Identification of Novel Protein Biomarkers and Targets to Overcome PARP Inhibitor Resistance

Poly(ADP-ribose) polymerase (PARP) inhibitors (PARPis) can be used to treat ovarian cancer at nearly all phases of a patient's cancer journey. Although the clinical impact of agents like olaparib, niraparib, and rucaparib have changed clinical practice over the last several years, there is still a need to optimize the way we use PARPis to maximize their therapeutic potential. Relying on current biomarkers to predict response to single-agent PARPi is not good enough. For instance, only about 20% of patients will have a somatic (in the tumor) or germline (in the blood) mutation in BRCA1 or BRCA2. While the benefits of a PARPi in such tumors are clear, the majority of patients will have a normal BRCA gene. For these patients, tests have also been developed to predict whether a tumor will have 'BRCA-ness,' meaning it lacks a BRCA mutation but 'behaves' like a tumor with a BRCA mutation. This phenomenon is reflected in a commercially available test called the homologous recombination deficiency (HRD) score. Although the HRD score can enrich for PARPi responses, it does not accurately predict which patients will respond. Importantly, the field of ovarian cancer PARPi development has moved beyond single-agent PARPi studies and is actively looking for combination partners with the greatest potential to improve the efficacy of PARPis. However, none of the aforementioned tests (BRCA mutations or HRD score) can tell doctors whether a second drug could be partnered with the PARPi to make the PARPi work more effectively.

This proposal addresses two ways to improve the impact of PARPis in the clinic: (1) to discover new biomarkers to predict response and (2) to identify new combinations that work synergistically with existing PARPis. Regarding the latter, there is substantial effort to accomplish this in the laboratory and in clinical trials. Although clinical trials have investigated new PARPi combination partners (like blood vessel blockers cediranib or bevacizumab) and inhibitors of cancer cell pro-survival pathways (such as alpelisib), a comprehensive approach is needed to discover other targets that could improve the efficacy of PARPis. In particular, there is tremendous need to improve the efficacy of PARPis for patients who would otherwise be resistant to this form of therapy. Accordingly, this proposal will focus on ovarian tumors with inherent resistance to a PARPi. The ultimate impact of the proposed research for patients with ovarian cancer could be improved regression of disease while on a PARPi and/or a longer duration of response.

To accomplish these goals, we have proposed specific aims that will take a comprehensive look at the protein changes that are occurring inside cancer cells to determine which proteins are most likely to be contributing to resistance. The unique approach in this proposal is to use patient tumors that have been transplanted into mice. So-call patient derived xenograft (PDX) models are thought to mimic the behavior of patient tumors more closely than traditional cell lines. Accordingly, they will be used as patient surrogates to build the foundation for all three aims of this proposal. First, the latest proteomics technology will be used to fully characterize the levels of total proteins as well as activated/inactivated proteins (e.g., phosphoproteins) in PARPi sensitive vs. resistant PDX tumors. It is noteworthy that the research team has developed the world's largest single institution collection of ovarian cancer PDX models and a decade of experience using these models for drug development. This first aim will reveal which proteins are significantly different between sensitive and resistant tumors and will provide the initial glimpse into the differences that may drive resistance in some tumors. In addition, it is conceivable that important resistance proteins are at low basal levels and only rise as a direct result of PARPi exposure. In this case, baseline protein levels by itself may not be sufficient. For this reason, we also intend to look at the protein levels before and after PARPi exposure in PDX models.

Second, our strong focus on protein changes will not be performed in isolation, but rather will be combined with other molecular data. This is important because every function of a cell is dependent on the transcription of the genetic code from DNA to RNA, followed by the translation of the RNA code to proteins. While proteins are the macromolecules that impart function within a cell, the DNA and RNA molecules have direct bearing on the protein levels and function. Moreover, mature proteins can be activated or inactivated by other proteins, such as kinases, by adding phosphoryl groups on the mature proteins. Taken together, the most comprehensive examination of PARPi-resistant tumors would include a careful integration of the proteome (total protein and phosphoprotein), genome (mutations in DNA repair genes), and transcriptome (RNA production of genes). This aim will result in a filtered list of proteins that are most the likely candidates for contributing to resistance.

Third, the findings from the first and second aims would be validated by testing new drugs and combinations with PARPis in primary patient tumors obtained from consenting patients who have recurrent cancer, which is presumed to be platinum-sensitive. Such patients would already be candidates for maintenance PARPi, so we would be able to validate a protein biomarker for PARPi response. Importantly, patient tumors would also be tested with the novel drug combinations as suggested by the previous two aims. To complete this aim, we will first need to narrow the list of potential drug candidates because biopsy specimens might provide only scant tumor tissue (depending on the size and location of the tumor). As such, we have developed methods to culture tumors in a way that retains the non-cancerous and cancerous tissue components grown in three dimensions (3D) as spheroids, which has advantages over traditional two-dimensional cultures. We have previously shown that this 3D methodology accurately reflects the clinical response seen in patients, and it will be used to screen PDX tumors in culture. Since we can also make these tumors grow in mice, we will further demonstrate the activity of a new drug combination by testing the best candidate combinations in PDX tumors in mice. Afterwards, we will be prepared to test the activity of novel combinations in precious patient samples, also in 3D. If we can show that the new PARPi combinations are synergistic in patient tumors, this would lay the groundwork for a future clinical trial in patients with inherent resistance to PARPis. Since we will also follow the clinical progress of each patient, we will know which patients are PARPi-resistant, as evident by a short duration of response to a PARPi in the clinic. We can take this information and compare it to the protein signature results from the first and second aims. This will help validate our signature of PARPi-resistance.