The Nomogram Based on Multiple Sequence Magnetic Resonance Imaging (MRI) of PD-1 and CTLA-4 to Predict the Long-Term Survival Rate of NPC Patients
Abstract
Purpose
The study aims to develop and validate a radiomics model using multiple sequence (MS) -MRI to predict the OS rate in patients diagnosed with NPC.
Methods
A total of 124 consecutive NPC patients treated from March 2013 to June 2021 were included in this study. Clinical risk factors associated with NPC of enrolled patients, including age, sex, smoking, TNM stage, and clinical stage before treatment. In addition, the immunological indexes of hemoglobin, albumin, globulin, white ball ratio, lactate dehydrogenase, cytokeratin 19 and squamous cell carcinoma antigen were included.Radiomics features were extracted using a radiomics package based on the Embedded Radiomics Computing module, which enables feature calculation in AccuContour software (version 3.0, Management Medical Technologies, Ltd., Xiamen, China).
Results
13 Rad scores and 13 nomograms were developed for risk assessment and to predict the prognosis of NPC patients. We found that CD4 + PD-1 + (P=0.022) was an independent predictor of overall survival (OS) in NPC patients, and that RS-T1 was significantly associated with the expression of CD8 + CTLA-4 + (P=0.029). Combining CD4 + PD-1 + with clinical factors and combined imaging omics, the resulting model had an accuracy of 0.965, accuracy of 0.923, ROC of 0.988, and 0.810, accuracy of 0.921 and accuracy of 0.909 in the validation group, and the results were better than the other 12 models.
Conclusion
Compared with other imaging omics, other immune cell or immune checkpoint indicators, and clinical line maps, combining imaging omics with CD4 + PD-1 +, CD8 + CTLA-4 + with clinical factors has the best discriminatory power and has higher clinical utility.