The availability of ICIs that target PD-1, PD-L1, and CTLA-4 has given oncologists a new way to treat cancers. We now have a chance to cure some patients with metastatic solid tumors using immunotherapy. Very few patients experience serious side effects from these agents. On the flip side, some individuals derive no benefits.

There is an immense desire to identify those patients who are on a cure trajectory with ICI therapy. We also want to know which patients will achieve no benefits, more modest benefits, or a limited duration of response from ICI therapy. PD-L1 expression, mismatch repair deficiency (MMRD), and tumor mutational burden (TMB) are the 3 markers that are most commonly used to assess the potential for ICI response. High levels of microsatellite instability may occur together with MMRD. While these markers are helpful indicators, they do not give us a complete picture of the potential for response, and we seek greater precision. Deep learning (abstract 2619) and artificial intelligence (abstract 2621) approaches incorporating immune biomarkers were studied toward this end, and results were presented at the 2022 ASCO Annual Meeting. It is critical that we obtain the most complete information that we can about both the immune background and the genetic background of the tumor at diagnosis. This includes next-generation sequencing (NGS) testing and PD-L1 expression levels.

Of the 3 markers, PD-L1 levels are the most useful overall. Data from the meeting suggest that, among patients with advanced non–small cell lung cancer and very high levels of PD-L1 (≥50%), the use of chemoimmunotherapy did not appear to provide additional benefits beyond anti–PD-1/PD-L1 therapy alone (abstract 9000). This finding exemplifies why it is so important to determine PD-L1 levels.

As previously noted, in addition to immune markers such as PD-L1, we need to know about the genetic background of the tumor. In non–small cell lung cancer, one should test for actionable drivers such as EGFR mutations before considering ICI therapy, which is not necessarily intuitive and highlights the importance of NGS testing. Comprehensive genomic testing identifies patients with actionable alterations such as EGFR mutations and ALK rearrangements, who benefit from targeted therapy and are much less likely to benefit from ICI therapy.

MMRD represents a small subset of patients in oncology, but the dramatic benefits from ICI therapy in those with this deficiency make it critically important to identify and treat them quickly. The importance of MMRD was highlighted in abstract LBA5 by Cercek et al at the 2022 ASCO Annual Meeting. In rectal cancer, multimodality care is the standard. This study, in a carefully selected population, flipped the paradigm on its head by starting with immunotherapy alone. The first 12 patients with MMR-deficient rectal cancer had complete responses following treatment with single-agent dostarlimab, an anti–PD-1 agent. There was no evidence of disease on imaging or endoscopy. These were patients who were destined to have chemotherapy, surgery, and radiation, but, thus far, none of them have received any of these standard modalities because all 12 patients have responded. Individuals both with and without germline abnormalities benefit.

TMB is another informative marker that can be measured using NGS. This was also evidenced in the study by Cercek and colleagues where, in addition to MMRD, patients had a very high TMB. An analysis by NGS confirmed microsatellite instability in all patients for whom testing was performed and revealed the high TMB as well.

There were other data from the meeting that reflect researchers’ efforts to develop other parameters to predict ICI responses. These include assessments of tumor-infiltrating lymphocytes, the identification of an inflamed phenotype via radiomic techniques, 12-gene expression scores, and microbiome types. While all of these parameters have potential, none of them are robust ways of identifying who is most likely to benefit from ICI therapy. Thus, despite the worldwide efforts to identify new biomarkers of response, PD-L1, MMRD, and TMB remain the most useful. 

Finally, it should be noted that the use of ICIs in earlier lines of treatment may influence the response characteristics, and, indeed, the predictive markers. All patients with rectal cancer who achieved dramatic benefits in LBA5 were treatment naive. This has also been observed in other malignancies such as lung cancer. ICIs may prove to be more useful as upfront agents at all stages of disease, not just in advanced or metastatic disease.