Predicting the Response to Checkpoint Inhibitors

Checkpoint events involve recognition by immune system cells of tumor antigens, primarily through pathways involving cytotoxic T lymphocyte antigen 4 (CTLA4) or the programmed cell death 1 (PD1) receptor.

Checkpoint inhibitors (CPIs) are monoclonal antibody (mAb) drugs that interfere, through either stimulatory or inhibitory mechanisms, with the immune response to cancer. By blocking the interactions between PD1/CTLA4 ligands on the tumor and the receptors to those ligands on T cells, CPIs prevent the tumor from “turning off” the immune response that would normally occur.

CPI drugs are used to treat various solid tumors, particularly those that are unable to correct errors in DNA copying. These include malignancies of the lung, liver, skin, and stomach, as well as some blood cancers. The U.S. Food and Drug Administration has approved eight CPI drugs, seven of which target the PD1/PDL1 pathway.

Ipilimumab (Yervoy^®), which works through the CTLA4 pathway, was the first CPI drug approved (in 2011) and remains in many ways the model for current development and approvals. Ipilimumab, a humanized mAb, provides significant survival benefit for a variety of tumors, both as a monotherapy and through co-administration with other drugs. Early studies suggested an improved overall survival of 10 months among metastatic melanoma patients treated with ipilimumab plus the glycoprotein GP100, compared with 6.4 months for those taking GP100 alone.

When they work, CPI agents add many months or years to a patient’s life, but they are not magic bullets.

As noted by the Johns Hopkins School of Medicine, “immunotherapy drugs work better in some cancers than others, and while they can be a miracle for some, ...[o]verall response rates are about 15% to 20%” and “[w]e don’t know who will respond best.”

CPI drugs are associated with numerous side effects, particularly those associated with hyper-stimulation of the immune system. More than half of those receiving these agents develop serious treatment-related conditions, and approximately 1% of ICI-treated patients  die.

Given the cost of CPI ($100,000 to $1 million per course of treatment), plus the resources involved and the personal cost to individuals undergoing therapy, identifying patients with a high probability of responding (or of being harmed) is a top priority for drug developers.

Desperately seeking biomarkers

Scientists have identified numerous factors affecting the success of CPI therapy. Unfortunately, individual response biomarkers used to monitor how well the drugs are working, which include DNA mutation and neoantigen loads, immune profiles, and PDL1 expression, are weak predictors of the overall response to treatment. Researchers therefore have attempted to combine these factors with host germline genetics, the effects of prior cytotoxic treatment(s), and features of the tumor microenvironment and the gut microbiome into a predictive model. In addition, researchers are examining biomarkers of toxicity both to lessen side effects and to understand better the relationship between toxicity and response.

This work is at its very earliest stages, so one should not put too much stock in the utility of biomarker types that have previously shown poor correlation. For example, tumor genetics alone are generally poorly predictive, but the discovery of dysfunctional DNA methylation via the 5mC (methylation) and 5hmC (hydroxymethylation) pathways may eventually serve as reliable prognostic and predictive biomarkers.

Tumor neoantigens, the “foreign” proteins expressed only by tumor cells, are on the surface the ideal biomarker since normal cells do not express them. Emerging evidence suggests these proteins mediate the tumor-specific T-cell-mediated antitumor responses and might be utilized, either through therapeutics or diagnostics, to improve prospects for patients identified as candidates for CPI treatment. Research has shown that patients expressing a high degree of tumor mutational burden or tumor neoantigen burden often respond well to CPI therapy.

Patients undergoing CPI treatments have typically undergone previous cytotoxic regimens, in other words classic chemotherapies. These treatments, which include methylating agents, disruptors of microtubule formation, antimetabolites, topoisomerase inhibitors, and cytotoxic antibiotics, have been used for decades with little or no regard to their immune-modulating activity. Recent studies suggest they may also contribute to the restoration of immunosurveillance and the initiation or maintenance of immunogenic cell death. Both mechanisms support CPI therapies, but applying this knowledge to predicting the success of immunotherapies will require the validation of appropriate treatment-associated biomarkers.

Special role of the TME

The tumor microenvironment (TME) has been implicated in nearly every aspect of cancer etiology and progression, and the success or failure of treatments. The TME consists of various tumor, structural, supportive, and immune system cells that are in constant chemical communication. Because of its biological and biochemical complexity, TME factors driving the success or failure of CPI therapy will be difficult to uncover and validate, but researchers are trying nonetheless.

The first thing one notices about the TME is the presence or preponderance of infiltrating tumor-fighting cells such as cytotoxic T lymphocytes, type 1 helper CD4+ lymphocytes, and natural killer (NK) cells, which participate in the elimination and equilibrium phases of immunosurveillance. Elimination refers to the removal of cancer cells by the immune system, while equilibrium is the state where cells that have escaped elimination take hold and expand.

During the escape stage, when cells finally overcome the host’s immune system to form tumors, levels of infiltrating lymphocytes and NK cells fall, while populations cells that suppress immunity, such as regulatory T cells and immunosuppressive myeloid cells, rise.

Given the complexity of cancer and individual responses to treatments, the discovery of a magic bullet biomarker that accurately predicts a patient’s response to CPI therapy is unlikely. Yet, considering the high cost of these treatments, the resources involved, and the human costs, a great need exists for biomarker panels that accurately predict therapeutic responses.

While the cost and benefits of CPI treatments justify the application of complex, time-consuming, and expensive assays, validating such testing regimens will be extremely difficult. The answer may therefore lie in a much simpler (and familiar) analytic platform, metabolomics.

A recent study suggested that a metabolomic approach could work for predicting the response to nivolumab among patients treated for non-small cell lung cancer. Although the authors claim an 80% success rate in predicting positive treatment responses, the article has been cited only four times in the past ten months. Metabolomics has the advantage, compared with assaying multiple cell types and signaling pathways, of requiring very little sample preparation and a unified analytic platform, mass spectrometry.

Immunotherapies, particularly CPI treatments, have shown great potential but in their current form benefit fewer than 20% of patients. Investigators have identified dozens of potential biomarkers that can be used to explain, and in some instances predict, responses to treatment. However, given the heterogeneity of both cancer and immunity, prediction based on individual biomarkers or biomarker panels will probably be limited to the specific situations in which they are studied. Meanwhile, immunotherapy awaits the discovery of universal or generalized biomarkers that may be more broadly applied to most or all CPI drugs, targets, and prospective patients.