CHARLOTTE, North Carolina (GenomeWeb) – At the annual meeting of the Association for Molecular Pathology here this week, representatives from NanoString Technologies and Merck Pharmaceuticals provided an update on their ongoing companion diagnostic collaboration, in which they are evaluating a gene expression-based assay, using NanoString's nCounter technology, to predict response to Merck's anti-PD1 immunotherapy Keytruda (pembrolizumab).
Representatives from both companies also spoke about how platforms like NanoString's are likely to find broader use in the growing cancer immunotherapy field, both for predicting response to additional immune-mediating drugs and for measuring other signatures that might complement the current 18-gene tumor inflammation assay that Merck is advancing for Keytruda.
Matt Marton, Merck's director of genomics and companion diagnostics, said that he could not share results from the ongoing clinical trials yet, in which the 18-gene signature is being evaluated. However, he said that the company has been seeing a high assay success rate in terms of being able to measure gene expression in clinical samples from the trials.
"We are well above 95 percent, which I think is pretty amazing," he said. "There are a lot of less complex assays that have had a lower assay success rate."
Marton and NanoString senior vice president of R&D Joe Beechem also discussed additional details from the preclinical validation of the signature, and how the partners strategized translation of what was initially a research assay into a clinical trial test that would serve as a future companion diagnostic.
The markers in the test panel fall roughly into one of four categories: T cell presence, natural killer cell presence, antigen presenting biology, and interferon gamma signaling. An algorithm creates a score that divides those likely or unlikely to benefit from treatment, based on being above or below an expression cutoff point for the genes.
Merck began working with NanoString to look at the feasibility of developing a CDx in mid-2015, and about nine months later signed the agreement to develop the test. However, by the time that milestone was reached, trials of Keytruda, in which Merck hopes to advance the NanoString assay, were already underway.
When it first garnered approval for Keytruda, Merck did so with an immunohistochemistry-based companion diagnostic developed by Dako that gauges expression of PD-L1.
Last month, the US Food and Drug Administration then expanded the approval for Keytruda and the accompanying Dako assay to first-line treatment of non-small cell lung cancer patients.
Marton called this recent approval a marker of the "huge undeniable success," of this drug and companion test.
However, he said, that leads to an obvious question regarding the company's new work with NanoString: "Given this huge success, why do you need a gene-expression assay?"
There are a lot of reasons, Marton said. First, it is clear that not all responders to checkpoint inhibitors like Keytruda are PD-L1 positive. Secondly, IHC detects a single analyte, and there is growing evidence that multiple factors may play a role in response or resistance to immunotherapy drugs. Additionally, IHC is a subjective testing method, which can suffer from variability based on the individual performing it.
"Gene expression profiling really should be less subjective and more quantitative," Marton said. Moreover, he added, as combination therapies are developed, a multiplex platform like NanoString's has the ability to potentially host multiple different gene signatures, which could be needed to determine optimal treatment for patients.
After deciding to explore the NanoString platform, Merck began discovery and validation of a gene expression signature on a research basis. This took place using new patient samples as they became available, starting with a very small initial discovery cohort of melanoma samples, and progressing through new sets from a variety of cancer types.
First, the company identified a 57-gene signature. Then, as it moved toward developing a clinical companion test, it refined this down to 18 genes.
At the AMP workshop, Marton shared data on Merck's experiments to assess the robustness and reproducibility of the assay, both of which he said were strong.
"The assay is very insensitive to having non-tumor present, but the best correlation is with more than 50 percent tumor, so that is what was chosen for the assay," he explained.
When the company looked at variability between samples of different sizes, "variation was less than 5 percent" he added, which suggests that using small core needle biopsies should yield comparable results to larger samples.
Also, he said, in reproducibility experiments in more than 10 tumor types, Merck saw low variance.
Importantly, Marton said, when the company worked to translate the research assay into the clinical trial form, it was relieved to find that it could easily transition the assay's cut-off point using a linear fit model.
Moving forward, Merck is advancing the 18-gene assay in several tumor-specific trials, as well as in the Phase II Keynote-158, a basket trial recruiting patients for treatment with Keytruda who have one of about 10 specific solid tumor types or any non-colorectal cancer with high microsatellite instability.
If the assay successfully stratifies patients who respond to Keytruda, the company will have powerful evidence to advance it as a CDx.
Stepping back, Marton also reflected on the evidence that platforms like NanoString's are likely to find broader use in immuno-oncology precision medicine. They are able to multiplex large numbers of markers, he said, and can potentially provide multiple biomarker signatures that speak to different aspects of biology that contribute to either response or resistance to this new class of drugs.
"This has been very exciting but I think it's only the first step of a trend for the use of gene expression profiling in immuno-oncology," he said.
"We are going to have a lot of different signatures yet to be discovered that could target non-responders, or speak to combination therapy," he explained, "so you need a platform where you can put a lot of different signatures."
Moving forward, he said, it will also be important to learn from the example of IHC-based PD-L1 assays, where siloed development has led to multiple assays that don't have perfect concordance with one another.
"With PD-L1, so far we have four tests, and it’s a mess now with three being similar and the other different," Marton said. "We have to come up with pre-competitive collaborations to prevent that from happening with gene expression."
NanoString has said that it is looking to connect with additional pharmaceutical companies to investigate its platform as a companion diagnostic for other cancer immunotherapies.
The company's deal with Merck stipulates that NanoString keeps rights to the test and can try to pair it with drugs from other manufacturers.