ASCO18 Update: Using Biomarkers to Improve Immunotherapy

Immunotherapy has firmly established itself as one of the major pillars of cancer treatment, but far from being a magic bullet, many challenges still need to be overcome to improve immunotherapy’s effectiveness as well as its safety. On the final day of ASCO18, several sessions took an in-depth look at these issues and discussed where we are now and how we can get to where we need to be.

Much focus has been placed on biomarkers recently, but for the most part, at least in the context of immunotherapy, their potential has remained largely untapped. Several biomarkers, such as expression of PD-L1 or certain immune gene signatures—have proven useful in guiding immunotherapy decisions, but there remain many more possible biomarkers to be discovered, validated, and then applied in the clinic, and Tuesday’s sessions explored several approaches that may help achieve that.

Mutations are one class of biomarkers that has emerged as being clinically useful. Aside from individual mutations that can drive the development of certain types of cancers, a tumor’s overall mutational burden—known as TMB—can influence a patient’s odds of responding to immunotherapy, as discussed in our Day 3 and Day 4 recaps.

David Gandara, MD, of the University of California, Davis, and Vamsidhar Velcheti, MD, of the Cleveland Clinic’s Taussig Cancer Institute, kicked off the final day at ASCO by discussing a retrospective analysis to gauge TMB’s value as a biomarker in seven trials involving patients with advanced cancer who were treated with atezolizumab, a PD-L1 immunotherapy. Gandara and his colleagues found that there was a correlation between mutational load, predicted neoantigens (the abnormal markers that arise from mutated genes), and objective response rate: at all TMB cutoffs examined, patients’ response rates increased with higher TMB levels.

For the purpose of balancing these improved response rates with mutational loads commonly seen in the clinic, the high TMB cutoff of ≥16 mutations per DNA Megabase pair was chosen for further analysis. This high TMB cutoff was able to identify a patient population distinct from those identified with PD-L1 expression analysis. Using these independent biomarkers together in a complementary fashion enabled Gandara’s team to reveal that patients whose tumors were characterized by both high TMB and high PD-L1 expression had significantly higher response rates.

In lung cancer patients who received immunotherapy in the second line setting, the high TMB / high PD-L1 group had a 38% response rate, compared to 20% for low TMB / high PD-L1 group, 8% for the high TMB / low PD-L1 group, and 9% for the low TMB / low PD-L1 group. In patients with bladder cancer who received immunotherapy in either the first- or second-line setting, the high TMB / high PD-L1 group had a response rate of 50%, compared to 27% for low TMB / high PD-L1 group, 25% for the high TMB / low PD-L1 group, and 12% for the low TMB / low PD-L1 group.

Whereas Gandara discussed work that analyzed tumor tissue itself to determine TMB and its association with patient responses in a retrospective manner, Velcheti, discussed a trial that is using blood samples to determine TMB and evaluate its relationship to immunotherapy’s effectiveness in a prospective manner. (We’ll come back to the important implications of this blood-based approach soon.)

Velcheti presented interim results from the B-F1RST trial in which atezolizumab was used as a first-line therapy for patients with advanced lung cancer. Among the 58 patients who were treated and evaluated for TMB, eleven were classified as high TMB (≥16 mutations per DNA Megabase pair) and had a response rate of 36.4%, whereas 47 were classified as low TMB and has a response rate of just 6.4%. The high TMB group was also characterized by increased progression-free survival—9.5 months versus 2.8 months—and a 49% reduction in the risk of progression compared to patients whose tumors were characterized as low TMB.

Following Gandara and Velcheti was Hossein Borghaei, D.O., of the Fox Chase Cancer Center, who discussed their presentations and put the current efforts in biomarker development in perspective. While Borghaei acknowledged that TMB looked promising as a predictor of improved efficacy in response to PD-1/PD-L1 checkpoint immunotherapy, he noted that there is currently no data regarding overall survival benefit, though improved response and progression-free survival rates bode well for those anticipated results.

Presuming those overall survival benefits do emerge with more follow-up time, Borghaei mentioned several other issues that must still be addressed in order ensure that the TMB biomarker is implemented most effectively into clinical practice.

First, there are at least several platforms that can be used to determine TMB, and not all do so in the same way. Standardizing the methods through which TMB is determined—or at the very least ensuring that they are mutually compatible and comparable—will be crucial in order to ensure consistency between different studies so that insights from different platforms used by different institutions might eventually be combined to more efficiently shed light on the bigger picture while minimizing the potential for needless replication.

Second, there currently is no agreement on what constitutes a “high” tumor mutational burden. The above studies with atezolizumab defined it as ≥16 mutations per DNA Megabase pair, but other studies, particularly the CheckMate-227 trial involving nivolumab, used ≥10 mutations per DNA Megabase pair as the cutoff in their retrospective analysis.

This second observation may be less of a problem (and more of a necessary question to be answered) than it seems at first, as the TMB thresholds that enable useful stratification of patients may well vary by cancer type as well as by what immunotherapy is being evaluated. In some cancers, or with some immunotherapies, patients might begin to reap the benefits associated with “high” TMB at different points in the TMB spectrum. Future efforts will need to determine the clinical significance of TMB in a setting-by-setting basis.

In addition to TMB’s significance in checkpoint immunotherapy, pre-existing immune responses against tumors, as measured by the extent of “killer” T cell infiltration into tumors, and, to a lesser extent, PD-L1 expression by tumors and immune cells in the tumor microenvironment, have also been correlated with positive responses to checkpoint immunotherapy.

However, these three biomarkers still can’t fully explain why some patients respond and others don’t, nor can they help doctors predict with a high degree of certainty whether an individual patient will benefit. After all, in the high TMB / high PD-L1 patient group referenced above in Gandara’s talk, no more than half of the patients responded to checkpoint immunotherapy in any cancer type. Even more importantly, their value has really only been demonstrated in a couple tumor types and only in the context of checkpoint immunotherapy.

As a result, Borghaei stressed the importance of discovering and validating new biomarkers, to ensure that existing immunotherapies as well as new ones are administered to individual patients so as to maximize their benefit. In all likelihood, even for a single type of immunotherapy, there won’t ever be one single perfect biomarker; utilizing panels of multiple biomarkers will almost certainly be important, and the important biomarkers to consider may very well differ according to the type of cancer type as well as the type of immunotherapy.

Borghaei also pointed out one last hurdle that will need to be overcome for the field to improve the science behind biomarkers for cancer immunotherapy: the tumor tissue that supplies the source for many of these studies.

As of now, tumor tissues are the gold standard when it comes to biomarker analysis, as they provide a look into the tumor microenvironment in its entirety and allow for the spatial relationships between different cells and molecules to be discerned.

Unfortunately, acquiring this tumor tissue is an invasive process for most cancers and it is often in scarce supply. The less tumor tissue that biomarker analyses require as a source, the better. In other words, the more easily that biomarker information can be gleaned without having to invasively acquire limited tumor tissue from patients, the better we’ll be able to advance our understanding of specific biomarkers in specific contexts.

In our Day 4 recap, we highlighted one radiomic approach that appears to have potential value in assessing T cells within tumors, and above Velcheti showcased the ability to determine TMB through a simple blood draw, but there is still much progress that must be made in developing alternative approaches for gathering this crucial information on tumors and the immune system.

To that end, a later session on Day 5 focused on next-generation diagnostics that could potentially minimize the field’s dependence on tumor tissues for biomarker assays, at least in certain situations so that those precious tumor samples can be used for other clinical and research investigations.

Patrick Ma, MD, of the West Virginia University Cancer Institute, began the session by highlighting two FDA-authorized companion molecular diagnostic tests—MSK-IMPACT and FoundationOne CDx—that enable the interrogation of tumor genomes.

MSK-IMPACT can provide information on 468 different cancer-associated mutations or alterations, whereas FoundationOne CDx can detect cancer-associated alterations in 324 different genes as well as the presence of two types of genomic signatures, including high microsatellite instability (MSI-hi). With that information, doctors can determine whether a patient might benefit from certain targeted therapies, or in the case of MSI-hi tumors, from checkpoint immunotherapy. While these tools don’t eliminate the need for tumor tissue, they do ensure that the limited supplies can be used more efficiently.

Aside from tumor tissue, one of the most promising places to acquire information on tumors and the immune system is the blood, which we know provides a potentially rich source of immune cells, circulating tumor cells, circulating tumor DNA, cell-free DNA, proteins, and metabolites.

These components in the blood can be analyzed through liquid biopsies. As noted by Sai-Hong Ignatius Ou, MD, PhD, of the University of California Irvine School of Medicine, liquid biopsies have already been used to identify specific driver mutations (those that fuel cancerous behavior) as well as mutations associated with resistance to certain treatments. They are also beginning to be used to measure overall TMB and to detect low levels of leukemia, known as minimal residual disease, which might not be found through traditional analyses. Doctors could also use the information obtained through liquid biopsies to determine patient prognosis, predict responses to specific treatments, and even monitor responses or side effects throughout the course of treatment.

And it’s not just blood-based biomarkers that could revolutionize our approach to treating cancer with immunotherapy. Our breath contains thousands of volatile organic compounds that can shed light on our overall health and disease, especially lung cancer. Saliva and urine biopsies have been used to detect EGFR mutations, which are commonly found in several types of cancer and can be used to guide treatment decisions.

Perhaps most fascinating, as we’ve come to appreciate how bacteria in the gut can influence cancer, the immune system, and responses to immunotherapy, stool samples have been used to provide insight into the different species that dwell within our intestines. Some species have been linked to improved immunotherapy responses, while a lower bacterial diversity appeared to be associated with worse patient outcomes.

As the speakers during this session made clear, there are abundant sources from which doctors can obtain this potentially valuable information. The next step will be to conduct large scale studies that help separate the signals from the noise, and then to develop innovative clinical strategies through which these insights can be applied most effectively in the clinic to ensure that each patient is treated in the way that is most likely to benefit him or her.

And with that, our coverage of ASCO18 comes to a close. We know we’ve provided a lot of complex information over the past week. To help you make sense of it all, we’ll soon be providing an overall recap that highlights the major takeaways from the conference, so check back in with our blog soon!

In the meantime, you can watch this short video of a panel discussion we convened involving three immunotherapy experts who share their views on the most important new data that came out of ASCO this year that patients should know about.