Using Liquid Biopsies to Predict Treatement Responses in Prostate Cancer
Liquid biopsies may be one way to help strengthen the understanding of prostate cancer biology, says Howard I. Scher, M.D.
“Over the past few years, we have learned that there are distinct molecular subtypes of prostate cancer for which certain drugs are available,” says Scher, chief of Genitourinary Oncology Service at Memorial Sloan Kettering Cancer Center. “Some are already approved and others are in development. The issue that we have learned is that an individual’s cancer changes biologically as it progresses and will also change differently, depending on what particular treatments have been given.”
Scher spoke on predicting sensitivity to prostate cancer treatment in a presentation during an interview with CURE. He explains the biology known thus far about multiple prostate cancer subtypes and how liquid biopsies will likely shape the future treatment landscape of the disease.
What are the current challenges with predicting response to treatment?
We know that androgen receptor (AR)–directed therapies, when given the first time, work very well. However, when you go from one to the other, the response rate is significantly lower. Most patients don’t respond well, but there is a subset for which it does work well. The question is, “How do we identify those patients ahead of time so they benefit?” How do we identify the patients unlikely to benefit, so they will be spared of the toxicity and cost of treatment?
If we look at the genetics of prostate cancers from patients who are first diagnosed with tumors localized to the prostate versus those with metastatic disease, and we look at the metastatic lesion in addition to the primary site — or we look at patients with metastatic castration-resistant prostate cancer (mCRPC) — we see that the frequency and types of changes are very different.
If we are focused on mCRPC, it is imperative that we study the cancer at the time we are making a decision to choose drug A versus drug B versus drug C. In many tumors, this is done via a biopsy of a metastatic lesion, but the most common metastatic spread is to the bone in prostate cancer. Here, the ability to do consistent molecular profiling is actually quite low. Our experience using directed biopsies, where we know exactly where the lesion is, is only about 50 percent. It’s an invasive procedure and is costly. If a patient has 10 individual lesions, they are not all biologically the same. Inadvertently, we may biopsy a lesion, identify a specific gene or pathway, and that pathway is not the key driver of the resistant cell population.
A number of us have been focusing on what has been categorized as the liquid biopsy. Obviously, taking a blood test is much easier than biopsying a metastatic lesion. There are a number of different tumor products that can be identified in blood. These include circulating tumor cells (CTCs), DNA, and RNA from tumors, and vesicles called exosomes that contain DNA, RNA, and protein, which help you characterize the disease.
Our focus has been primarily on CTCs and there are two techniques used. One is called a capture, or selection, method where you will use an antibody to capture cells that express the particular target of that antibody and then characterize in different ways. You can look at it on a slide, visually in a chamber, or deposit it on a slide.
The method that we are most recently using is called a nonselection method where you take a blood sample, deposit the cancer cells on a slide, perform various stains, and use software that can identify and localize each of the cancer cells present. This is the technology developed by Epic Sciences.
With this technology, we have been studying a protein made by the AR-V7 splice variant. The AR loses the portion of the protein that binds testosterone and the antiandrogen such as Xtandi (enzalutamide) or Casodex (bicalutamide), which inhibit its function. When the AR-V7 splice variant is present, it alone can drive the cancer without a ligand. The AR-V7 encodes for a protein that you can identify in cells. It can also do a pathological complete response reaction to identify the gene.
What we have been focusing on is looking at cells where AR-V7 is present and looking at patients who are first progressing on hormones, those who received one first-generation hormonal agent, and patients who had two of the life-prolonging agents. By drawing blood at each of these time points, we have seen the frequency of this splice variant is relatively low in the first-line setting. It gets higher with each course. Each time we identify it, the patients treated with AR-signaling directed therapy do not respond.
In contrast, there is no relationship between presence of AR-V7 and response to chemotherapy—in this case, docetaxel or Jevtana (cabazitaxel). What we have shown is that the survival of patients is improved in patients with AR-V7 present who receive a taxane, and it is inferior for those who receive AR-directed therapy.
It is very important to understand that the splice variant is in the nucleus where it binds to DNA and functions, in order to be very specific to the test results. If AR-V7 is present, then we know that these are patients who should not be treated with an AR-directed drug. This was one experience that we published earlier this year, now it is going through a validation process to confirm that this is true.
In this particular case, we are trying to predict and identify what biologic feature in this patient’s case will predict for response or nonresponse of a particular drug.
It sounds like there are a lot of developments taking place with regard to biomarkers.
When you think about it, you can actually look at a patient’s blood and you can see the cells that are very homogenous and look very similar. Those are tumors that tend to respond to a targeted agent—particularly those who are more advanced. It looks as if you are treating 20 cancers at once.
But recently the focus is the heterogeneity—how chaotic does it look? The earlier in the course it is, the less chaotic it is. As it moves down the road in second- and third-line settings, it gets much more heterogeneous. That’s where you can make the argument that a cytotoxic or biologic agent may be preferable. Those are the types of questions being asked.
Liquid biopsies have been discussed in great detail this year. What role do you see them having in prostate cancer?
In lung cancer, there were two liquid biopsy tests approved as companion tests. These are tests for CTC DNA, which are looking for a specific mutation. In prostate cancer, the numbers of mutations that clearly predict for sensitivity are fewer. It’s harder to identify the types of changes that are predictive using CTC DNA, even though the technology is improving very quickly.
Research has shown that they can see if there is overexpression or amplification of the AR, which is turning out to be a predictor of nonresponse. Now, the assays have to get to the point of establishing analytical validity, which means they are done at a performance level that’s consistent anywhere the assay is done.
In many cases, that’s almost harder to do to get to that level of performance than it is to develop the drug. The whole field of companion diagnostics is really where the field is headed.
Ideally, you would like to be able to do a blood test and say, “This is the drug for you.” You want to be able to monitor the change over time in a tumor to understand when a drug should be added to what patients are already getting versus changing therapy completely.
What do you believe the field can accomplish in the next five years?
I would like to be part of the accomplishment. In his Cancer Moonshot program, Vice President Joe Biden is looking for 10 years of progress in five years. I am totally optimistic that it can be achieved.
We recently met in Washington, DC, as a group — the Blood Profiling Atlas — with the enthusiasm of the assay developers and regulatory agencies. The spirit of collaboration is just incredible. There is importance to sharing information so things can go forward.