Many experts say that out of all the therapies developed to harness the power of the immune system to fight cancer over the past several decades, PD-1/PD-L1 inhibitors look to be the most promising, even though it is still relatively early days in their development. Maria Fe Paz and Patrick Considine of Diaceutics suggest that there is an urgent need to standardize and harmonize the assays used to evaluate PD-L1 expression, and for integrative initiatives to support the ongoing collaborations amongst the many different key stakeholders.
There is no doubt that immuno-oncology is a paradigm shift holding much promise in cancer treatment. The idea of boosting the immune system to fight against its own malignant cancer cells has provided a strong rationale and focus to researchers for decades, working to translate their studies from the bench to the clinic. Understanding the checkpoints responsible for the immunomodulation was key to facilitating the identification of drug targets, and development of monoclonal antibodies that are showing promising results in the clinical setting.
Not only the medical community, but many other stakeholders, are following with excitement the development of these immune checkpoint inhibitors. Some of the major pharmaceutical companies have initiated clinical trials, using immune checkpoint inhibitors as single therapies or in combination with other therapies for a range of cancers and an increasing scope of indications. The relatively mild safety profile makes them ideal for combinations, and they will very likely become the backbone of most cancer treatment schemes in the next years. Many experts say that out of all of the therapies developed to harness the power of the immune system to fight cancer over the past several decades, PD-1/PD-L1 inhibitors look to be the most promising, even though it is still relatively early days in their development.
In this competitive landscape, in which the market share will be at least partially influenced by the first comers to market, there is a growing need to improve the clinical outcomes to grant faster approvals. A recent Goldman Sachs review [i] suggests some companies with PD-1/PD-L1 inhibitors are betting initially on an ‘all-comers’ strategy, i.e., no requirement to segment patients based on biomarker profile. Such a strategy appears to fit those first into the market with a specific indication such as melanoma and other second line or smaller third indications. BMS have embraced this strategy for Yervoy in melanoma indication and for Opdivo in melanoma indication and third line squamous NSCLC. Merck has adopted the same strategy for Keytruda in melanoma indication and first NSCLC indication. Whilst the early clinical studies may support the ‘no biomarker’ strategy, the move of PD-1/PD-L1 therapies into first line may require the incorporation of a biomarker strategy.
The complexity of the interaction between the immune system and the tumor cells, the role of the microenvironment and other factors yet to be determined, are challenging the application and validation of the one single biomarker of response. Given the scenario in which many of the indications will require a combination treatment, the validation of a signature of response may be of paramount relevance to guide therapy decisions.
Immuno-oncology has been revolutionized with Programmed cell death-1 (PD-1) and its ligand, PD-L1, as the foremost targets of immunotherapies, with more than six of the major pharmaceutical companies involved in the development of antibodies and combinations with them. PD-L1 has been shown to be a potential predictive biomarker in exploratory trials, and has been incorporated as inclusion criteria in the recruitment of patients in an increasing amount of studies. BMS, Merck, Roche/Genetech and AstraZeneca clinical studies have included patient stratification based on PD-L1 status for some NSCLC indications. The application of the PD-L1 diagnostic biomarker as a predictor of therapy response presents a number of technical and clinical utility challenges. There is the unique scenario whereby each PD-1/PD-L1 therapy has its own proprietary PD-L1 CDx test. Independently, each CDx will be technically and clinically validated to achieve regulatory approval. However, there are some technical hurdles with the test and the technology itself. PD-L1 is an inducible biomarker, which is both expressed by tumor cells and also by tumor infiltrating lymphocytes, which are part of the microenvironment of the tumor. Some companies are measuring the staining of PD-L1 just in tumor epithelial cells, whilst others are including tumor infiltrating lymphocytes.
An additional complication is the diverse range of antibody clones used to measure PD-L1. Each assay may have a unique technical performance (sensitivity and specificity) and its own cut-off point to define PD-L1 expression positive. Some companies are considering a tumor as PD-L1 positive when staining more than 50 per cent of the tumor cells, while others are setting a three-score criteria similar to HER2, with different percentages of positivity (1-5 per cent, 5-10 per cent, >10 per cent). Nevertheless, IHC is semi-quantitative and based on subjective assessment, so there might be considerable variability and lack of reproducibility between pathologists. Collectively, the availability of different tests, variability of measurements and cut-offs, the hurdles of the technique itself along with the attempts to make it more accurate, reproducible and quantifiable, are presenting very difficult challenges to pathologists who will be charged with providing quality PD-L1 results to enable therapy decisions. In order to reduce the subjective nature of the assessment, a clear standardized definition of each category should be created and agreed upon. Once the technical standardization has been achieved, interpretative standardization can be addressed to facilitate physicians in their therapeutic decisions.
There is emerging evidence that PD-L1 may not be the ideal companion diagnostic biomarker to predict response to PD-1/PD-L1 therapies. BMS management have stated that clinical experts remain sceptical of PD-L1 status driving decision-making. Many studies do not separate the technical staining component and the interpretative assessment of staining, which can limit the impact of outcomes. Solid tumor tissue is inherently heterogeneous, containing a mixture of healthy and malignant cells in varying proportions. It is common for tumor content to alter between sections from FFPE blocks. The availability of sufficient homogeneous material may be a stumbling block for PD-L1 characterization. For the same reason, single core biopsy may not be the most adequate sample to determine tumor PD-L1 expression. Furthermore, the correlation of the measurements of PD-L1 in metastatic tissue and matched primary tumor are weak, which means that primary tumor testing may not be the ideal surrogate to determine the expression in metastasis and predicting response in metastatic patients.
There is ever growing evidence of objective and durable responses granting faster approvals of the drugs by regulating agencies, regardless of PD-L1 status. Partly as a consequence of the factors discussed above, but also due to the dynamic inducible nature of the immune response and evolving immunophenotype during the course of the disease and treatment; there is not a robust concordance among PD-L1 positivity and clinical outcomes, especially in some indications. PD-L1 expression as a biomarker seems only moderately useful, and the use as a single biomarker of response may miss patients that would respond despite being negative for the biomarker. There is a higher probability of response to the therapies by those patients expressing higher rates of PD-L1 on tumor cells, but there is still a significant number of negative patients who would respond. The turmoil among physicians is generating an additional layer of confusion to the existing bewilderment among pathologists, who see a complication for physicians to take ‘go/not-go’ therapy decisions given the lack of clear cut binary results on patients’ PD-L1 status. And in spite of it all, many physicians would not preclude a negative PD-L1 patient from benefiting from a breakthrough therapy if there are still immeasurable chances of response, or possibilities to synergize the effect with a combination, particularly in patients with limited therapeutic options.
The lack of clinical evidence supporting PD-L1 alone as a single biomarker of response is prompting the clinical research of additional measurements and biomarkers, such as mutation rates, immune scores/cytokine profiling, CD8+ T cell ratios, or neo-epitopes, gene signatures or RNA expression profiles both in the stroma or tumor infiltrating lymphocytes. It may well happen that the signature of response will vary among different indications, as may other tissue-specific or tumor-specific co-stimulatory molecules in the PD-1/PD-L1/PD-L2 pathway that could be influencing the response as well.
Looking to overcome the limitations of PD-L1 as a biomarker, there are many groups looking to develop additional quantitative technologies based on RNA expression of PD-L1, with some groups investigating measurements of biomarkers in liquid biopsies (peripheral blood and serum), to allow for serial analysis during the course of the treatment and disease monitoring. Interestingly, a recent retrospective study[ii] found that patients with NRAS-mutant melanoma seemed to respond better to immunotherapy compared with patients whose tumors had other genetic subtypes, and this was especially true for patients treated with anti–PD-1/PD-L1 therapies.
There is an urgent need to standardize and harmonize the assays used to evaluate PD-L1 expression, as lack of harmonization will be a barrier to drug/CDx adoption. It could also potentially lead to false negative results which would have a detrimental effect on patient care. PD-L1 is a complex testing landscape and thus laboratories would benefit from a collaborative study to accelerate the clinical readiness of the test. It is essential to define a reference method/material, or a unified test or algorithm, so as to provide traceability to a reproducible assessment of the threshold. Most importantly, generated data cannot be aggregated to develop guidelines until the results are standardized, and guidelines cannot be issued and implemented until all the methods are harmonized.
Finally, there is an urgent need for integrative initiatives to support the ongoing collaborations amongst the different stakeholders, including academic research institutions and university hospitals, pathologists, physicians, pharmaceutical and diagnostic companies and key bodies.
[i] Goldman Sachs (2015) Americas: Healthcare: Pharmaceuticals – Pharma Bus Trip: Takeaways from BMY, MRK, PFE and ZT (Published 27th February 2015) https://360.gs.com/research/portal//?st=1&action=action.binary&d=18911921&fn=/document.pdf&a=b7d239e796be4d8682ff584ca13328b2