Sun R, et al. - In patients included in phase 1 trials of anti-programmed cell death protein (PD)-1 or anti-programmed cell death ligand 1 (PD-L1) monotherapy, researchers developed and independently validated a radiomics-based biomarker of tumor-infiltrating CD8 cells. They also assessed the correlation between the biomarker, and tumor immune phenotype and clinical outcomes of these patients. In three independent cohorts, the radiomic signature of CD8 cells was validated. A promising way to predict the immune phenotype of tumors and to infer clinical outcomes for patients with cancer who had been treated with anti-PD-1 and PD-L1 were provided by this imaging predictor.

Methods

• In this retrospective multicohort study, four independent cohorts of patients with advanced solid tumors to develop and validate a radiomic signature predictive of immunotherapy response by combining contrast-enhanced CT images and RNA-seq genomic data from tumor biopsies to assess CD8 cell tumor infiltration.
• In order to develop the radiomic signature of CD8 cells, they used the CT images and RNA sequencing data of 135 patients with advanced solid malignant tumors enrolled into the MOSCATO trial between May 1, 2012 and March 31, 2016, in France (training set).
• They used the genomic data, which are based on the CD8B gene to estimate the abundance of CD8 cells in the samples and data were then aligned with the images to generate the radiomic signatures.
• Cancer Genome Atlas [TGCA] database dataset including 119 patients who had available baseline preoperative imaging data and corresponding transcriptomic data on June 30, 2017 validated the concordance of the radiomic signature (primary endpoint).
• A radiomics-based predictor of the CD8 cell expression signature was built by use of machine learning (elastic-net regularized regression method) from 84 input variables used for the machine-learning method (78 radiomic features, five location variables, and one technical variable).
• They used two other independent cohorts of patients with advanced solid tumors to evaluate this predictor.
• They randomly selected the immune phenotype internal cohort (n=100) from the Gustave Roussy Cancer Campus database of patient medical records based on previously described, extreme tumor-immune phenotypes: immune-inflamed (with dense CD8 cell infiltration) or immune-desert (with low CD8 cell infiltration), irrespective of treatment delivered; these data were used to analyze the correlation of the immune phenotype with this biomarker.
• Finally, they used the immunotherapy-treated dataset (n=137) of patients recruited from December 1, 2011 to January 31, 2014, at the Gustave Roussy Cancer Campus, who had been treated with anti-PD-1 and anti-PD-L1 monotherapy in phase 1 trials to assess the predictive value of this biomarker in terms of clinical outcome.

Results

• As per data, experts developed a radiomic signature for CD8 cells that included eight variables, which was validated with the gene expression signature of CD8 cells in the TCGA dataset (area under the curve [AUC]=0.67; 95% CI 0.57–0.77; p=0.0019).
• The signature, in the cohort with assumed immune phenotypes, was also able to discriminate inflamed tumors from immune-desert tumors (0.76; 0.66-0.86; p < 0.0001).
• Findings suggested that in patients treated with anti-PD-1 and PD-L1, a high baseline radiomic score (relative to the median) was related to a higher proportion of patients who achieved an objective response at 3 months (vs those with progressive disease or stable disease; p=0.049) and a higher proportion of patients who had an objective response (vs those with progressive disease or stable disease; p=0.025) or stable disease (vs those with progressive disease; p=0.013) at 6 months.
• Results demonstrated an association of a high baseline radiomic score with improved overall survival in univariate (median overall survival 24.3 months in the high radiomic score group, 95% CI 18.63–42.1; vs 11.5 months in the low radiomic score group, 7.98-15.6; hazard ratio 0.58, 95% CI 0.39–0.87; p=0.0081) and multivariate analyses (0.52, 0.35–0.79; p=0.0022).