Why do the majority of patients not respond at all, or only partially or transiently, to immunotherapy?

Abstract

The response to immunotherapy is often dramatic but relatively uncommon and until recently difficult to improve upon. The modern era of immunotherapy was arguably ushered in with the discovery of William Coley’s (eponymously named) toxins. The development of cancer vaccines and cytokines, such as Interleukin-2 (IL-2), continued to induce marked clinical responses but only in a minority of patients (rarely above 20%) [1]. The advent of the checkpoint inhibitors (CPIs), such as Ipilimumab (ipi), Nivolumab (nivo), and Pembrolizumab (pembro) have led to responses of up to 40% in melanoma when used as single agents but more importantly increased survival, although at the expense of significant toxicity (a feature also shared with high dose IL-2 based regimens) especially when given in combination. The holy grail of optimizing these treatments has been to find biomarkers that will predict responders to these agents [2]. Such is the depth and volume of research worldwide in this area, there is no shortage of leads. These include the direct assessment of PD-L1 expression on tumor cells as a predictor of response to anti-PD1 antibodies. However, initial enthusiasm for this marker has waned as 10–20% of patients with negative expression will benefit from treatment. As a single biomarker it is not adequate for determining clinical response [3]. The importance of tumor infiltrating lymphocytes (TILs) has been recognized since Rosenberg and colleagues sought to improve the response rate to high dose IL-2. It is now well recognized that a dense TIL is associated with an increased response rate to the CPIs, especially if the density increases following treatment. However, it is not absolute enough to establish a level at which it is not worth treating [4]. Another marker used for patient selection is the mutational or neo-antigen burden produced by somatic mutations. Patients with colorectal cancer with mismatch repair deficient (MMRD) whose mutational burden is 20X more than those without, are many times more likely to respond to CPIs, together with an increased overall survival [5]. Within the tumor, immune response genes (IRG’s), such as gamma interferon, are likely to be over-expressed >2,5 fold in responders. A 28 gene panel associated with clinical benefits in melanoma patients treated with Pembrolizumab has been presented as a model to predict responders [6]. Combinations of these approaches are being studied to increase the specificity and sensitivity of predicting responders