Skip to main content

Tool Measuring PD-1/PD-L1 Interaction May Add Precision to Immunotherapy Response Prediction

Premium

NEW YORK – As approved indications for immune checkpoint inhibitors continue to expand, so does the need for an accurate, specific, and affordable method for determining which patients are likely to benefit from these agents.

According to a recent Cancer Research paper authored by an international team of researchers, that method may require a shift in focus, from simply measuring the expression of PD-L1 on patients' tumors to assessing the functional interaction between PD-L1 receptors on tumors and PD-1 receptors on immune cells.

Currently, PD-L1 expression status is a key biomarker that oncologists use to direct immunotherapy. The US Food and Drug Administration has approved several anti-PD-1 drug indications for immune checkpoint inhibitors where patients' tumors must have a certain level of PD-L1 expression, as determined by immunohistochemistry (IHC) staining, in order to receive the therapy. But PD-L1 expression cutoffs identifying responders and non-responders can vary based on the tumor type, the immunotherapy agent, and the IHC assay.

And while studies have shown that in certain cancers higher PD-L1 expression is associated with better response to checkpoint inhibitors, there are outliers. Oncologists have seen enough patients with low PD-L1 expression who respond to immunotherapy that many in the community feel this is an imprecise biomarker for directing treatment and that better solutions are needed.

For example, pembrolizumab (Merck's Keytruda) is approved as a first-line treatment for advanced non-small cell lung cancer patients and metastatic patients who have progressed on platinum chemotherapy if their tumors have a PD-L1 proportion score of at least 1 percent. However, among patients who are considered to have very high PD-L1 expression, with a tumor proportion score of 50 percent or more, overall response to single-agent pembrolizumab is 41 percent.

Looked at another way, around 60 percent of NSCLC patients, despite having high-PD-L1 expression in their tumors, still don't respond to pembrolizumab. As such, many oncologists feel the jury is still out as to the best way to identify best responders to immunotherapy.

Peter Parker of the Francis Crick Institute in London, one of the authors of the recently published Cancer Research paper, is among those who believe that PD-L1 expression status isn't an accurate reflection of a patient's ability to respond to immunotherapy and that better approaches are needed. To explain why, he offered an analogy: "If you want to know what the ambient light state is in a room, you don't go around and count the number of light switches lying around," he said. "You need to work out what state they're in — whether they're on or off — in order to make that judgment."

Parker and his colleagues have developed a tool to measure PD-1/PD-L1 interaction levels in patients' tumors, which they believe could potentially improve the precision with which oncologists are able to identify if patients will or won't respond to checkpoint inhibitors. PD-1 is a checkpoint protein on T cells and when these receptors attach to PD-L1 ligands on other cells, the T cells get the signal to not attack them. Cancer cells use PD-L1 ligands to interact with PD-1 receptors on T cells to evade an immune system attack.

Checkpoint inhibitors try to block this interaction between PD-1 and PD-L1 receptors so T cells can recognize and attack cancer cells. Tumor cells with high PD-L1 expression, the thinking goes, may have more opportunity to engage with PD-1 receptors and signal T cells not to attack. The method developed by Parker and colleagues actually measures how well tumor cells are exploiting this interaction.

The tool is rooted in an imaging assay that provides a quantitative readout of immune-checkpoint interaction between cells. The technology, called immune Förster resonance energy transfer (iFRET), detected by fluorescence lifetime imaging microscopy (FLIM) measures cell-cell interactions in the range of 1 to 10 nanometers — a significantly smaller distance than other assays developed to measure checkpoint interaction. The distance allows for the specification of cell-to-cell interaction as opposed to mere proximity.

To develop the iFRET for measuring PD-L1/PD-1 interactions, the researchers started with the commercially validated Promega Blockade Bioassay, which was originally designed to measure antibody blockade of PD-1/PD-L1 and CTLA-4/CD80 interaction by luminescence, and adapted it to detect intercellular interaction of the ligand-receptor pairs. To do this, they used two primary antibodies to detect the ligand and the receptor, respectively, then mapped the proximity that appeared within that crucial 1 to 10 nanometer range.

"This is a technology that has been around for a long time in terms of being able to tag with fluorescent probes and look by [FRET]," Parker said. "But what we have done is develop a secondary antibody system so that you don't need to label all your primaries … you can have a much more flexible system to look through these secondary [antibody-binding fragments] with an amplification step that gives you a fairly sensitive way of measuring protein-protein interactions in situ in pathological samples."

Though this technology has been around for some time, Parker and his colleagues didn't expect that they would be able to home in on how a T cell interacts with a tumor cell. "But it turns out that, actually, if the proximity is appropriate and you make the right choice of antibodies, you can," Parker said.

To validate the test's ability to detect ligand-receptor interaction, the researchers applied it to cells from formalin-fixed paraffin-embedded (FFPE) tumor samples from a number of donor cohorts, including from patients with clear cell renal cell carcinoma, melanoma, and NSCLC. In the melanoma cohort, notably, 58 out of 117 patient samples that were categorized as PD-L1 negative by conventional IHC methods actually showed a PD-1/PD-L1 interaction state when tested with the iFRET method, and 42 patient samples considered PD-L1 positive, conversely, did not demonstrate an interaction, despite the higher concentration of the ligands being present in the tumors.

With the metastatic NSCLC sample cohort, for which the researchers had access to multiple samples per patient, as well as patients' clinical outcomes after treatment with pembrolizumab and nivolumab (Bristol Myers Squibb's Opdivo) for 40 of the samples, the test was applied retrospectively to validate its predictive capacity. Researchers were blinded to the patients' outcomes with checkpoint inhibitors at first.

After using the iFRET test and stratifying the samples from 60 NSCLC patients into two groups — those with the lowest 60 percent of median FRET efficiencies and those with the highest 40 percent of FRET efficiencies — the researchers determined that patients with the lowest FRET efficiencies had worse clinical outcomes on immunotherapy, according to Kaplan-Meier survival analysis. When the researchers evaluated outcomes based on PD-L1 expression status in the same cohort, they did not find a significant survival difference between PD-L1 positive and negative groups.

"This again shows the shortcomings of using PD-L1 expression levels to determine patient outcome," wrote Parker and co-authors.

As of now, Parker and his fellow researchers have only validated the predictive capability of their ligand-receptor interaction method retrospectively and in patients treated with anti-PD-1 agents. Their plan next is to replicate the method with other immune pathways beyond PD-1/PD-L1 and prospectively validate the approach.  

According to Parker, prospective validation trials are already in the works. Though he did not disclose specifics, he shared that these trials will be developed through partnerships with academic institutions and diagnostic and pharmaceutical companies. Further research will also be conducted to assess the tool's utility with other immune checkpoint pathways.

Additionally, because FRET technology already exists and involves a straightforward procedure, the results could theoretically be read by nonspecialized personnel.

"Probably five or 10 years ago, you'd need to employ people a lot smarter than I am to actually run these sorts of analysis," Parker said. "But now, one can use algorithms to extract the data that you need to extract. We're not replacing pathologists, but I think what we're doing is providing adjunct molecular data in combination with pathology to determine areas of interest."

The cost and turnaround time, Parker suspects, would be comparable to IHC assays currently used. And as far as the total cost of cancer care goes, immunotherapies are expensive, and improving the ability to determine which patients are unlikely to respond could save the healthcare system money.

Stephen Ward of the University of Bath, one of Parker's co-authors on the study, noted in a statement about the research that, in sparing the cost of expensive checkpoint inhibitors for patients who are unlikely to benefit from them, the method could ultimately have cost efficiency implications for the UK's National Health Service.

However, before the iFRET interaction measurement method can be used more readily in the clinic to direct immune checkpoint therapy for patients, a direct comparison will be needed comparing the predictive performance of this method with PD-L1 IHC tests. "You'd really want to see post-approval trials to see whether you would do better in terms of an alternative companion diagnostic," Parker said. "You'd just want to look at them head-to-head in a given setting for a particular agent."

From a patient perspective, he added, "If you know that the decisions that are going to be made for your treatment could be improved, you'd want to see that happen."