Fig. 1

TIS scores are associated with response to anti-PD-1 in the CERTIM multi-tumor cohort. a Volcano plot of p-value versus log2 fold change of the differential expression between responders (PR/CR) and non-responders (PD/SD) in the whole cohort. The test for differential expression was done by fitting the log2 normalized count to the response with linear model. The p-values were adjusted by the Benjamini–Hochberg Procedure. Dots corresponding to genes that are significant at FDR < 0.1 are labelled in red. b Boxplot of TIS scores in responders and non-responders (The lower whisker of the responders is not visible as the length is 0). The response was fit to TIS scores with logistic regression and p-value = 0.008, indicating that high TIS scores are predictive of response to anti-PD-1 treatment. The odds ratio is 2.64, 95% confidence interval (1.37, 5.95). c The Kaplan–Meier curves of TIS score groups for the whole patient cohort. The TIS scores were categorized into three groups by tertiles. The Kaplan–Meier curves show that the high TIS score group have higher survival rate than the other two groups (which are combined on the graph into the “low” group). The survival time was fit to TIS score group (high vs low and intermediate) with Cox proportional hazard model. Hazard ratio is 0.374, 95% confidence interval (0.18, 0.76), p-value = 0.005, meaning the high TIS score group has a decrease of the hazard by 63%. d Heatmap showing the individual TIS genes normalized expression, as well as TIS global score, histological subtype and overall response to anti-PD1. NSCLC non small cell lung carcinoma, RCC renal cell carcinoma, SCLC small cell lung carcinoma, nivo nivolumab, pembro pembrolizumab, ORR overall response according to RECIST v1.1, CR complete response, PR partial response, SD stable disease, PD progressive disease