Identifying biomarkers to guide immunotherapy treatment of cancer of unknown primary

Abstract

Cancer of unknown primary (CUP) is a metastatic cancer where the primary tissue of origin (ToO) evades detection despite extensive clinical investigation. Recent approaches to the management of CUP have involved gene-expression profiling to resolve the likely ToO and mutational profiling to identify targeted treatments. Current evidence is that both of these methods can have some clinical benefit but only in a minority of cases. Immunotherapy may also be effective in CUP, but clinical trials looking at this are yet to report and the best way to select patients who may benefit is uncertain. Methods: Through a national cohort study called Solving Unknown Primary Cancer (SUPER), 245 CUP patients had molecular profiling performed and clinical follow-up. Variant detection from DNA sequencing using a targeted panel of 386 genes (SureSelect Agilent) was used to calculate tumour mutation burden (TMB). Additionally, a custom gene expression assay including 225 genes associated with ToO and 35 immune genes (SUPERDx, nCounter NanoString) was used to classify the likely ToO and immune profile the tumours. We used the genomic data from 217 of these patients to explore interdependencies between immune gene-expression patterns and TMB. We also retrospectively reviewed the outcomes of patients treated with anti-PD1/PD-L1 immune checkpoint inhibition (ICI) therapy in whom genomic data was available. Results: TMB was weakly correlated with a gene-expression score for immune response (IR) based on averaged z-score values of six genes (CD8A, IFNG, PRF1, PD-L1, GZMA, GZMB) (Spearman rho = 0.21) with many outliers exhibiting low TMB but high IR scores. Among twenty patients treated with ICIs, with both molecular data and response data available, 6 had a partial response (PR), 5 had stable disease (SD) and 9 had progressive disease (PD). We found that an elevated IR score was more predictive of ICI response than TMB. Most patients with either PR or SD to ICI treatment had a high IR score (9/10), and most patients with PD had low IR scores. However, only 3/10 patients with TMB available and a PR or SD had a high TMB ( > 10). Conclusions: Our results demonstrate that ICI therapy may benefit some CUP patients and that an immune response gene-expression scoremay predict ICI treatment response.