Share：
Share this content in WeChat
X
[Chinese] [PDF] 13 1
• Clinical Article •
Conventional clinicopathological features combined with MRI-based radiomics model for predicting programmed death-ligand 1 expression
SUN Xin  QIN Fengying  TIAN Mingke  WEI Yuze  GAO Xiaozhuo  DONG Yue 

DOI:10.12015/issn.1674-8034.2026.01.012.


[Abstract] Objective To establish a combined model based on MRI-based radiomics features and clinicopathological characteristics for evaluating the programmed death-ligand 1 (PD-L1) level in cervical cancer.Materials and Methods A retrospective analysis was conducted on 327 cervical cancer patients who underwent MR enhanced scans at Liaoning Cancer Hospital & Institute, from January 2021 to September 2024. The samples were randomly divided into a training set (n = 228) and a validation set (n = 99) in a 7∶3 ratio. The PD-L1 combined positive score (CPS) ≥ 10 was used as the cut-off value and divided the patients into high and low expression groups. Radiomics feature selection was generated through the χ2 test, the analysis of variance and random forest. An extreme gradient boosting (XGBoost) classifier was employed for model construction. Univariate logistic regression analysis was used to analyze the clinicopathological data. Radiomics modesl, clinicopathological models and combined models were developed for predicting the level of PD-L1. The predictive performance of the model was evaluated using the receiver operating characteristic (ROC) curve, calibration curve and decision curve analysis (DCA).Results There were significant differences in human papillomavirus (HPV) infection and degree of differentiation between the high and low PD-L1 cervical cancer expression groups (all P < 0.05). The AUC of the clinical model in the training and validation sets were 0.672 [95% confidence interval (CI): 0.598 to 0.745] and 0.698 (95% CI: 0.578 to 0.819), respectively. Seven radiomics features were selected from 2261 extracted radiomics features to construct the model, and the AUC was 0.788 (95% CI: 0.728 to 0.848) and 0.712 (95% CI: 0.593 to 0.832) in the training and validation sets, respectively. The AUC of the combined model in the training and validation sets were 0.932 (95% CI: 0.898 to 0.967) and 0.805 (95% CI: 0.694 to 0.916), respectively.Conclusions PD-L1 expression can be effectively predicted using an MRI-based radiomics model combined with clinicopathological characteristics to identify patients who may benefit from anti-PD-L1 immunotherapy.
[Keywords] gynecologic oncology；cervical cancer；programmed death-ligand 1；magnetic resonance imaging；radiomics

SUN Xin   QIN Fengying   TIAN Mingke   WEI Yuze   GAO Xiaozhuo   DONG Yue*  

Department of Radiology, Liaoning Cancer Hospital & Institute, China Medical University Clinical Oncology College, Shenyang 110042, China

Corresponding author: DONG Y, E-mail: dyy1026@sina.com

Conflicts of interest   None.

Received  2025-08-23
Accepted  2025-12-06
DOI: 10.12015/issn.1674-8034.2026.01.012
DOI:10.12015/issn.1674-8034.2026.01.012.

[1]
BRAY F, LAVERSANNE M, SUNG H, et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries[J]. CA Cancer J Clin, 2024, 74(3): 229-263. DOI: 10.3322/caac.21834.
[2]
O'MALLEY D M, OAKNIN A, MONK B J, et al. Phase Ⅱ study of the safety and efficacy of the anti-PD-1 antibody balstilimab in patients with recurrent and/or metastatic cervical cancer[J]. Gynecol Oncol, 2021, 163(2): 274-280. DOI: 10.1016/j.ygyno.2021.08.018.
[3]
O'MALLEY D M, NEFFA M, MONK B J, et al. Dual PD-1 and CTLA-4 checkpoint blockade using balstilimab and zalifrelimab combination as second-line treatment for advanced cervical cancer: an open-label phase II study[J]. J Clin Oncol, 2022, 40(7): 762-771. DOI: 10.1200/JCO.21.02067.
[4]
LOU H M, CAI H B, HUANG X, et al. Cadonilimab combined with chemotherapy with or without bevacizumab as first-line treatment in recurrent or metastatic cervical cancer (COMPASSION-13): a phase 2 study[J]. Clin Cancer Res, 2024, 30(8): 1501-1508. DOI: 10.1158/1078-0432.CCR-23-3162.
[5]
YE H Y, LIAO W J, PAN J L, et al. Immune checkpoint blockade for cancer therapy: current progress and perspectives[J]. J Zhejiang Univ Sci B, 2025, 26(3): 203-226. DOI: 10.1631/jzus.B2300492.
[6]
STEFANOUDAKIS D, KAROPOULOU E, MATSAS A, et al. Immunotherapy in cervical and endometrial cancer: current landscape and future directions[J/OL]. Life (Basel), 2024, 14(3): 344 [2025-11-09]. https://pubmed.ncbi.nlm.nih.gov/38541669/. DOI: 10.3390/life14030344.
[7]
DOROSHOW D B, BHALLA S, BEASLEY M B, et al. PD-L1 as a biomarker of response to immune-checkpoint inhibitors[J]. Nat Rev Clin Oncol, 2021, 18(6): 345-362. DOI: 10.1038/s41571-021-00473-5.
[8]
LIU J, CHEN Z, LI Y, et al. PD-1/PD-L1 checkpoint inhibitors in tumor immunotherapy[J/OL]. Front Pharmacol, 2021, 12: 731798 [2025-11-09]. https://pubmed.ncbi.nlm.nih.gov/34539412/. DOI: 10.3389/fphar.2021.731798.
[9]
Pathology Quality Control Center, Chinese Society of Pathology, Pathology Committee of Chinese Society of Clinical Oncology, et al. Consensus on the immunohistochemical tests of PD-L1 in solid tumors (2021 version)[J]. 2021(7): 710-718. DOI: 10.3760/cma.j.cn112151-20210228-00172.
[10]
SCAPICCHIO C, GABELLONI M, BARUCCI A, et al. A deep look into radiomics[J]. Radiol Med, 2021, 126(10): 1296-1311. DOI: 10.1007/s11547-021-01389-x.
[11]
ZWANENBURG A, VALLIÈRES M, ABDALAH M A, et al. The image biomarker standardization initiative: standardized quantitative radiomics for high-throughput image-based phenotyping[J]. Radiology, 2020, 295(2): 328-338. DOI: 10.1148/radiol.2020191145.
[12]
MEIßNER A K, GUTSCHE R, GALLDIKS N, et al. Radiomics for the non-invasive prediction of PD-L1 expression in patients with brain metastases secondary to non-small cell lung cancer[J]. J Neurooncol, 2023, 163(3): 597-605. DOI: 10.1007/s11060-023-04367-7.
[13]
GONG X Q, LIU N, TAO Y Y, et al. Radiomics models based on multisequence MRI for predicting PD-1/PD-L1 expression in hepatocellular carcinoma[J/OL]. Sci Rep, 2023, 13(1): 7710 [2025-11-09]. https://pubmed.ncbi.nlm.nih.gov/37173350/. DOI: 10.1038/s41598-023-34763-y.
[14]
LIU Z H, YAO Y M, ZHAO M M, et al. Radiomics models derived from arterial-phase-enhanced CT reliably predict both PD-L1 expression and immunotherapy prognosis in non-small cell lung cancer: a retrospective, multicenter cohort study[J]. Acad Radiol, 2025, 32(1): 493-505. DOI: 10.1016/j.acra.2024.07.028.
[15]
LI W, LI L J, LIU Z N, et al. Correlation study between 18F-FDG PET/MR imaging radiomic features and PD-L1 expression in cervical cancer[J]. Chin J Magn Reson Imag, 2024, 15(7): 32-38, 45. DOI: 10.12015/issn.1674-8034.2024.07.006.
[16]
LIU K H, YANG W, TIAN H P, et al. Combined model of clinical, pathological, MRI features and IVIM parameters for predicting programmed cell death protein 1(PD-1)/PD-1 ligand(PD-L1) expression in cervical cancers[J]. Chin J Med Imag Technol, 2023, 39(2): 235-240. DOI: 10.13929/j.issn.1003-3289.2023.02.019.
[17]
JI J F, PANG W Q, SONG J L, et al. Retrospective analysis of the predictive value of 18F-FDG PET/CT metabolic parameters for PD-L1 expression in cervical cancer[J/OL]. Diagnostics (Basel), 2023, 13(6): 1015 [2025-11-09]. https://pubmed.ncbi.nlm.nih.gov/36980323/. DOI: 10.3390/diagnostics13061015.
[18]
ZHOU L W, LIU J W, CHEN M C, et al. Influence of HPV infection on expression of FOXP3, PD-1 and PD-L1 proteins in cervical cancer tissues[J]. Chin J Nosocomiology, 2021, 31(10): 1571-1575. DOI: 10.11816/cn.ni.2021-202582.
[19]
BIRRER M J, FUJIWARA K, OAKNIN A, et al. The changing landscape of systemic treatment for cervical cancer: rationale for inhibition of the TGF-β and PD-L1 pathways[J/OL]. Front Oncol, 2022, 12: 814169 [2025-11-09]. https://pubmed.ncbi.nlm.nih.gov/35280818/. DOI: 10.3389/fonc.2022.814169.
[20]
RISCHIN D, GIL-MARTIN M, GONZÁLEZ-MARTIN A, et al. PD-1 blockade in recurrent or metastatic cervical cancer: Data from cemiplimab phase I expansion cohorts and characterization of PD-L1 expression in cervical cancer[J]. Gynecol Oncol, 2020, 159(2): 322-328. DOI: 10.1016/j.ygyno.2020.08.026.
[21]
GULIJIANATI, ZHAO S J;ZHAO M, et al. Expression levels and significance of PD-1/PDL1 in peripheral blood of cervical cancer and cervical lesions[J]. J Clin Exp Med, 2020, 19(23): 2532-2537. DOI: 10.3969/j.issn.1671-4695.2020.23.020.
[22]
MUNARI E, MARIOTTI F R, QUATRINI L, et al. PD-1/PD-L1 in cancer: pathophysiological, diagnostic and therapeutic aspects[J/OL]. Int J Mol Sci, 2021, 22(10): 5123 [2025-11-09]. https://pubmed.ncbi.nlm.nih.gov/34066087/. DOI: 10.3390/ijms22105123.
[23]
SAGLAM O, ZHOU J M, WANG X F, et al. PD-L1 expression correlates with young age and CD8+ TIL density in poorly differentiated cervical squamous cell carcinoma[J]. Int J Gynecol Pathol, 2020, 39(5): 428-435. DOI: 10.1097/PGP.0000000000000623.
[24]
LIU L U, XIE B, ZHU W, et al. High expression of PD-L1 mainly occurs in non-small cell lung cancer patients with squamous cell carcinoma or poor differentiation[J]. Oncol Res, 2023, 31(3): 275-286. DOI: 10.32604/or.2023.028227.
[25]
CHANG S F, DANG Y M, ZHANG X L, et al. Analysis of PD-L1 level and P53 mutation in colorectal cancer tissues and their clinical significances[J]. Chin Clin Oncol, 2022, 27(12): 1119-1123. DOI: 10.3969/j.issn.1009-0460.2022.12.010.
[26]
HU W J, DENG S D, ZHANG J M, et al. Levels of PD-1 and PD-L1 in peripheral blood and cancer tissues of cervical cancer patients with HPV16 and 18 infections[J]. Chin J Nosocomiology, 2020, 30(16): 2502-2506. DOI: 10.11816/cn.ni.2020-192780.
[27]
HUANG R S P, HABERBERGER J, MURUGESAN K, et al. Clinicopathologic and genomic characterization of PD-L1-positive uterine cervical carcinoma[J]. Mod Pathol, 2021, 34(7): 1425-1433. DOI: 10.1038/s41379-021-00780-3.
[28]
ZHU H. Squamous cell carcinoma antigen: clinical application and research status[J/OL]. Diagnostics (Basel), 2022, 12(5): 1065 [2025-11-09]. https://pubmed.ncbi.nlm.nih.gov/35626221/. DOI: 10.3390/diagnostics12051065.
[29]
TURATO C, SCARPA M, KOTSAFTI A, et al. Squamous cell carcinoma antigen 1 is associated to poor prognosis in esophageal cancer through immune surveillance impairment and reduced chemosensitivity[J]. Cancer Sci, 2019, 110(5): 1552-1563. DOI: 10.1111/cas.13986.
[30]
GAO W, SHAO Z Y, ZHENG Z G, et al. Expression of PD-L1 in cervical cancer and its relationship with clinicopathological features and prognosis of patients[J]. Zhejiang Med J, 2022, 44(13): 1392-1396, 1466. DOI: 10.12056/j.issn.1006-2785.2022.44.13.2021-415.
[31]
SU G H, XIAO Y, YOU C, et al. Radiogenomic-based multiomic analysis reveals imaging intratumor heterogeneity phenotypes and therapeutic targets[J/OL]. Sci Adv, 2023, 9(40): eadf0837 [2025-11-09]. https://pubmed.ncbi.nlm.nih.gov/37801493/. DOI: 10.1126/sciadv.adf0837.
[32]
BERA K, BRAMAN N, GUPTA A, et al. Predicting cancer outcomes with radiomics and artificial intelligence in radiology[J]. Nat Rev Clin Oncol, 2022, 19(2): 132-146. DOI: 10.1038/s41571-021-00560-7.
[33]
VAN GRIETHUYSEN J J M, FEDOROV A, PARMAR C, et al. Computational radiomics system to decode the radiographic phenotype[J/OL]. Cancer Res, 2017, 77(21): e104-e107 [2025-11-09]. https://pubmed.ncbi.nlm.nih.gov/29092951/. DOI: 10.1158/0008-5472.can-17-0339.
[34]
JIANG Y S, CHEN M, NIE H, et al. PD-1 and PD-L1 in cancer immunotherapy: clinical implications and future considerations[J]. Hum Vaccin Immunother, 2019, 15(5): 1111-1122. DOI: 10.1080/21645515.2019.1571892.
[35]
MESSERSMITH L, WILLIAMS G, BROADWATER` D R, et al. Heterogeneity and geographical patterns of PD-L1 positive tumor infiltrating immune cells in colonic adenocarcinoma[J/OL]. Am J Clin Pathol, 2020, 154(Supplement_1): S156 [2025-11-09]. https://doi.org/10.1093/ajcp/aqaa161.340. DOI: 10.1093/ajcp/aqaa161.340.
[36]
MCCAGUE C, RAMLEE S, REINIUS M, et al. Introduction to radiomics for a clinical audience[J]. Clin Radiol, 2023, 78(2): 83-98. DOI: 10.1016/j.crad.2022.08.149.
[37]
LI J H, GE S S, SANG S B, et al. Evaluation of PD-L1 expression level in patients with non-small cell lung cancer by 18F-FDG PET/CT radiomics and clinicopathological characteristics[J/OL]. Front Oncol, 2021, 11: 789014 [2025-11-09]. https://pubmed.ncbi.nlm.nih.gov/34976829/. DOI: 10.3389/fonc.2021.789014.
[38]
BRACCI S, DOLCIAMI M, TROBIANI C, et al. Quantitative CT texture analysis in predicting PD-L1 expression in locally advanced or metastatic NSCLC patients[J]. Radiol Med, 2021, 126(11): 1425-1433. DOI: 10.1007/s11547-021-01399-9.
[39]
ZHENG Y M, ZHAN J F, YUAN M G, et al. A CT-based radiomics signature for preoperative discrimination between high and low expression of programmed death ligand 1 in head and neck squamous cell carcinoma[J/OL]. Eur J Radiol, 2022, 146: 110093 [2025-11-09]. https://pubmed.ncbi.nlm.nih.gov/34890937/. DOI: 10.1016/j.ejrad.2021.110093.

PREV Prediction of lymph node metastasis in cervical cancer using virtual magnetic resonance elastography
NEXT Prognostic value of an MRI-based radiomics machine learning model for patients with locally advanced cervical cancer undergoing concurrent chemoradiotherapy
  


LINK


Tel & Fax: +8610-67113815    E-mail: editor@cjmri.cn