Immunotherapy: Extending Its Reach, Predicting Responses

Alexander M. Castellino, PhD

November 26, 2014

Drugs that act as immune checkpoint inhibitors are seen as an important advance in oncology practice, with impressive responses seen in melanoma and indications of activity seen in many other solid tumor types. But a big issue with these drugs is that only about a third of patients respond, and there are intense research efforts underway to find ways to identify patients who are likely to respond.

Some clues have been identified in a series of letters published online November 26 in Nature, which provide insights into the impact of immune checkpoint inhibitors on other cancers and an understanding of the way they work.

The five letters "reveal a growing list of cancers that respond to checkpoint blockade and describe characteristics of those patients who respond to therapy," write Jedd D. Wolchok, MD, PhD, and Timothy A. Chan, MD, PhD, both from the Memorial Sloan Kettering Cancer Center in New York City, in a related comment.

Expanding Immunotherapies Beyond Melanomas

The letters highlight how the programmed cell death protein (PD-1), along with its ligand (PD-L1), has been the recent thrust of several immunotherapeutic agents.

In one letter, Roy S. Herbst, MD, PhD, from the Yale Comprehensive Cancer Center at the Yale School of Medicine in New Haven, Connecticut, and colleagues report that the PD-L1 antibody, MPDL3280A (under development by Roche/Genentech/Chugai), was shown to have responses in 175 patients with small-cell lung cancer, melanoma, renal cell carcinoma, and other solid tumors. The overall objective response rate (ORR) for all the solid tumors was 36%, and 6-month progression-free survival was 42%.

Another letter, by Thomas Powles, MD, from Barts Cancer Institute, the Queen Mary University of London, and Barts Experimental Cancer Medicine Centre in London, United Kingdom, and colleagues, documented durable responses with MPDL3280A in patients with urothelial bladder cancer.

A significant aspect of the study protocols was the inclusion of immunohistochemistry (IHC) screening and biomarkers to correlate clinical response with immune response, and the use of T-cell trafficking to predict response to therapy. IHC was done on pretreatment biopsies from archived paraffin-embedded tissue. PD-L1 expression was scored as 0, 1, 2, or 3, indicating the level of staining for PD-L1 in tumor and immune cells infiltrating into the tumor.

Dr Powles and colleagues showed that ORRs were higher in patients with tumor IHC scores of 2 or 3 than in patients with lower tumor IHC scores (43% vs 11%).

Dr Herbst and colleagues showed that for all tumor types, ORR decreased with PD-L1 expression. ORR was higher in tumors with an IHC score of 3 than in those with a score of 0 (46% vs 13%).

In non-small cell lung cancer, ORR was 83% in tumors with an IHC score of 3, and 6-month progression-free survival also of 83%.

"I think this is a launching point to use these findings as a predictive biomarker," said F. Stephen Hodi, MD, director of the Center for Immuno-Oncology and the Melanoma Treatment Cancer Center at the Dana-Farber Cancer Institute in Boston, who was senior author of the letter by Dr Herbst's team.

However, in their comment, Drs Wolchok and Chan are more cautious. Ever since the earliest reports of the effects of PD-1 blockade, PD-L1 expression by tumor cells has been the focus of studies looking for biomarkers to predict therapeutic response. Although it is clear that expression of PD-L1 on tumor cells makes it more likely that a patient will respond to PD1-pathway blockade, this is not a binary, static predictive marker, they write.

Expanding the Understanding on How to Predict Responses

In addition to observations that PD-L1 expression provides a likelihood of response, another letter, by Paul C. Tumeh, MD, from the Jonsson Comprehensive Cancer Center at the University of California, Los Angeles, and colleagues, indicates that "PD-1 blockage induces responses by inhibiting adaptive immune responses."

"We've had amazing clinical success treating patients battling advanced melanoma with pembrolizumab (Keytruda, Merck & Co). The challenge is that it only works in approximately 30% of patients with melanoma," Dr Tumeh told Medscape Medical News.

Now we have established a platform to determine the phenotype and density and location of different immune cells, and have found that CD8+ T-cells at tumor invasive margins most accurately predict response to therapy, he explained.

In a retrospective analysis of tissue samples from 46 patients, this team examined the density of CD8+ T-cells (or cytotoxic T-cells), PD1, and PD-L1 in the invasive margin where tumors infiltrate normal stromal cells and in tumor centers before and after treatment with pembrolizumab.

Dr Tumeh and his colleagues found that patients who responded to therapy had the highest number of CD8+ T-cells at the invasive margins of tumors, whereas patients who progressed had the lowest number. In addition, they found a high expression of PD1 in CD8+ T-cells and a high of expression PD-L1 in melanoma cells, macrophages, and lymphocytes.

The density of CD8+ T-cells increased from baseline to postdosing biopsy samples, indicating that releasing the PD-1 immune checkpoint with pembrolizumab increased T-cell proliferation and increased effector function, which was correlated with tumor shrinkage.

The predictive model was validated in the pretreatment biopsies of 15 patients; in this sample, treatment outcome was blinded.

"We need to validate these observations prospectively in a clinical trial," Dr Tumeh told Medscape Medical News. In addition, we need to classify "signatures" for the nonresponders, he added. The signatures relate to the expression of CD8+ T-cells, PD1, and PD-L1 before and after treatment in tissue biopsies.

A recent study provided insights into predicting response to the first checkpoint inhibitor, ipilimumab (Yervoy, Bristol-Myers Squibb), as reported by Medscape Medical News. That group correlated the likelihood of response to ipilimumab with a high mutational burden and the emergence of neoepitopes.

Two of the Nature letters expand on this, focusing on a preclinical approach to define how "passenger mutations" play a key role in tumor immunity using murine tumor cell lines and mouse tumor model.

One letter, by Mahesh Yadav, PhD, from Genentech in South San Francisco, and colleagues, reports on neoepitopes identified in mouse tumor cell lines. When exome sequencing, done on protein coding genes, of these tumor cell lines was compared with the mouse genome, the researchers identified seven neoepitopes in the MC-38 cell line using mass spectroscopy. Indeed, three of these neoepitopes elicited a CD8 T-cell response when vaccinated with adjuvant into the strain C57BL/6 strain of mice.

The other letter, by Matthew M. Gubin, PhD, from the Department of Pathology and Immunology at the Washington University School of Medicine in St Louis, Missouri, and colleagues, expands our understanding of neoantigens in the context of tumor biology. In a mouse tumor model, this team showed that neoantigens are causally related to immune checkpoint inhibition — an observation that cannot be made in human subjects.

This is the most convincing evidence that unmasking neonatigens is the causation related to immune checkpoint therapy, Dr Chan told Medscape Medical News. Using genomics and bioinformatics, Dr Gubin's team identified tumor-specific mutant proteins "as a major class of T-cell rejection antigens following anti PD-1 and/or anti-CTLA-4 therapy of mice bearing progressively growing sarcomas." In addition, they showed that they could use synthetic long-peptide vaccines that incorporate these mutant epitopes to elicit tumor rejection in much the same manner as checkpoint blockade immunotherapy.

The Future of Predicting Responses

These studies point to the future of predicting response to immunotherapies.

According to Dr Chan, the cellular changes seen in some of these reports are a result of the underlying mutational profile. Mutational profiles are hardwired into the DNA, he said. The cellular phenomenon could be a manifestation of the mutational profile, he said.

Dr Tumeh indicated that PD-L1 expression might be part of clinical decision making in the future. Once validated prospectively, the signature based on CD8+ T-cells, PD1, and PD-L1 could be used to determine the probability of a response with immune checkpoint blockade.

Dr Chan indicated that biomarker studies, such as the ones by Dr Herbst's team and by Dr Powles' team, should be integrated into future clinical protocols. Also, identifying neoepitopes in tumors might be the future of predicting response, he added.

The research into predicting which patients are likely to respond to immunotherapies is very important, Dr Chan said. These drugs have significant toxicities and cost, he pointed out, and being able to identify patients who are not likely to benefit will spare patients both.

Some of the authors report that they have a pending patent application related to determinants of cancer response to immunotherapy, and some report receiving personal fees, grant support, and/or nonfinancial support from industry. Details are available at

Nature. Published online November 26, 2014. Comment, Herbst extract, Powles extract, Tumeh extract, Yadav extract, Gubin extract


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.