Skip to main content
Premium Trial:

Request an Annual Quote

Oxford BioDynamics Gathering Real-World Data on Checkpoint Inhibitor Response Predictor


NEW YORK – Oxford BioDynamics recently launched a blood test in the US with the goal of improving oncologists' ability to personalize checkpoint inhibitor treatment for patients, regardless of the type of cancer they have, stage of disease, or specific drug they're considering.

The Oxford, England-based company began taking orders from US healthcare providers in February for the EpiSwitch Checkpoint Inhibitor Response Test, or EpiSwitch CiRT. The liquid biopsy qPCR test evaluates patients' blood samples for an eight-biomarker signature before or during checkpoint inhibitor treatment and predicts their likelihood of response.

Oxford BioDynamics is offering its test in the US through partner Next Molecular Analytics' CLIA-certified laboratory. After doctors send patients' blood samples to the lab, they receive a report about a week later that indicates whether their patients have a high or low probability of responding to checkpoint inhibitors.

Oxford BioDynamics identified the eight-biomarker signature using its EpiSwitch discovery platform, which analyzes changes in the 3D structure of the human genome. The platform can identify alterations in the interactions between two genomic regions, called chromosome conformation signatures (CCS), which are key events in gene regulation.

"In our approach, we're looking across the whole genome and we're trying to capture very stable settings in a very complex cross-interaction between tumor and the patient's immune system," said Oxford BioDynamics CSO Alexandre Akoulitchev. "These [CCS] biomarkers represent hubs of the network that is established in the patient. Depending on the way that network is set, it's either conducive or not conducive for response to immune checkpoint inhibitors."

Oxford BioDynamics published a paper in December via the preprint server MedRxiv, in which researchers described how they relied on genomic data from 229 patients with 20 different kinds of cancer and treated with three immune checkpoint inhibitors — Merck's Keytruda (pembrolizumab), Genentech's Tecentriq (atezolizumab), and AstraZeneca's Imfinzi (durvalumab) — to train and validate the test. The study showed that EpiSwitch CiRT had 93 percent sensitivity and 82 percent specificity in predicting cancer patients' responses to checkpoint inhibitors.

In the initial stage of the study, researchers analyzed whole-genome sequencing results and clinical outcomes from 32 patients enrolled in an observational study at Mount Miriam Cancer Hospital in Malaysia to identify and select the relevant CCS biomarkers. In this discovery cohort, researchers found a total of 20 genes related to immune regulation that were associated with the eight markers ultimately included in the panel.

The most significant marker of response researchers identified is associated with the CD274 and PDCD1LG2 loci, suggesting interactions with the genes encoding for PD-L1 and PD-L2. "Immunohistochemistry tests have been approved by FDA that biologically deliver the same message, that there is something going on with expression of PD-L1 in certain statistical groups that have a trend for better response or non-response," Akoulitchev explained.

But the EpiSwitch test goes further, he continued, by considering those additional immune biomarkers and regulatory mechanisms associated with response to checkpoint inhibitors. The goal with EpiSwitch CiRT, Akoulitchev said, is to "bring to practice something that is robust, reproducible, and can be used on each individual patient reliably," to predict response.

Once the researchers narrowed the panel to eight biomarkers, they trained the predictive 3D genomic classifier model on a cohort of 77 patients with 11 tumor types, who received one of the three checkpoint inhibitors. They then tested the biomarker panel in two independent validation cohorts.

The first validation cohort was 24 patients and balanced between responders and non-responders. In that group, the test demonstrated 83 percent positive predictive value and 83 percent balanced accuracy, a measurement that averages the sensitivity and specificity of a model.

The second, larger validation cohort included 128 patient samples, again representing several treatments and tumor types. This cohort, however, had a larger proportion of non-responders. EpiSwitch CiRT predictive calls demonstrated 76 percent accuracy, 78 percent sensitivity, and 76 percent specificity in the larger cohort.

Better than current biomarkers?

Oxford BioDynamics is hoping to show the advantages of EpiSwitch CiRT over other methods of predicting checkpoint inhibitor response, such as PD-L1 expression or tumor mutation burden (TMB). In Akoulitchev's view, EpiSwitch CiRT's main advantage over these other biomarkers lies in its ability to characterize checkpoint inhibitor response across cancer types. The ability of PD-L1 expression and TMB to reliably separate responders from non-responders, by comparison, requires tumor-specific cutoffs, he noted.

"We saw that at this level of regulation in this biomarker modality, there is a very clear, consistent shared profile in patients who do not respond to PD-1 or PD-L1 interactions, [and] that's the core on which the test is built," Akoulitchev continued. "That is the advantage of this test against many others. We do not have to generate different thresholds [associated with treatment response and non-response] depending on the disease or stage."

For instance, PD-L1 expression testing can have a wide range of thresholds, depending on the treatment and tumor types, for determining when a patient is a responder. The US Food and Drug Administration has approved checkpoint inhibitors for cancer patients who have tumor PD-L1 expression levels in 1 percent to 50 percent of cancer cells. A 2017 review of predictive biomarkers for checkpoint inhibitor response suggested that the different thresholds of PD-L1 positivity contribute to the "poor reliability" of the biomarker to predict response.

The review authors noted additional limitations of PD-L1 expression measured by IHC, for example the fact that PD-L1 expression is regulated by various mechanisms, that it can be heterogeneous even within a single tumor, and that IHC does not take into account immune markers or tumor microenvironment factors that could also affect checkpoint inhibitor response.

"On the basis of these findings, PD-L1 immunohistochemistry alone is not yet an adequate biomarker for routine clinical use in deciding which patients to offer anti-PD-1 or anti-PD-L1 therapy to, and which patients would benefit equally from monotherapy versus combination anti-PD-1 or anti-PD-L1 therapies," the review concluded.

In a presentation at the 2019 Society for Immunotherapy of Cancer meeting, Oxford BioDynamics demonstrated the shortcomings of PD-L1 testing when compared to EpiSwitch CiRT by analyzing data from 99 non-small cell lung cancer patients who received Merck KGgA's checkpoint inhibitor Bavencio (avelumab) in a Phase II JAVELIN trial. That study stratified patients across three PD-L1 expression levels: greater than 1 percent, between 1 percent and 80 percent, or greater than 80 percent of cells. Across all three thresholds, PD-L1 expression levels were neither predictive nor prognostic of patients' overall survival or progression-free survival on Bavencio. The EpiSwitch test results, however, did correlate with patients' overall survival and progression-free survival on treatment.

"No matter where they put the PD-L1 expression cut off, they just couldn't tell responders versus non-responders by IHC," Oxford BioDynamics CEO Jon Burrows explained. "In the same cohort, when we took the day zero blood and ran it [on EpiSwitch], we completely recapitulated the responder versus non-responder profile, and that's really the power of looking beyond a single marker and at all the markers of gene regulation."

The company hasn't presented or published comparisons of EpiSwitch CiRT against TMB. In 2020, Keytruda received FDA accelerated approval as a treatment for refractory patients with solid tumors that have a high tumor mutational burden defined as having at least 10 mutations/Mb. However, since that approval, researchers have debated the value of TMB as a tissue-agnostic biomarker.

Several studies have been published showing that the TMB-high cutoff predicting immunotherapy response differs across tumor types. Researchers have pointed out that the endpoint of the trial that led to Keytruda's tissue-agnostic approval — response rate by radiographic scan — did not fully capture clinical outcomes like overall survival or quality of life endpoints. Others have gone further to suggest that TMB may not even be the right predictive biomarker to focus on because it's not necessarily what's driving patients' tumors. Rather, underlying biological mechanisms, like mismatch repair deficiency, may be what's causing the accumulation of tumor mutations and actually driving tumors. Merck has continued to back its tissue-agnostic approval for Keytruda with additional TMB data in a wider range of tumor types.

Oxford Biodynamics executives noted that Merck KGaA collected TMB data in the JAVELIN study, the same study from which Oxford used data to compare its test with PD-L1 expression testing at the 2019 SITC meeting. Although Merck KGaA decided not to present the TMB comparison at the SITC meeting, the two companies may do so in an upcoming paper, Oxford BioDynamics said.

Shoring up real-world data

Oxford BioDynamics is confident that its newly launched test will help oncologists better select patients for checkpoint inhibitor treatment compared to available biomarkers. The company expects doctors to consider EpiSwitch CiRT results alongside other patient data, like comorbidities or quality-of-life index, when deciding which treatment to prescribe.

For the initial launch, Oxford BioDynamics is targeting a network of early-adopter oncologists. The company will also market the test to doctors at oncology meetings and promote it via digital marketing in medical journals.

Some doctors have been eager to try out its test. Oxford BioDynamics provided one oncology practice in Florida with a number of sample collection kits, so that the practice can offer the test more easily without having to order each one individually, Burrows noted.

After shoring up some real-world experience with the test and collecting additional data, Oxford BioDynamics expects to submit EpiSwitch CiRT for FDA approval.

With its test, Oxford BioDynamics is hoping to address the unmet need for predictive biomarkers for personalizing immunotherapy. One meta-analysis found around 40 percent of patients across tumor types responded to first-line treatment with Bavencio, Tecentriq, and Keytruda. On average, the meta-analysis also showed across treatment lines that less than 20 percent of patients responded to single-agent PD-1 and PD-L1 inhibitors and more than 45 percent responded to checkpoint inhibitors in combination with chemotherapy.

In the absence of reliable predictive biomarkers, Burrows noted that most doctors don't want their patients to miss out on the chance to benefit from immunotherapy, and simply try giving them checkpoint inhibitors if they are eligible for it.

"The upside of that is you're not going to miss anybody that would respond, but the downside is that you're going to put someone on a treatment that might not help them," Burrows said. "It could take at least six to nine months to figure out that it's not helping them. In that time, they might miss the opportunity to be on a more effective treatment sooner."