cancer

The advent of immune checkpoint blockade therapies that use programmed death 1 (PD-1) or programmed death ligand 1 (PD-L1) inhibitors for the treatment of multiple cancer types represents a major step forward in the care of patients with cancer, and the discovery that certain genetic subtypes of cancers (particularly those with mutations in DNA mismatch-repair genes) may be remarkably sensitive to PD-1 inhibitor therapies has moved us closer to using tumor genomic status to inform our interventions (an example of precision oncology). A related discovery is that cancers that have mutations in genes that repair DNA damage using homologous recombination, especially cancers of the breast and ovary, may respond well to treatment with poly(adenosine diphosphate [ADP]–ribose) polymerase (PARP) inhibitors or platinum chemotherapies. In the context of advanced prostate cancer, both of these genomic classifications are clinically relevant. Approximately 2 to 5% of metastatic prostatic cancers have mismatch-repair deficiency and 20 to 30% have defects in homologous recombination.
Wu et al.1 recently described a new molecular subclass of advanced prostate cancers: those defined by biallelic somatic loss-of-function mutations of the tumor-suppressor gene CDK12, which encodes cyclin-dependent kinase 12. This enzyme was previously thought to maintain DNA repair through the regulation of DNA damage response genes (BRCA1, FANCD2, and ATR), and it had been suggested that the enzyme was associated with PARP inhibitor sensitivity when it was genetically inactivated.2 Wu et al. found biallelic inactivation of CDK12 in 25 of 360 biopsy samples (6.9%) obtained from patients with metastatic castration-resistant prostate cancer and in only 6 of 498 samples (1.2%) obtained from patients with primary prostate cancers. Notably, all the inactivating CDK12 variants were somatic alterations; no pathogenic germline mutations were identified in any patient, which suggests a vital role for CDK12 in embryogenesis.

Wu et al. also discovered a unique role for CDK12 in maintaining genomic stability. The genomes of prostate-cancer tumors with CDK12 inactivation had a novel pattern of widespread focal tandem duplication (i.e., copy-number gains) of stretches of DNA exceeding 8 kb that were dispersed throughout the noncoding and coding portions of the genome, with enrichment in gene-dense regions. Unlike cancers with genomic instability associated with deficiencies in DNA repair by means of homologous recombination, cancers that have biallelic loss-of-function variants in CDK12 very rarely had large (chromosomal or arm-level) amplifications or deletions. Furthermore, CDK12 mutations almost never co-occurred with pathogenic variants in genes encoding mismatch-repair proteins, in genes encoding homologous-recombination repair proteins (other than CDK12), or in other genetic subtypes of prostate cancer (e.g., those with ETS-family gene fusions and SPOP mutations). The focal tandem duplications that were observed in CDK12-variant tumors were characterized by a bimodal distribution in the length of the duplicated segment — either approximately 2.5 Mb or 0.5 Mb, the same size as DNA replication domains — a striking finding that was consistent with their occurring as a result of aberrant DNA re-replication during the S phase of the cell cycle. The authors designated these unique genomic rearrangements as CDK12-associated focal tandem duplications.

Next, in a series of elegant experiments and bioinformatics approaches, Wu et al. found that CDK12-variant prostate cancers were enriched for gene fusions (resulting from focal tandem duplications that occur within coding regions) and that their gene-fusion burden was at least three times that of homologous repair-deficient or ATM-variant prostate cancers. The authors reasoned that such gene fusions would induce neoantigens by creating chimeric open reading frames that generate fusion proteins. The neoantigen-prediction methods confirmed a higher level of fusion-induced neoantigens (FINAs) in CDK12-variant prostate cancers than in all the other molecular subclasses of prostate cancer. When Wu et al. examined the total neoantigen burden (encompassing all forms of neoantigen generation), only prostate cancers that were deficient in mismatch repair surpassed CDK12-variant cancers in their neoantigen load, by virtue of their huge numbers of single nucleotide variant–induced neoantigens.

Concordant with this finding, CDK12-variant tumors had higher overall levels of T-cell infiltration and larger numbers of expanded T-cell clones than all other genomic subtypes of prostate cancer (except those deficient in mismatch repair) and also had increased expression levels of certain chemokines and their receptors. Finally, preliminary data from four patients with CDK12-inactivated advanced prostate cancer who were treated with PD-1 inhibitors suggested that two of these patients had robust declines in the prostate-specific antigen level, shrinkage of metastatic deposits induced by the immune checkpoint blockade therapy, or both.

The clinical implications of the study conducted by Wu et al. are profound and immediate. Recent clinical trials have shown that only approximately 5 to 15% of patients with advanced prostate cancer have a favorable response to PD-1 inhibitor treatment,3,4 unlike patients with other cancer types, among whom response rates are higher. However, uncovering the genomic determinants of PD-1 inhibitor sensitivity in patients with prostate cancer has remained elusive. Although prostate cancers that are deficient in mismatch repair may represent a subclass that is more responsive to immune checkpoint inhibition, they represent only 2 to 5% of all castration-resistant prostate cancers, which means that most of the favorable responses are still unexplained. If the tumor responses that were observed in the present study are replicated in prospective clinical trials, CDK12-variant prostate cancer may become the second genomically defined tumor subtype that may benefit from anti–PD-1 therapy.

Furthermore, CDK12 loss-of-function mutations are not restricted to prostate cancer. A recent large-scale genomic study of more than 10,000 cancer genomes encompassing different cancers has shown CDK12 alterations in many tumor types, including gastrointestinal, bladder, uterine, and ovarian cancers.5 Because CDK12 mediates DNA repair by means of homologous recombination in addition to replication-associated repair, combination therapy comprising a PD-1 inhibitor and a PARP inhibitor may also be an effective approach in patients with CDK12-deficient cancers. Finally, in cancers with wild-type CDK12 status, one wonders whether treatment with a CDK12 inhibitor may induce a sensitivity to immune checkpoint blockade therapy, generating a new form of synthetic lethality. Designing the clinical trials to answer these questions will be our challenge for the next several years.

Emmanuel S Antonarakis, From the Johns Hopkins University Sidney Kimmel Comprehensive Cancer Center, Baltimore.

1Wu YM, Cieślik M, Lonigro RJ, et al. Inactivation of CDK12 delineates a distinct immunogenic class of advanced prostate cancer. Cell 2018;173(7):1770-1782.e14.

The New England journal of medicine. 2018 Sep 13 [Epub]
Source: Antonarakis, Emmanuel S. 2018. "Cyclin-Dependent Kinase 12, Immunity, and Prostate Cancer". New England Journal of Medicine 379 (11): 1087-1089. Massachusetts Medical Society. doi:10.1056/nejmcibr1808772.