A whole-blood RNA transcript-based gene signature is associated with the development of CTLA-4 blockade-related diarrhea in patients with advanced melanoma treated with the checkpoint inhibitor tremelimumab

Abstract Background Anti-CTLA-4 immune checkpoint blockade is associated with immune-related adverse events (irAEs). Grade 3–4 diarrhea/colitis is the most frequent irAE requiring treatment discontinuation. Predicting high-risk diarrhea/colitis patients may facilitate early intervention, limit irAE...

Full description

Bibliographic Details
Main Authors: Philip Friedlander, Kevin Wood, Karl Wassmann, Alan M. Christenfeld, Nina Bhardwaj, William K. Oh
Format: Article
Language:English
Published: BMJ Publishing Group 2018-09-01
Series:Journal for ImmunoTherapy of Cancer
Subjects:
Online Access:http://link.springer.com/article/10.1186/s40425-018-0408-9
Description
Summary:Abstract Background Anti-CTLA-4 immune checkpoint blockade is associated with immune-related adverse events (irAEs). Grade 3–4 diarrhea/colitis is the most frequent irAE requiring treatment discontinuation. Predicting high-risk diarrhea/colitis patients may facilitate early intervention, limit irAE severity, and extend treatment duration. No biomarkers currently predict for anti-CTLA-4 immunotherapy related severe diarrhea. Methods Whole-blood was collected pre-treatment and 30 days post-treatment initiation from patients with stage III or IV unresectable melanoma who received 15 mg/kg tremelimumab at 90 day intervals in two clinical trials. The discovery dataset was a phase II study that enrolled 150 patients between December 2005 and November 2006. The validation dataset was a phase III study that enrolled 210 patients between March 2006 and July 2007. RT-PCR was performed for 169 genes associated with inflammation, immunity, CTLA-4 pathway and melanoma. Gene expression was correlated with grade 0–1 versus grade 2–4 diarrhea/colitis development. Results Pre-treatment blood obtained from the discovery dataset (N = 150) revealed no gene predictive of diarrhea/colitis development (p < 0.05). A 16-gene signature (CARD12, CCL3, CCR3, CXCL1, F5, FAM210B, GADD45A, IL18bp, IL2RA, IL5, IL8, MMP9, PTGS2, SOCS3, TLR9 and UBE2C) was identified from 30 days post-tremelimumab initiation blood that discriminated patients developing grade 0–1 from grade 2–4 diarrhea/colitis. The 16-gene signature demonstrated an AUC of 0.814 (95% CI 0.743 to 0.873, p < 0.0001), sensitivity 42.9%, specificity 99.2%, positive predictive value (PPV) 90.0%, and negative predictive value (NPV) 91.4%. In the validation dataset (N = 210), the 16-gene signature discriminated patients developing grade 0–1 from grade 2–4 diarrhea/colitis with an AUC 0.785 (95% CI 0.723 to 0.838, p < 0.0001), sensitivity 57.1%, specificity 84.4%, PPV 57.1% and NPV 84.4%. Conclusion This study identifies a whole-blood mRNA signature predictive of a clinically relevant irAE in patients treated with immune checkpoint blockade. We hypothesize that immune system modulation induced by immune checkpoint blockade results in peripheral blood gene expression changes that are detectable prior to clinical onset of severe diarrhea. Assessment of peripheral blood gene expression changes in patients receiving anti-PD-1/PD-L1 immunotherapy, or combination anti-CTLA4 and anti-PD-1/PD-L1 immunotherapy, is warranted to provide early on-treatment mechanistic insights and identify clinically relevant predictive biomarkers. Trial registration Clinicaltrials.gov, NCT00257205, registered 22 November 2005
ISSN:2051-1426