Background: Few studies have focused on the subclinical cardiotoxicity of immune checkpoint inhibitors (ICIs) in cancer patients. This study aimed to evaluate the manifestations of subclinical cardiotoxicity of ICI therapy using cardiovascular magnetic resonance (CMR) and to explore whether CMR parameters can help predict cardiotoxicity at the early stage of ICI therapy.Methods: A prospective, longitudinal study was conducted among patients with lung cancer. The patients were planned to undergo serial CMRs before (baseline), 3 weeks after (1st follow-up), and 3 months after (2nd follow-up) the initiation of ICI therapy, respectively. Patients with 3 CMRs were included in the analysis. Serial quantitative measurements based on CMR were compared using one-way repeated measures analysis of variance (RM-ANOVA). On the basis of cancer therapy-related cardiac dysfunction (CTRCD) observed at the second follow-up, patients were categorized into a CTRCD group and a non-CTRCD group. Baseline clinical and CMR parameters and the relative reduction of left ventricular global strain at the second follow-up was compared between the CTRCD group and the non-CTRCD group. Receiver operating characteristic (ROC) analysis was used to identify CTRCD that developed 3 months after ICI therapy.Results: A total of 36 patients with 3 CMRs (60.7 +/- 9.2 years old, 77.8% male) were included in the analysis. Left ventricular-global radial strain (LV-GRS) decreased significantly at the second follow-up (37.9%+/- 8.5% vs. 33.1%+/- 1.0%; P=0.014), but left ventricular ejection fraction (LVEF) did not change significantly (51.5%+/- 6.0% vs. 49.2%+/- 6.5%; P>0.05). A total of 7 patients (19.4%) had developed CTRCD by the second follow-up. Baseline clinical and CMR parameters did not differ between the CTRCD group and the non-CTRCD group (P>0.05 for all). In the CTRCD group, the left ventricular-global circumferential strains (LV-GCSs) showed significant reductions at both the first and second follow-up (P=0.008 and 0.035, respectively), but the LVEF only showed a significant reduction at the second follow-up (P<0.001). The relative reduction of LV-GRS at the second follow-up was significantly higher in the CTRCD group than in the non-CTRCD group (29.8%+/- 25.8% vs. 6.8%+/- 20.4%; P=0.036) and was used to predict CTRCD developed at the 3-month timepoint after ICI therapy [area under the curve (AUC) =0.759; P=0.036].