Relationships between intravoxel incoherent motion parameters and expressions of programmed cell death-1 (PD-1) and...

To determine the associations of intravoxel incoherent motion (IVIM) parameters with expression of programmed cell death-1 (PD-1) and programmed cell death ligand-1 (PD-L1), and evaluate the performance of the combined model established based on IVIM and clinicopathological parameters in predicting PD-L1and PD-1 status of cervical cancer (CC) patients.

Seventy-eight consecutive CC patients were enrolled prospectively and underwent magnetic resonance imaging (MRI) including IVIM. IVIM quantitative parameters were measured, compared, and correlated with PD-L1 and PD-1 expression. Independent factors related to PD-L1 and PD-1 positivity were identified and were used to establish the combined model. The combined model's diagnostic performance was evaluated using the receiver operating characteristic (ROC) analysis. The Shapley additive explanation (SHAP) algorithm was used to explain the contribution of each parameter in the combined model.

The real diffusion coefficient (D) value was significantly lower in the PD-L1-positive group than in the PD-L1-negative group (0.64 ± 0.12 versus 0.72 ± 0.11, p=0.021). The PD-1-positive and PD-1-negative groups showed similar trends (0.63 ± 0.13 versus 0.73 ± 0.09, p=0.003). Parametrial invasion, lymph node status, pathological grade, FIGO (International Federation of Gynecology and Obstetrics) staging, and D values were independently associated with PD-L1 and PD-1expression. A combined model incorporating these parameters showed good discrimination with the sensitivity, specificity of 90.9%, 82.6% for PD-L1, and 93.5%, 72% for PD-1. According to the SHAP value, FIGO staging and pathological grade were the most influential features of the prediction model.

IVIM parameters were found to correlate with PD-L1 and PD-1 expression. The combined model, incorporating parametrial invasion, lymph node status, pathological grade, FIGO staging, and D values, showed good discrimination in predicting PD-L1 and PD-1 status, providing the basis for CC immunotherapy.