In an interview with Oncology Learning Network, Stephen J. Bagley, MD, MSCE, Assistant Professor of Medicine, University of Pennsylvania Perelman School of Medicine, Philadelphia, discussed the potential role of plasma cell-free DNA (cfDNA) as an effective prognosticator and substitute for tumor burden in patients with newly diagnosed glioblastoma.

What existing data led you and your co-investigators to conduct this research?

In multiple other solid tumors, liquid biopsy has shown value as a noninvasive means of tumor molecular profiling, a prognostic tool, and a way to monitor the tumor’s response or progression on therapy. Most of the liquid biopsy in clinical oncology to date has centered around plasma cfDNA, which is short fragment DNA that is present in the blood. A small proportion of the overall plasma cfDNA pool in cancer patients is tumor-derived.

In certain tumors, such as non–small-cell lung cancer, liquid biopsy using plasma cfDNA is increasingly utilized in routine clinical practice. For example, resistance mutations to EGFR targeted therapies can be detected in the blood of lung cancer patients, allowing the oncologist to choose a more effective targeted therapy without the need for an invasive tissue biopsy.

In many other tumors, this technology is not used as routinely in the clinic but has become an important correlative biomarker in clinical trials, particularly with regard to monitoring the tumor’s molecular evolution on treatment. In glioblastoma, however, which is the most common malignant primary brain tumor in adults, use of liquid biopsy has been more challenging due to lower levels of plasma cfDNA compared to other solid tumors.

As a result, this has been a relatively understudied area in glioblastoma research. We decided to design a prospective pilot study to get a sense of the potential clinical utility of plasma cfDNA in this disease.

Please briefly describe your study and its findings.

We enrolled 42 patients with newly diagnosed glioblastoma and began banking plasma from these patients even before their first surgical resection, which is the first step in the care of these patients. We therefore were able to quantify plasma cfDNA concentration at a timepoint when these patients’ tumor burden is at its highest.

We then continued serial plasma cfDNA collection during the course of their standard front-line treatment, which consists of adjuvant radiation with concurrent temozolomide chemotherapy, followed by 6 months of maintenance temozolomide.

The plasma cfDNA collections corresponded to routine MRI scans that were obtained to monitor the tumor radiographically over time. We therefore ended up with a dataset that consisted of paired cfDNA collections and MRI scans for these 42 patients from initial diagnosis through the time of their first disease progression.

There were 3 main findings from our study. The first was that the baseline plasma cfDNA concentration (ie, the plasma cfDNA obtained prior to initial surgical resection) was independently associated with subsequent clinical outcomes. The higher the cfDNA concentration at baseline, the worse progression-free and overall survival the patients went on to have.

The second key finding was that at certain, but not all time points, the plasma cfDNA concentration was correlated with radiographic tumor burden. Importantly, there were some patients in whom the plasma cfDNA spiked in the blood prior to radiographic evidence of tumor progression. Although our study was too small to prove this, we hypothesize that plasma cfDNA may eventually be used as a marker of tumor progression that can complement what is happening on the patient’s MRI scan.

Our third main finding was that >50% of our patients who had plasma cfDNA next generation sequencing had at least one tumor somatic mutation detected in the blood at their baseline (pre-operative) blood draw. Although the concordance with matched tumor tissue DNA sequencing was low, this raises the question of whether liquid biopsy may be able to detect tumor somatic mutations in glioblastoma beyond what is detected in the limited amount of tissue that is typically subjected to next generation sequencing in this disease.

Because of the pilot nature of our study and limited sample size, these results are not immediately translatable to clinical practice. We are currently expanding our initial pilot study up to about 200 total patients. We hope to obtain more conclusive results with this larger sample size and to validate our findings and determine how liquid biopsy can be optimally integrated into the clinical care of patients with glioblastoma.

Eventually, we aim to develop plasma cfDNA as a prognostic tool and noninvasive biomarker of tumor burden in this disease. There may also be a role for combined tissue and plasma next generation sequencing in patients with glioblastoma if we can ultimately show that this provides additional clinically valuable information beyond tissue sequencing alone.

Is there anything else pertaining to your research and findings that you would like to add?

Cerebrospinal fluid for liquid biopsy in glioblastoma is also an appealing research direction moving forward. Work from other research groups has shown that cerebrospinal fluid has higher yield for detecting tumor-derived cfDNA compared to plasma.

Although this requires an invasive procedure, typically a lumbar puncture, it is possible that this could also become an extra tool in taking care of patients with glioblastoma. Although this was not a focus of our current study, we also aim to explore this in more detail in the future.