What Has the Checkpoint Inhibitor Experience in Melanoma Taught Us About Immunotherapy for Other Cancers?

What Has the Checkpoint Inhibitor Experience in Melanoma Taught Us About Immunotherapy for Other Cancers?


Dr. Jeffrey S. Weber

By Jeffrey S. Weber, MD, PhD

Article Highlights

  • Lessons learned from patients with melanoma will help accelerate the development of checkpoint inhibition for other cancers. These lessons may also suggest new strategies for treating patients with immunotherapy-resistant cancers, such as prostate, pancreatic, and non–microsatellite instability high colon cancers.
  • The existence of the “tail on the curve” of survival in melanoma, and now other cancers, suggests that some responders to checkpoint inhibition may be cured or at least have long-term freedom from progression of their disease and do not need to be treated until progression.
  • Successful prediction of outcome to checkpoint protein inhibition will likely require an amalgamated biomarker that combines tumor cell intrinsic and host T-cell–specific determinants.

Since the checkpoint inhibitors ipilimumab and tremelimumab were first tested in patients with melanoma in 2002, the field of immunotherapy for cancer has exploded with hundreds of new trials and an increasing presence in the developmental therapeutics oncology field. Ipilimumab, nivolumab, and pembrolizumab have become mainstays of treatment for metastatic melanoma, but a far more important clinical development is the U.S. Food and Drug Administration (FDA) approval of three different anti–PD-1/–PD-L1 agents for non–small cell lung, renal cell, head and neck, and bladder cancers, as well as Hodgkin lymphoma. Additional approvals in different tumor types are not far off, and numerous combination trials are underway in an effort to optimize the use of checkpoint inhibition. Lessons learned from patients with melanoma will undoubtedly help accelerate the development of checkpoint inhibition for other cancers and may suggest new strategies for treating patients with immunotherapy-resistant cancers, such as prostate, pancreatic, and non–microsatellite instability high colon cancers.

Kinetics of Response

Early in the development of checkpoint inhibitors for melanoma, their unique kinetics of response became apparent. Patterns of response included slow regression over 6 to 12 months, mixed responses with subsequent regression, and progression followed by regression. Although these unusual patterns were observed in up to 10% of patients, they raised the issue of how long to keep treating after what seemed like RECIST progression of disease, or whether to keep treating in the face of a mixed response. These issues brought about the use of the modified immune response, or irRECIST, criteria. Under irRECIST criteria, progression of disease for patients without clinical or biochemical deterioration required confirmation with a repeat short-interval scan. For these patients, greater tolerance was allowed for an increase in tumor burden before progression of disease was declared.1,2 These concepts will need to be applied to other histologies treated with checkpoint inhibition.

Duration of Response

A key issue in the cancer immunotherapy field is how long to treat patients with checkpoint inhibitors. Early trials in melanoma allowed treatment until progression; then, 2 years was the maximal duration, and several reports have examined the outcome in patients who achieved a response and stopped therapy after 2 to 3 years.3,4 In most cases, patients who achieved a complete response and subsequently stopped therapy maintained their remission, and many of those who did not achieve a complete response and whose disease progressed were able to respond to further immunotherapy. Similar data have been observed for those who achieved a partial remission. The existence of the “tail on the curve” of survival in melanoma, and now other cancers, suggests that many responders to checkpoint inhibition may be cured or at least have long-term freedom from progression of their disease and do not need to be treated until progression. A duration of not less than 1 year and not more than 2 years seems reasonable.

Immune-Related Toxicities

The unique pattern of side effects observed with ipilimumab and noted with PD-1/PD-L1 inhibitors presents a challenge for physicians who are inexperienced with the use of these drugs. Algorithms have been established for the successful management of these immune-related adverse events, which are mechanism related and directly tied to breaking tolerance as a mode of action of checkpoint inhibition.5-7 In virtually all cancers in which these drugs are used, specific patterns of side effects will be observed based on the combination regimens used and the unique pathophysiology of each cancer. There are some data suggesting that the onset of overall immune-related adverse events is associated with better outcomes, but it may be the need for immune modulators such as steroids and anti–tumor necrosis factor antibodies (e.g., infliximab) that are actually associated with benefit. In patients with lung cancer, pneumonitis is more common with PD-1 blockade with or without CTLA-4 blockade than in patients with melanoma, and vitiligo and other skin effects may be more common in melanoma than in other cancers. Key lessons to take away are that clinicians must have a low threshold for ruling out endocrinopathies with non-specific symptoms of fatigue, that patients may benefit from the occasional use of short steroid regimens to manage grade 2 side effects, and that clinicians must be aware of the need for longer-term steroid regimens to manage grade 3 to 4 immune-related side effects.

Biomarkers

The most vexing question in the field of checkpoint inhibition is whether biomarkers can be defined that predict regression from the use of these drugs and that allow practitioners to choose patients who are most likely to respond to them. In patients with melanoma or non–small cell lung cancer, there have been a number of studies suggesting that tumors can be divided into three categories: those that are infiltrated with T cells and tend to have an “inflammatory” or “hot” profile of tumor gene expression, tumors that are devoid of any T cell or inflammatory infiltrate on histology and have a non-inflamed or “cold” profile, and tumors that have T cells and other immune cells at the periphery but not within the tumor.8,9 The “hot” tumors are the ones most likely to respond to PD-1/PD-L1 blockade and have been primed but have T cells with high levels of PD-1. Many studies have evaluated the role of PD-L1 tumor and/or immune cell immunohistochemical staining and its association with outcome with PD-1/PD-L1 blockade. Although most studies are in agreement that the higher the level of membranous tumor PD-L1 the better the outcome with PD-1/PD-L1 blockade, it is clear that patients whose tumors stain PD-L1 negative may still gain benefit from checkpoint inhibition.10 This negates the utility of PD-L1 to choose patients for therapy because it is unable to define those who should not be treated.

The number of tumor-infiltrating CD8+ T cells expressing PD-1 and/or CTLA-4 appears to be a key indicator of success with checkpoint inhibition, and both PD-1 and CTLA-4 blockade may increase the infiltrating T cells.11 A “focused” T-cell receptor repertoire is associated with a good outcome with PD-1/PD-L1 blockade,8 whereas a more diverse repertoire is associated with benefit from CTLA-4 antibodies.12 The T-cell repertoire reflects the host immune response to cancer, but the tumor itself is a key determinant of success with checkpoint inhibition because there is a relationship between increased nonsynonymous variants or mutations in tumors and outcome with checkpoint inhibition. The number of neoantigens, or mutated proteins that are expressed and could be recognized as an antigen by T cells, which is related to the total mutational burden, is critically associated with outcome for checkpoint inhibitors.13-17

The nature of the tumor microenvironment also plays an important role in resistance or susceptibility to checkpoint inhibition. A tumor gene expression signature that reflects a series of interferon-gamma–inducible genes may define a “hot” inflamed tumor and is associated in several studies with a good outcome with checkpoint inhibition; its loss is associated with resistance to ipilimumab therapy.18-20 Melanomas that are class II major histocompatibility complex positive respond to PD-1/PD-L1 blockade and may share the interferon-gamma responsive signature.21 In contrast, resistant tumors display a transcriptional signature (called the IPRES or innate anti–PD-1 resistance), which is associated with increased expression of genes involved in the regulation of the epithelial-mesenchymal transition, cell adhesion, extracellular matrix remodeling, angiogenesis, and wound healing.18

Deletion of the PTEN gene, commonly found in melanoma, has a deleterious effect on antitumor immunity with checkpoint inhibition and leads to a “cold” tumor with high levels of immune-suppressive cytokines with sparse and inactive T cells.22 There is also an association between tumor activation of the WNT/β-catenin signaling pathway and absence of a T-cell gene expression signature, which leads to deficiencies of infiltrating dendritic cells and a “cold” tumor microenvironment.23 This might be overcome with the use of a STING agonist, which can augment expression of interferon-gamma pathway genes.24 Clinical examination of host biomarkers from large clinical trials of PD-1 blockade has shown that neutrophil to lymphocyte rations, baseline lactate dehydrogenase, and eosinophil numbers are associated with outcome to checkpoint blockade, although none of these markers can reliably define a patient who will not benefit from treatment.25,26

In tumors that exhibit baseline resistance to checkpoint inhibition or that develop adaptive resistance to therapy after an initial response, the induction of tumor JAK1 and JAK2 mutations or deletion of β-2 microglobulin may be responsible, leading to impaired T-cell immunity and inability to detect tumor antigens.27,28

There will, undoubtedly, be common pathways of innate and adaptive resistance to checkpoint protein inhibition across many different tumor types. Successful prediction of outcome to these drugs will require an amalgamated biomarker that combines tumor cell intrinsic and host T-cell–specific determinants.  

About the Author:​ Dr. Weber is deputy director of the Laura and Isaac Perlmutter Cancer Center and a professor of medicine at the NYU-Langone Medical Center in New York.