This is an important area for future research because we have not yet maximized precision medicine in this disease, and genomic evaluations will continue to evolve. At this point, we have a good characterization of the DNA alterations that are associated with prostate cancer. However, there are other genomic predictors at the RNA level that may evolve in the future to help us identify new strategies for treating this disease. It is quite possible that biology such as RNA splicing is an important aspect of prostate cancer heterogeneity. We already have the example of alternative androgen receptor splicing, which is a mechanism of interest in the expression of resistance to androgen receptor–targeted strategies.

Genetic changes tend to occur early on, and I think that there will be an opportunity to use this information more than we currently do in early prostate cancer. In the future, we may be able to better harness the knowledge of distinct molecular subsets of prostate cancer. Even if, at this point, these alterations and subsets are not predictive, they can still be helpful as prognostic factors to determine concomitant or layered treatment approaches. Perhaps more importantly, some of these alterations might be discovered to be targetable as this area evolves.

In the CRPC disease space, understanding biologic heterogeneity is extremely important. Liquid biopsies will become very useful because they can be performed in series over time to ascertain how these patterns are evolving. My hope is that this type of profiling becomes as routine as prostate-specific antigen monitoring. Hopefully, the costs will be modest enough to allow us to do that, as it will provide insights into how the disease is responding to various therapies, which may allow us to better layer or adapt our therapies to the changing biology.

We are currently conducting much more germline testing in men with localized disease than we did in the past. We are seeing a remarkable uptake in the use of germline testing in men with clinically localized disease, high-risk features, and/or a strong family history of early onset lethal disease. And this may be important down the line. Findings do not necessarily change our approach to treatment right now, but testing identifies mutations, and this might be important information if the patients require novel forms of systemic therapy.

We have plenty of prognostic markers to identify which patients will or will not do well, so what we truly need right now are predictive biomarkers to identify the optimal approach to the treatment of patients with mCRPC. 

RNA expression profiling is often conducted in men with clinical localized disease whom we are considering for active surveillance. But is it predictive? It is a prognostic marker for progression, as are findings from magnetic resonance imaging. So, we are quite interested in testing to determine predictive biomarkers. Current examples are luminal vs basal expression for men who might be candidates for hormonal therapy. The Post-Operative Radiation Therapy Outcomes Score for the use of adjuvant radiation therapy is also playing a role; it is not quite as well defined as it is in advanced disease, but it is increasingly applicable.

When we talk about genomic testing, the first thing to ask is: What are we testing? We have germline testing, which can be done with a blood draw, and then we have testing for somatic mutations, which can be done by sampling the tumor tissue itself and by circulating tumor DNA testing.

Germline testing is important for several reasons, one of which is that these mutations can be carried in the family. Special surveillance can be put in place for those family members who have the altered genes, so germline testing has implications beyond that of a predictive biomarker. Additionally, some genetic mutations, including BRCA1 and BRCA2, are predictive biomarkers in that we have poly ADP ribose polymerase (PARP) inhibitors, such as olaparib and rucaparib, to use in that setting. 

Sampling the tumor may not be as easy as the blood draw for germline testing, but it can be helpful for gaining additional insight. Many alterations that we observe in the tumor genome are actually present from the beginning, in the germline, but some patients have BRCA1 or BRCA2 within the genome of their tumor, even though it was not in the genome that they were born with. This is called a somatic mutation, as opposed to a germline mutation. Somatic mutations are common and expected in tumors. Analyzing a sample of the tumor can also assess DNA mismatch repair genes, such as MSH2 and MSH6, as well as other homologous recombination gene mutations, such as PALB2. Mutations in mismatch repair genes MSH2 and MSH6 are rare but predict good responses to pembrolizumab. With circulating tumor DNA, we can also test for BRCA1, BRCA2, and ATM, and we now have US Food and Drug Administration approvals for circulating tumor DNA testing to serve as a predictive biomarker for the use of PARP inhibitors. 

Lastly, prostate-specific membrane antigen (PSMA) is a predictive biomarker that could indicate that the patient will respond to precision radioligand therapies (ie, PSMA-targeted lutetium-177 therapies). PSMA is also being studied as a target for immunotherapy using bispecific T-cell engagers, which are infused into the patient and activate the immune system specifically toward PSMA-targeted cells. In this way, PSMA may serve as a predictive biomarker that is quite distinct from the genetic markers applicable to PARP inhibitors.