PARP Inhibitors: The Journey From Research Hypothesis to Clinical Approval

Kishan AT Naipal; Dik C van Gent


Personalized Medicine. 2015;12(2):139-154. 

In This Article

Predictive Biomarkers for PARP Inhibitor Sensitivity

Patient selection is very important to facilitate clinical PARP inhibitor trials but in the future also to select patients eligible for PARP inhibitor treatment. The synthetic lethal approach of PARP inhibitors is known to work effectively in HR-deficient tumors. Currently, germ line BRCA status is the major (only) biomarker for PARP inhibitor treatment in clinical setting. However, somatic BRCA mutations and BRCA1 promoter methylation also account for a sizable fraction of BRCA gene deficiencies in cancers that could be targeted with PARP inhibitors. Assays are being optimized to identify somatic BRCA mutations in cancers.[53,54] These assays should be clinically implemented to increase the number of BRCA mutation associated tumors eligible for PARP inhibitors.

BRCA1 promoter methylation has also been proposed as a marker for HR deficiency, as functional methylation of the BRCA1 promoter leads to repression of BRCA1 mRNA expression.[55,56] Specific assays to detect BRCA1 promoter methylation in clinical cancer samples, including tumor biopsies, have been developed and validated.[57] However, validation of these assays in clinical trials has not yet been performed and no retrospective analysis on material of completed clinical trials has been published, although BRCA1 promoter methylation has been reported to cause increased sensitivity to PARP inhibitors and Platinum drugs.[58,59]

In the past decade the term BRCAness has been used many times to indicate a biological condition of tumors resembling BRCA1 or BRCA2 deficiency, which may therefore predict PARP inhibitor sensitivity of these tumor cells. This term was first mentioned by Turner et al. (2004)[56] and has been refined many times.[60] Most importantly, the original intention of therapeutically exploiting BRCAness in tumors (e.g., with PARP inhibitors) remains a promising perspective. However, up to this moment no standardized method is available to detect BRCAness, which includes, but is not limited to, the previously mentioned BRCA mutations and BRCA1 promoter hypermethylation.

BRCAness without any BRCA gene defect could be caused by other HR gene defects, which are known to cause PARP inhibitor sensitivity. Several different approaches have been taken to identify these genes as biomarkers for BRCAness. A straightforward approach is the identification of single-gene biomarkers for PARP inhibitor sensitivity. This is generally done in experiments on cell lines and is the most common approach at the moment. A number of genes that predict PARP inhibitor response has been found in this way, with sensitivities ranging from extreme responses (e.g., BRCA1/2 genes with a thousand fold sensitivity) to gene defects that result in a few fold sensitivity. Interesting HR associated single-gene biomarkers are: MRE11, NBS1, RAD50, RAD51C and PALref-2. However, DDR genes other than core HR genes are also biomarkers for increased cytotoxic response to PARP inhibitors, for example, ATM, CHEK2, ERCC1, CDK12 and γ-H2AX.[19,61–69] Other single-gene biomarkers not directly involved in DDR include PTEN, which shows defects in a large fraction of prostate cancers, and STAG2 a component of the cohesin complex.[70,71] This specific approach of identifying and validating single-gene biomarkers for PARP inhibitor sensitivity is ongoing and also high-throughput systems are being developed to systematically and effectively identify these markers.[64,72]

The various approaches to identify single-gene markers for PARP inhibitor treatment are very productive. In addition to identification of possible biomarkers for PARP inhibitor sensitivity this approach has also extended our knowledge and understanding of the role of DNA repair in PARP inhibitor sensitivity and other elements of the DDR. Even very subtle DDR pathway modulations, for example, via protein modifications and miRNAs, can alter sensitivity to PARP inhibitors.[73,74] However, one has to bear in mind that this research on biomarkers is mostly performed in cancer cell lines in a laboratory setting. It is very challenging to apply these single-gene biomarkers in a clinical setting, as the number of markers is rapidly increasing and the clinical relevance of identified markers has not been clarified in the majority of cases. Therefore, development of functional markers predicting HR activity or even directly measuring PARP inhibitor sensitivity would be powerful tools.[75] One such functional approach to detect HR activity is the formation of RAD51 foci in response to a DSB inducing treatment as a way to predict PARP inhibitor sensitivity.[37–38,61,76–77] The ability to form RAD51 foci can be tested in fresh tumor material as a predictive assay for HR capacity. This assay detects the defects, that could cause HR deficiency, upstream of RAD51 accumulation at DNA ends, including BRCA1 promoter methylation (Figure 3).[77,78] In this way, testing of a large set of single-gene biomarkers could be replaced by functional testing based on RAD51 focus formation. On the other hand, there are some important drawbacks for implementation of this assay in the clinic. First, a sizeable fraction of single-gene biomarkers, especially those not directly involved in HR, cannot be replaced by this assay (MRE11, CHEK2, PTEN, etc.) and thus parallel screening of these biomarkers will remain necessary.[61,79] Second, this assay demands DNA damage induction in fresh tumor material and a subsequent incubation period to allow time for focus formation, which is not feasible in most clinical settings at this moment. However, this assay may facilitate patient selection in a trial setting at specialized hospitals and the discovery of novel, clinically relevant, single-gene biomarkers, which can subsequently be used for patient selection in general.

Actual PARP-1 activity has been shown to be significantly higher in tumor cells that are more sensitive to PARP inhibitors. Interestingly, cells that are deficient for core HR proteins show higher PARP-1 activity than HR-proficient cells.[61,65] Thus, indications are that HR deficiency indirectly results in increased PARP-1 activity. Furthermore, in these studies, the decreased RAD51 focus formation capacity in HR-deficient cells correlated well with high PARP-1 activity levels suggesting that PARP-1 activity levels may be a functional biomarker for HR defects. However, the mechanism of increased PARP-1 activity in HR-deficient cells remains to be elucidated. Also, experimental and clinical confirmation is needed to know whether this biomarker, just like RAD51 focus formation, is restricted to HR-deficient cells or whether it also accounts for HR-proficient cells that are sensitive to PARP inhibitors.

Novel strategies in biomarker discovery are determination of the genomic or expressional signatures of PARP inhibitor sensitive tumors. The rationale behind this approach is that one can define a specific profile of, for example, HR-deficient tumors. RNA profiling or gene expression arrays have been used for this purpose and show potential in identifying PARP inhibitor sensitive tumors.[80,81] Furthermore, a specific genomic 'scar' may be associated with BRCAness. This 'scar' includes a specific pattern of genome-wide short insertions and deletions frequently observed in BRCA1/2-deficient tumors. Strictly defining this scar and validating it for PARP inhibitor sensitive tumors could result in a robust biomarker for PARP inhibitor sensitive tumors. However, the costs of next generation whole genome sequencing (the only way to find the genomic 'scar') are considerable and more research on larger panels of tumors will be required to validate this approach for clinical use.