07 Dec Daniel S. Chen, Genentech: Interview Q&A

Q: Immunotherapy has rapidly evolved into a game-changing cancer treatment strategy for a wide array of different cancers, though some patients respond better than others.How important are biomarker assays for patient-selection to guide clinicians to use immunotherapies and for monitoring response and why?

A: We’re only beginning to scratch the surface of the possibilities in cancer immunotherapy, and biomarkers are going to play a critical role in realizing its full potential. At the same time, it’s important to understand that biomarker strategies for immunotherapy may be different than what many of us are familiar with today.

Right now, we’re using biomarkers to guide the use of medicines that target specific genetic abnormalities. Our ability to identify these abnormalities, like the overproduction of HER2 gene in a quarter of breast cancers, allows us to treat them as a distinct subtype of the disease. We can develop medicines and design clinical trials all focused on a specific genetic abnormality of the cancer itself. This type of personalized medicine allows us as oncologists and scientists to improve success rates of medicines in clinical trials, reduce uncertainty and ultimately improve cancer care.

But with cancer immunotherapy, the biomarkers are going to be different, and more complicated. We have to not only account for the features of a tumor but also the fact that everyone’s immune system will have a different reaction to the tumor. The exact same tumor could elicit a completely different response depending on the patient. We are going to have to be able to account for that as we develop the next generation of biomarkers.

I believe that eventually immunotherapies will become as personalized as targeted therapies, but the science is still young. The human immune system is one of the most complex defense mechanisms in the entire world, and there is a lot left to learn.This also means we’ll need to go beyond the “one biomarker, one drug approach” for cancer immunotherapy. The good news is that the science is moving quickly.

At Genentech, we are studying how to effectively use multiple biomarkers simultaneously, especially since one of the most promising paths for immunotherapy is the ability to combine them with many other types of medicine, ranging from chemotherapies to personalized targeted therapies.

Q: What are the current successful examples of applying biomarkers to patient selection and or optimizing immunotherapies?

A: The examples of biomarkers for immunotherapy are interesting because they illustrate how complicated and early the science is. Let’s take a look at three common cancer types – lung, colon and breast cancer.

In lung cancer, the level of a protein called PD-L1 is thought to be important to predict response to the anti-PD1 and anti-PDL1 class of medicines. In other words, PD-L1 “positive” tumors are thought to be more likely to respond favorably to these types of immunotherapies. But what we’ve seen so far is that depending on how stringently and at what level you set the threshold for PD-L1 positivity, and which type of diagnostic you use to test for PD-L1 expression, a clinical trial could be successful or not.If the level is set too low, you may not see an effect in a clinical trial; if you set it too high, you may be only treating a fraction of those who could respond.

For colorectal cancers, about 15% are genetically defined as microsatellite instability (MSI) high. This means that the DNA repair mechanisms in the tumor cells are faulty and as a result there are many different aberrations that make the tumor even more foreign to immune system, potentially making these good candidates for immunotherapy. Indeed, the data we’ve seen so far is that using MSI as a biomarker could help us identify people likely to respond to an anti-PD1 medicine. But a significant portion of the other 85% of people with this disease may also respond to an anti-PDL1 or anti-PD1 medicine. So how do we find those people? In colorectal cancer, a single biomarker may not be enough.

Another interesting example is in triple-negative breast cancer (TNBC). TNBC has some molecular characteristics that does not make it very noticeable to the immune system, which suggests that immunotherapies may not work as well in this type of disease. In addition, TNBC is typically aggressive and may advance before immunotherapies have a chance to kick in.

Our team discovered a potential way to overcome these barriers by combining a specific type of chemotherapy with our anti-PDL1 cancer immunotherapy. We found that in this case, measuring the amount of immune system cells in the tumor could potentially be used as a biomarker to predict response. When we used this as a biomarker we found that we could shrink tumors in nearly 40% of women with a specific type of TNBC.

The main takeaway is that when it comes to immunotherapy we need to look at multiple biomarkers, beyond PD-L1 protein expression, and multiple combinations to determine which approach is right for each patient and each tumor type.

Q: There is a controversial belief that at this moment we do not have a good baseline that can tell who is going to respond to treatment or not.How important is it to understand the immune regulatory influences within the tumor microenvironment and what are the challenges associated with it? What are we missing?

A: I agree that it may currently appear difficult to predict responses to cancer immunotherapies, but the field is actively testing a number of new hypotheses with very promising early results.

One hypothesis is that these types of immunotherapies only work in people whose immune system has previously recognized cancer cells. This has led to a number of new insights into how immunotherapies interact with the immune system and with different cancers.

I don’t think we’re truly missing anything, but we do know that a single biomarker alone is unlikely to be sufficient to predict responses to cancer immunotherapies, as it probably does not capture all of the relevant biology. Other immune-related factors are also important to consider, like the overall number of mutations in a tumor that leads to higher amounts of cancer-specific proteins called neoantigens. We know that higher mutational load can predict responses to anti-PDL1 and anti-PD1 medicines, so this is an important biomarker to continue studying. Other promising biomarkers related to immune regulation include gamma-interferon genes and T effector cell signatures.

The field is moving quickly, and I’m encouraged by these new insights into how the immune system interacts with immunotherapies. A more holistic view of protein biomarkers, immune cell signatures and mutational load together could provide a useful path forward for predicting responses to immunotherapies across a wide range of patients and tumor types.

Q: What is the status quo of developing companion diagnostics for immuno-oncology? What are some of the test limitations that we need to overcome?

A: The most well-known companion diagnostic for cancer immunotherapy is using immunohistochemistry (IHC) to measure PD-L1 protein expression. For these tests, we use antibodies that bind to PD-L1 and then visualize how many of those antibodies are bound. This lets us estimate how much PD-L1 is present in a given tumor sample.

But there are many different IHC tests, and as I mentioned above, the thresholds for PD-L1 positivity can also be different. So not only do these tests use different antibodies with different molecular characteristics, it is difficult to compare thresholds between them as well. In addition, some tests only look for PD-L1 expression on tumor cells, while other tests also measure expression on immune cells.

This leads to important differences between tests, which can impact the success of a clinical trial. These tests need to be rigorously validated before we can use them consistently and reliably across trials, which will take diligence and a lot of hard work.

This is why, in parallel, we are taking a comprehensive approach to biomarker development that integrates information about multiple diagnostics and multiple medicines simultaneously.

Q: Is there anything else you would like to add and share with the readership?

A: Personalized medicine is truly science in motion. For some types of medicines, we’re in the middle of the revolution, but for cancer immunotherapy it’s just the beginning.

Although the science is early, and navigating the complexity of the immune system is no easy task, I believe we are at a tipping point in the evolution of cancer treatments. All of the investments we have made in the human/cancer genome projects over the last few decades are now paying dividends, and there is an explosion of information about how each cancer can uniquely interact with the immune system. By translating this information into advanced biomarkers, I’m truly excited to see the full potential of immunotherapies realized in the near future.