Identifying a composite signature for predicting immune-related adverse events in advanced melanoma patients treated with immune checkpoint blockade
Authors
- 1. BostonGene, Corp., Waltham, MA
- 2. Massachusetts General Hospital, Boston, USA
Abstract
While immune checkpoint blockade (ICB) produces long-lasting remission in some patients, they may cause serious immune related adverse events (irAEs) that can result in morbidity and mortality. irAEs have been increasingly recognized to stem from autoimmune-like activation of the adaptive immune system. Here, we analyzed pre-treatment peripheral blood mononuclear cells (PBMCs) from patients with advanced melanoma treated with ICB to uncover processes underlying irAE occurrence and if irAE onset can be predicted.
Methods:
Pre-treatment PBMCs from patients with advanced melanoma were profiled using flow cytometry and bulk RNA sequencing. We developed a cellular irAE signature using 17 peripheral immune cell populations selected from open-source data. Principal component analysis was then used to evaluate these populations in samples from the melanoma cohort. Additionally, we developed a gene-based irAE signature from 55 reported irAE-associated genes using ssGSEA for gene signature calculation.
Results:
Pre-treatment PBMCs from a melanoma cohort (n=47) treated with either anti-PD-1 (pembrolizumab; n=23) or anti-PD-1 plus anti-CTLA-4 (nivolumab + ipilimumab; n=24) ICB were analyzed. The severe irAE incidence was 20% (10/47). Patients with severe irAEs had significantly different cellular signature (PC2) at baseline from those without irAEs (ROC-AUC=0.78, p=0.01; Figure 1). In the melanoma cohort, analysis of pre-treatment PBMCs showed that the irAE gene signature was differentially expressed in patients with and without irAEs (ROC-AUC=0.76, p=0.01; Figure 2). Finally, an independent differential cell population analysis of the melanoma cohort identified cellular clusters that distinguished between patients with and without irAEs (Figure 3). Interestingly, patients who developed severe irAEs post-ICB treatment presented with higher frequencies of memory and effector CD4+ and CD8+ T cells, which overlap with populations in our cellular irAE signature.
Conclusions:
We constructed complementary cellular and gene expression-based irAE signatures via a supervised approach using open-source datasets. These signatures, coupled with differential cell population analysis, identified distinct cellular and gene expression patterns in pre-treatment PBMCs among advanced melanoma patients who later developed severe irAEs post-ICB treatment. As such, our findings indicate the potential clinical utility of these signatures in identifying patients with a higher probability of developing irAEs, allowing physicians to mitigate these side effects in a timely manner to improve patient outcomes. Future studies will investigate why and how irAEs occur, based on signatures presented in patients receiving ICB treatment.
Latest publications
