In looking to the future of immuno-oncology, it is exciting to think about ways in which we can improve our ability to predict responses and treat patients at the individual level. The field is moving toward precision immunotherapy, a dramatic example of which was seen at the 2022 ASCO Annual Meeting in a study examining the use of PD-1 blockade in patients with mismatch repair–deficient, microsatellite instability (MSI)–high, locally advanced rectal cancer (abstract LBA5). Results demonstrated that all of the first 12 patients in the study had a complete response following treatment with single-agent dostarlimab, with no evidence of disease and no need for chemotherapy, surgery, or radiation during the follow-up period (range, 6-25 months). This is a prime example of precision immunotherapy, as included patients had MSI-high disease. When you have that type of change in that tumor, it is associated with robust neoantigen production and it predicts a high likelihood of response to immunotherapy.

Unfortunately, there are differences in markers of immune checkpoint inhibitor response between different cancer types, and, in non–small cell lung cancer (NSCLC), it has been challenging to find an MSI-high subset that responds in this way. Tumor mutational burden and PD-L1 expression are used to predict the likelihood of response to checkpoint blockade, but we would like to develop greater precision in our biomarkers of response.

Several potential biomarkers are being explored across tumor types, including immune phenotypic markers, gene expression signatures, gut microbiome signatures, and changes in the tumor microenvironment. One abstract described a pan-tumor, deep network machine learning analysis that could identify relevant prognostic biomarkers in patients treated with immune checkpoint inhibitors (abstract 2619). These approaches have the potential to increase the predictive power of the biomarkers that we already have and to identify new markers. This may be the future, but we need additional data, and, ultimately, novel markers need to be practical for use in routine clinical practice.

When considering immune checkpoints (eg, PD-1, CTLA-4, LAG3, and TIM-3), depending on the mechanisms involved, different biomarkers may be useful; however, I think that it will be important to have a single test, albeit perhaps a test that incorporates multiple variables. We will need to be able to integrate findings from the whole patient—looking at the clinical parameters, molecular parameters, circulating markers, and perhaps even microbiome characteristics, imaging biomarkers, and radiomics—in order to predict response. The more complicated part of all of this occurs outside of the tumor.

We also need to confirm that such a test is truly something that we can hang our hat on vs an expensive biomarker that essentially tells us what we already know. So, we will have to see how this develops over time. Intuitively, it feels right that we should be able to do some sort of gene expression profiling to predict responses, but I think that we are not quite there yet. Hopefully, improving biomarkers will lead to more trials such as the aforementioned rectal cancer trial, and we will be able to predict with greater certainty that an immunotherapy will achieve an outstanding response for an individual patient.

New therapies that target established checkpoints continue to be developed. Over time, new monoclonal antibodies might be engineered that mitigate some of the toxicities or improve an immune response in, say, a PD-L1–high patient; however, right now, there does not appear to be much of a difference in efficacy among the available antibodies that target the PD-1/PD-L1 pathway.

In regard to novel checkpoints, the LAG3 pathway is already being targeted in melanoma and may be of interest in NSCLC and in other tumors. Earlier this year, the US Food and Drug Administration approved relatlimab in combination with nivolumab for patients with advanced-stage melanoma. At ASCO 2022, based on the phase 2 trial by Felip and colleagues (abstract 9003), we are now seeing activity in NSCLC with eftilagimod alpha (a soluble LAG3 protein) and pembrolizumab in the first-line metastatic NSCLC setting, regardless of PD-L1 status. Eftilagimod alpha does seem to have some interesting characteristics in that it binds to a subset of major histocompatibility complex class II molecules to mediate activation of antigen-presenting cells and CD8+ T cells, and the subsequent T-cell recruitment may lead to a stronger antitumor response than with pembrolizumab alone. Investigators did not see a marked increase in serious side effects with the combination, and, certainly, the overall response rate of 37.3% was quite interesting. With a PD-L1 subgroup of 50% or higher, the response rate was 54.5%.

Adoptive cellular therapies have been shown to be effective in hematologic malignancies, and they are beginning to be evaluated in solid tumors. A number of chimeric antigen receptor T-cell and tumor-infiltrating lymphocyte (TIL) therapies are under development. For example, an oncolytic adenovirus armed with tumor necrosis factor alpha and interleukin 2 was shown to be generally safe in patients with advanced melanoma who were receiving adoptive TILs (abstract TPS9590). The adenovirus is designed to enable T-cell therapies and checkpoint inhibition, with the aim to expand the proportion of patients who are benefiting from immunotherapies. I am also still excited about the mesothelin chimeric antigen receptor or TIL therapies that are being developed for mesothelioma or mesothelin-expressing solid tumors. Further, it was interesting to see an autologous natural killer cell product with activity against heavily pretreated solid tumors (abstract 2644). These cellular therapies may advance the treatment of solid tumors, but more data are needed. We are on the precipice, but I think that it is going to take time to find the right cellular therapy in solid tumor malignancies.