However, she noted that immunotherapeutic approaches such as dendritic cell vaccines and oncolytic viruses may induce a host immune response at the tumor site, providing an environment in which checkpoint inhibitors exert clinical activity. She and her colleagues recently published a preclinical study14
of human tissue samples and a murine model that identified a population of tumor-infiltrating myeloid cells that increased with dendritic cell vaccine therapy and accounted for the majority of PD-L1 expression in the tumor microenvironment. Furthermore, treatment with a PD-1 inhibitor and colony stimulating factor 1 receptor inhibitor significantly improved survival in the mouse model, suggesting that the addition of a checkpoint inhibitor or another agent that blocks the tumor-infiltrating myeloid cells may help patients who demonstrate a less durable response to vaccine therapy.
“There are probably going to be subgroups of patients who have different responses to immune or vaccine therapy,” said Liau. “Going into the future, we need to figure out who will respond to which type of therapy.”Antibody-Drug Conjugates
EGFRvIII, a tumor-specific, constitutively active form of EGFR, is found in 20% to 30% of glioblastomas. However, monoclonal antibodies (eg, rituximab [Rituxan]) and small molecules targeting EGFR such as erlotinib (Tarceva) and gefitinib (Iressa) have not shown efficacy in GBM, in part because commonly used EGFR-targeted therapies do not work with the EGFR abnormalities, amplifications, or mutations in the extracellular domain in GBM, said Martin J. van den Bent, MD, PhD, of the Erasmus MC Cancer Center in Rotterdam, the Netherlands, in an interview with OncLive®
ABT-414 is composed of a tumor-specific antiEGFR antibody (ABT-806) linked to monomethyl auristatin F, a microtubule cytotoxin, and selectively targets cells with EGFR amplification, overexpression, or mutation (such as EGFRvIII).
According to van den Bent, ABT-414 acts like a “Trojan horse” because the tumor’s EGFR receptor is targeted to internalize the compound and increase the cytotoxic effect in the tumor cell. A pooled analysis of 126 patients with EGFRamplified recurrent GBM demonstrated an ORR of 10.4% and disease control rate of 52%.15
The follow-up phase IIb/III Intellance1 trial (NCT02573324) will randomize patients with newly diagnosed GBM to receive ABT-414 or placebo along with standard-of-care therapy. Positive outcomes in this trial could indicate an additional therapy to add to the standard of care for patients with EGFR-amplified GBM, as well as a proof of principle for more effective delivery of other targeted agents, van den Bent said.
According to Holland, the increase in concentration of cytotoxic agents within the tumor cells with this “Trojan horse” mechanism may also kill cells adjacent to the target, even if they do not express the EGFR mutation. However, he cautioned that the heterogeneity of the cells within GBM tumors makes it difficult to predict the success of therapies relying on a single target. “We’re looking for the therapeutic window, but the problem is that the population you’re targeting is very heterogeneous and they’re not all going to have the thing you want to target,” said Holland. “That’s been the general problem all along, from small molecules to antibodies to radiation.”Clinical Trial Design: Key to Optimizing Treatment
Until recently, the World Health Organization (WHO) classification of primary brain tumors has been based solely on histopathologic criteria. However, large-scale efforts such as The Cancer Genome Atlas (TCGA) Network demonstrating the clinical relevance of genetic and epigenetic alterations prompted the creation of diagnostic entities that integrate histopathology and molecular signatures in the 2016 WHO classification system.16
According to Holland, genetics are a major driver of tumor behavior and response to treatment and should be considered when assessing efficacy of a given treatment in clinical trials and predicting which patients will respond. “If your control arm happens to have a lot of patients who are genetically different from your study arm, you could make a drug that actually works quite well look no good at all,” he said.
A recent example of the control arm performing better than expected was the recently discontinued phase III ACT IV trial,8
in which the median OS was 21.1 months in the control group and 20.4 months in the experimental group receiving rindopepimut. Although the investigators are still researching why the control arm performed better than they expected with standard-of-care therapy (approximately 15 months), Holland stated that genetic analyses of tumors should be incorporated more broadly in the design of trials for GBM. “We have that technology now…That’s the kind of thing that, when we design trials, we should watch carefully to make sure we’re not running off the rails.