NEW YORK (GenomeWeb) – A group of researchers from the Netherlands Cancer Institute and the University of California, Los Angeles have proposed a multi-pronged approach they call a "cancer immunogram" for assessing an individual cancer patient's likelihood of responding to cancer immunotherapies.
The method, described in a perspective last week in Science, could give a more comprehensive picture of the interactions between a cancer and the immune system than individual genomic features or other biomarkers.
In their publication, the authors — Christian Blank, John Haanen, and Ton Schumacher from the Netherlands Cancer Institute (NKI), and Antoni Ribas of UCLA — described seven initial parameters for this immunogram, including a tumor's genomic (and in turn immunological) foreignness, a patient's general immune status, immune cell infiltration into tumors, and the absence or presence of specific checkpoints like PD-L1.
The full seven, they argue, represent a reasonable start to building a cancer immunogram, which will most likely change over time as more is understood about the interaction between cancer and the immune system.
Blank, leader of an NKI group focused on immunotherapy, told GenomeWeb this week that he and his colleagues in the Netherlands have been collecting data on how well they can predict patients' response to ipilimumab (Bristol-Myers Squibb's Yervoy) using a combined approach that involves at least some of the seven factors in the proposed immunogram.
"We have not collected everything, but with four parameters, we can already distinguish patients that have no benefit … [at] a much higher percentage than [other single strategies or combinations] published before," he said.
At its core, Blank and his colleagues' proposal rests on a recognition that cancer-immune interactions are based on a number of what they argue are largely unrelated factors, all of which can precipitate or contribute to response or resistance to immunotherapeutic treatment.
Taken as a whole, the framework views T-cell activity as the ultimate "effector mechanism." In other words, while other factors — like the microbiome — may hold promise as therapeutic targets either in their own right or in enhancing the effect of existing immunotherapies, their effectiveness is likely to be through an ultimate effect on T-cell activity itself.
Importantly, the authors wrote, it's clear that the relative value or contribution of each parameter that falls under this umbrella of T-cell activity can differ greatly between patients. This variability likely underlies the failure of individual genomic biomarkers or other clinical features — like PD-L1 expression, a patient's general immune status, or a tumor's genome-wide mutational load — to explain more than a small proportion of the differences in response to immunotherapy from one cancer patient to another.
Mutational load (and associated neoantigen load) in tumors has received much attention as a tool for predicting immunotherapy response. However, Blank and colleagues argued that mutational load is an imperfect marker for tumor foreignness because the immune system can be rallied against a cancer not only by mutation-derived antigens, but also by self-antigen recognition.
"Tumors don't only have mutational load but they can also overexpress a normal antigen. It's an amplification from a normal protein, it's not mutated, but you just express it higher and this could also be recognized by the immune system," he said.
Genome-wide sequencing to assess mutational load also remains impractical for routine clinical use, although researchers are now looking for ways to infer a patient's overall mutational load using smaller and cheaper targeted assays.
Meanwhile, PD-L1 expression tests have been advanced as companion diagnostic assays to cancer immunotherapies, but these biomarkers themselves don't tell the full picture either, according to Blank.
He said that some research has shown that PD-L1 levels may not have much added value over other factors like T-cell infiltration or mutational load, but rather are a reflection of these other factors. More studies taking a comprehensive immunogram-style approach could help determine how strong PD-L1 expression remains as a biomarker amongst other contributing factors.
A host of other molecular and non-molecular factors can also inform on a patient's likelihood of responding to immunotherapy, including simple lymphocyte counts, the presence or absence of soluble inhibitors like the proteins IL-6 and CRP, measures of tumor metabolism, or markers of a tumor's sensitivity to immune effectors.
The relative contribution of each may vary from patient to patient or cancer type to cancer type. For example, Blank said, in melanoma, blood levels of lactate dehydrogenase, or LDH, appear to be a more predictive factor than mutational load, which may be more of a determinant in other cancers.
Importantly, the complexity of the interaction of these factors makes it complicated in turn for clinicians to identify the best action to take or for researchers to choose the most promising investigation to pursue.
"In some patients, intratumoral inhibition of tumor-specific T cells will be the sole defect that needs to be addressed, whereas in other patients, the tumor may simply be insufficiently foreign to elicit a clinically relevant T-cell response in the first place," Blank and colleagues wrote.
A full immunogram allows the simultaneous consideration of many different clinical questions, including whether the immune system sees a tumor as foreign, whether a patient's immune system is even sufficient to allow immunotherapy to work, whether there are T cells infiltrating a tumor, whether there are specific biomarkers that could hamper the activity of these cells, and whether, if these checkpoints could be mediated, the tumor would be sensitive to the unleashed T-cell response.
It's an exceedingly complicated picture that requires a combination of tumor genomics, immunohistochemistry, and a variety of other blood assays to paint in full.
Helpfully, Blank and his coauthors shared in their description of the proposed immunogram an example of a radar plot of the seven proposed parameters with a gradient shift from desirable (meaning likely to contribute to a higher chance of therapy response) versus undesirable states for each.
In an online supplement to the perspective, they outlined several hypothetical patient examples with associated plots, and their potential clinical implications.
The authors noted that their cancer immunogram should evolve. "Currently if you look at all the retrospective data on individual biomarkers, these seven parameters are the strongest," Blank said. "This is the first outline of this concept. In ten years I'm sure we will have some other effects to include," he added.