Background: The drug approval and clinical use of immune checkpoint inhibitors (ICIs) were different among de novo and previously treated cancer patients. The purpose of this study is to investigate progression-free survival (PFS) and objective response rate (ORR) as earlier primary endpoints in previously treated advanced or metastatic solid cancers in immunotherapy era. Methods: This systematic review included phase 2 and 3 clinical trials in previously treated advanced or metastatic solid cancers receiving ICIs, through literature search on databases up to 2022. Quality control was performed, where studies with high risk of bias were excluded. Prediction models were first established using randomized controlled trials (RCTs), and then externally validated in the phase 2 non-randomized trials. Trial-level surrogacy analysis was conducted by correlating PFS hazard ratio (HR) and overall survival (OS) HR. Correlation analysis within checkpoint inhibitor arms was performed between ORR, 3-, 6-, and 9-month PFS rates and 12-month OS rate. The correlation was evaluated using the Pearson correlation coefficient r in weighted linear regression, with weight equal to patient size. High, medium strength, and low correlation was defined as r≥ 0.85, 0.7 < r< 0.85, and r≤ 0.7, respectively. Sensitivity analyses were performed to assess the consistency of predictive model by leaving trials of one cancer type out at a time. Results: A total of 64 phase 2 or 3 RCTs with 22,011 patients and 96 phase 2 non-randomized trials with 10,288 patients were included in modeling and validation, respectively. The most common primary endpoint in RCTs and non-randomized trials was OS (38 [59%]) and ORR (72 [75%]). The mostly used endpoint assessment criteria were RECIST v1.1 in both RCTs (59 [92%]) and non-randomized trials (88 [92%]). In trial-level surrogacy, PFS HR showed a low correlation with OS HR (r, 0.52; 95% confidence interval [CI], 0.23–0.78). Within the checkpoint inhibitor treatment arms in RCTs, there was a high correlation between 6-month PFS and 12-month OS (r, 0.85; 95% CI, 0.76–0.90), and 9-month PFS and 12-month OS (r, 0.85; 95% CI,0.80–0.90), with a medium strength correlation between 3-month PFS and 12-month OS (r, 0.79; 95% CI, 0.65–0.89) and a low correlation between ORR and 12-month OS (r, 0.13; 95% CI, 0.10–0.76). In external validation, when 6-month PFS was used to predict 12-month OS, there was a good calibration between actual and predicted 12-month OS. Sensitivity analysis demonstrated reasonable overall consistency. Conclusions: For phase 2 non-randomized, observational checkpoint inhibitor trials enrolling previously treated advanced or metastatic solid cancers, we recommended 6-month PFS rate as primary surrogate endpoint. For RCTs, gold standard endpoint OS should be persistently used and should not be replaced by earlier endpoints PFS or ORR.