Skip to main content
. 2022 Jun 4;72(1):23–37. doi: 10.1007/s00262-022-03221-5

Fig. 6.

Fig. 6

A Patients were stratified into response groups based on RECIST criteria (PR, SD, and PD), duration of OS, and duration of PFS. “Long-term survival with no durable clinical benefit (NDB_LS)” was defined as patients who achieved NDB with ICI therapy (OS > 1 year) (n = 37). An additional cohort of patients who achieved short-term survival (OS < 1 year) after ICI treatment with early tumor progression (PFS < 6 months) was considered separately (n = 16). B Pie charts demonstrating the proportions of NDB_LS patients. C Waterfall plot showing the correlation of BICS with clinical status (NDB_LS, NDB_SS, and DCB). BICS: peripheral-blood-immune-cell-based signature; DCB: durable clinical outcomes; NDB_LS: long-term survival with no durable clinical benefit; NDB_SS: short-term survival with no durable clinical benefit; NDB: no durable benefit; PD: progressive disease; PR: partial response; SD: stable disease; RCC: renal cell carcinoma; ESCC: esophageal squamous cell carcinoma; NSCLC: non-small cell lung cancer; HCC: hepatocellular carcinoma. D Bar plot showing the AUC of each model for classifying NDB versus NDB_LS& DCB in both the training and validation cohorts. E The accuracy and kappa generated by bootstrapping for classifying NDB versus NDB_LS& DCB using the SVM-RFE, RFB, elastic net, and BICS models. BICS: peripheral-blood-immune-cell-based signature; SVM-RFE: support vector machine-recursive and feature elimination; RFB: random forest and Boruta