Longitudinal Deep immunophenotyping in secondary anti-PD1 resistance – PI: Mark Linder

The introduction of immune checkpoint inhibitors (ICIs) which act through modulation of the host cytotoxic immune responses have dramatically improved durable response in the treatment of advanced melanoma. However, a majority of patients develop ICI resistance (1) . The literature reflects two divergent hypotheses regarding the development of secondary ICI resistance (2). One emphasizes changes in the localized interaction between the tumor and responding immune cells, while the other emphasizes immunophenotypic exhaustion, measured as increased expression of co-inhibitory modulators (LAG3, B2M, Tim-3 etc.) (3-5). In preliminary studies of patients with multiple distant metastatic lesions, who progressed while receiving anti-PD-1 therapy, we observed simultaneous progression of each independent lesion. These data support the hypothesis that secondary resistance to anti-PD-1 therapy is driven by changes in the host immune response. Prior studies to date have revealed evidence of immunophenotypic traits (e.g. expression of LAG3, B2M, Tim-3 by CD8+ PD1+ Ki67 + T cell sub-populations)(6) associated with ICI resistance (7-10). However, these studies have failed to demonstrate conclusive evidence of causal immunophenotypic changes due to overlapping expression patterns of discriminatory factors between radiographically defined responders and non-responders. We anticipate that these studies have been compromised by multiple confounding factors including: a) failure to distinguish between primary versus secondary ICI resistance, b) inclusion of both PD-1 and CTLA-4 directed therapies which differ in their mechanism of action, c) focusing on a limited number of immunophenotypic features on restricted immune cell lineages, and most importantly d) reliance on radiographically defined response (11). Our group and multiple others (12-17) demonstrate that increasing plasma circulating tumor DNA (ctDNA) concentrations reflect biological tumor progression weeks to months prior to radiographical progression. Consequently, longitudinal studies to identify causal immunophenotypic features associated with acquired resistance, that rely on radiographic response classification, are subject to misclassification of response status. Most notably, misclassification of status as responsive to immunotherapy, when in fact biological therapeutic non-response (not detected by imaging) has already developed. We hypothesize that secondary anti-PD-1 resistance is driven by compensatory host immune factors that diminish the effectiveness of PD-1/PD-L1 blockade. Profiling of plasma ctDNA to define anti-PD-1 resistance will enable a longitudinal study design which avoids confounding due to mis-classification of response status.