Yun Sun
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Recognitions
Yun Sun has a total of 293 co-authors. Here are the top co-authors by number of shared publications:
Fuyou Li
Department of Chemistry and State Key Laboratory of Molecular Engineering of Polymers and Collaborative Innovation Center of Chemistry for Energy Materials, Fudan University, Shanghai 200433, P. R. China
22 shared publicationsWei Feng
Department of Chemistry, State Key Laboratory of Molecular Engineering of Polymers, Fudan University, 220 Handan Road, Shanghai 200433, China
15 shared publications- Yuhao Li
Institute of Bismuth Science, School of Materials and Chemistry, Shanghai Collaborative Innovation Center of Energy Therapy for Tumors, University of Shanghai for Science and Technology, Shanghai 200093, China
11 shared publications - Fuyou Li
School of Chemistry and Chemical Engineering & Institute of Translational Medicine Shanghai Jiao Tong University Shanghai 200240 China
10 shared publications Juanjuan Peng
Department of Chemistry and State Key Laboratory of Molecular Engineering of Polymers, Fudan University, Shanghai 200433, PR China
10 shared publications- Xian Zeng
School of Pharmaceutical Sciences, Shanghai Engineering Research Center of Immunotherapeutics, Fudan University, Shanghai 201203, China
10 shared publications
Latest Publications
- Open Access
- 21 May 2026
- Published by Wiley in Interdisciplinary Medicine
Enhanced peripheral blood mononuclear cell lipid metabolism drives poor immunotherapy response via ferroptotic stress‐induced T cell exhaustion in hepatocellular carcinoma
Resistance to immunotherapy remains a central challenge in advanced hepatocellular carcinoma (HCC). The role of systemic factors, particularly the metabolic fitness of the circulating immune reservoir, in driving this resistance remains poorly understood. This study aimed to identify pre‐treatment metabolic signatures in peripheral blood mononuclear cells (PBMCs) that predict therapeutic response and to elucidate the underlying mechanisms of T‐cell dysfunction. We performed integrated multi‐omics analyses, including proteomics, metabolomics, lipidomics, and phosphoproteomics, on pre‐treatment PBMCs from a cohort of HCC patients receiving anti‐PD‐1 immunotherapy‐based combination treatment. Predictive metabolic signatures were identified, and key mechanisms were validated using flow cytometry, confocal microscopy, and molecular analyses. An enhanced lipid metabolic signature in pre‐treatment PBMCs was identified as a powerful and independent predictor of poor immunotherapy response and survival. We found that a lipid‐enriched plasma milieu in non‐responders imposes a metabolic constraint on PBMCs. Specifically, CD36 upregulation correlates with increased lipid uptake and accumulation of glycerophospholipids and sphingolipids, triggering sub‐lethal ferroptotic stress (characterized by ferroptosis molecular hallmarks and functional impairment in viable cells without overt cell death) and subsequent T cell exhaustion. This ferroptotic stress state exhibited lipid peroxidation, reduced GPX4 expression, and elevated intracellular iron (Fe 2+ ), and was mechanistically sustained by suppressed AKT survival signaling coupled with hyperactive inflammatory pathways (MAPK, NF‐κB). Importantly, this metabolic dysfunction was reversible upon blocking lipid peroxidation. The systemic metabolic fitness of circulating PBMCs is an important determinant of immunotherapy efficacy in HCC that complements tumor‐intrinsic factors. Our findings provide a new framework for developing non‐invasive predictive biomarkers and demonstrate that PBMC ferroptotic stress is mechanistically reversible, providing a rationale for evaluating systemic metabolic modulation as a strategy to enhance immunotherapy outcomes.
- 21 May 2026
- Published by Wiley in Interdisciplinary Medicine
Open AccessEnhanced peripheral blood mononuclear cell lipid metabolism drives poor immunotherapy response via ferroptotic stress‐induced T cell exhaustion in hepatocellular carcinoma
Resistance to immunotherapy remains a central challenge in advanced hepatocellular carcinoma (HCC). The role of systemic factors, particularly the metabolic fitness of the circulating immune reservoir, in driving this resistance remains poorly understood. This study aimed to identify pre‐treatment metabolic signatures in peripheral blood mononuclear cells (PBMCs) that predict therapeutic response and to elucidate the underlying mechanisms of T‐cell dysfunction. We performed integrated multi‐omics analyses, including proteomics, metabolomics, lipidomics, and phosphoproteomics, on pre‐treatment PBMCs from a cohort of HCC patients receiving anti‐PD‐1 immunotherapy‐based combination treatment. Predictive metabolic signatures were identified, and key mechanisms were validated using flow cytometry, confocal microscopy, and molecular analyses. An enhanced lipid metabolic signature in pre‐treatment PBMCs was identified as a powerful and independent predictor of poor immunotherapy response and survival. We found that a lipid‐enriched plasma milieu in non‐responders imposes a metabolic constraint on PBMCs. Specifically, CD36 upregulation correlates with increased lipid uptake and accumulation of glycerophospholipids and sphingolipids, triggering sub‐lethal ferroptotic stress (characterized by ferroptosis molecular hallmarks and functional impairment in viable cells without overt cell death) and subsequent T cell exhaustion. This ferroptotic stress state exhibited lipid peroxidation, reduced GPX4 expression, and elevated intracellular iron (Fe 2+ ), and was mechanistically sustained by suppressed AKT survival signaling coupled with hyperactive inflammatory pathways (MAPK, NF‐κB). Importantly, this metabolic dysfunction was reversible upon blocking lipid peroxidation. The systemic metabolic fitness of circulating PBMCs is an important determinant of immunotherapy efficacy in HCC that complements tumor‐intrinsic factors. Our findings provide a new framework for developing non‐invasive predictive biomarkers and demonstrate that PBMC ferroptotic stress is mechanistically reversible, providing a rationale for evaluating systemic metabolic modulation as a strategy to enhance immunotherapy outcomes.
- Top Cited Papers
- 11 May 2026
- Published by Elsevier in Free Radical Biology and Medicine
- Vol. 252, 400-410
Carbon ions collapse the hypoxic lipid shield to decouple survival from redox vulnerability in NSCLC
- 11 May 2026
- Published by Elsevier in Free Radical Biology and Medicine
- Vol. 252, 400-410
Top Cited PapersCarbon ions collapse the hypoxic lipid shield to decouple survival from redox vulnerability in NSCLC
- Open Access
- 26 March 2025
- Published by Elsevier in Materials Today Bio
- Vol. 32, 101705
A cerium nanocluster for effective alleviation of inflammatory bowel disease by scavenging RONS and regulating gut microbiome
- 26 March 2025
- Published by Elsevier in Materials Today Bio
- Vol. 32, 101705
Open AccessA cerium nanocluster for effective alleviation of inflammatory bowel disease by scavenging RONS and regulating gut microbiome
- Open Access
- 21 May 2024
- Published by Springer Nature in Cell Communication and Signaling
- Vol. 22 (1) , 1-17
Metabolic self-feeding in HBV-associated hepatocarcinoma centered on feedback between circulation lipids and the cellular MAPK/mTOR axis
Introduction: Hepatitis B Virus (HBV) is widely recognized as a “metabolic virus” that disrupts hepatic metabolic homeostasis, rendering it one of the foremost risk factors for hepatocellular carcinoma (HCC). Except for antiviral therapy, the fundamental principles underlying HBV− and HBV+ HCC have remained unchanged, limiting HCC treatment options. Objectives: In this study, we aim to identify the distinctive metabolic profile of HBV-associated HCC, with the promise of identifying novel metabolic targets that confer survival advantages and ultimately impede cancer progression. Methods: We employed a comprehensive methodology to evaluate metabolic alterations systematically. Initially, we analyzed transcriptomic and proteomic data obtained from a public database, subsequently validating these findings within our test cohort at both the proteomic and transcriptomic levels. Additionally, we conducted a comprehensive analysis of tissue metabolomics profiles, lipidomics, and the activity of the MAPK and AKT signaling pathway to corroborate the abovementioned changes. Results: Our multi-omics approach revealed distinct metabolic dysfunctions associated with HBV-associated HCC. Specifically, we observed upregulated steroid hormone biosynthesis, primary bile acid metabolism, and sphingolipid metabolism in HBV-associated HCC patients’ serum. Notably, metabolites involved in primary bile acid and sphingolipids can activate the MAPK/mTOR pathway. Tissue metabolomics and lipidomics analyses further validated the serum metabolic alterations, particularly alterations in lipid composition and accumulation of unsaturated fatty acids. Conclusion: Our findings emphasize the pivotal role of HBV in HCC metabolism, elucidating the activation of a unique MAPK/mTOR signaling axis by primary bile acids and sphingolipids. Moreover, the hyperactive MAPK/mTOR signaling axis transduction leads to significant reprogramming in lipid metabolism within HCC cells, further triggering the activation of the MAPK/mTOR pathway in turn, thereby establishing a self-feeding circle driven by primary bile acids and sphingolipids.
- 21 May 2024
- Published by Springer Nature in Cell Communication and Signaling
- Vol. 22 (1) , 1-17
Open AccessMetabolic self-feeding in HBV-associated hepatocarcinoma centered on feedback between circulation lipids and the cellular MAPK/mTOR axis
Introduction: Hepatitis B Virus (HBV) is widely recognized as a “metabolic virus” that disrupts hepatic metabolic homeostasis, rendering it one of the foremost risk factors for hepatocellular carcinoma (HCC). Except for antiviral therapy, the fundamental principles underlying HBV− and HBV+ HCC have remained unchanged, limiting HCC treatment options. Objectives: In this study, we aim to identify the distinctive metabolic profile of HBV-associated HCC, with the promise of identifying novel metabolic targets that confer survival advantages and ultimately impede cancer progression. Methods: We employed a comprehensive methodology to evaluate metabolic alterations systematically. Initially, we analyzed transcriptomic and proteomic data obtained from a public database, subsequently validating these findings within our test cohort at both the proteomic and transcriptomic levels. Additionally, we conducted a comprehensive analysis of tissue metabolomics profiles, lipidomics, and the activity of the MAPK and AKT signaling pathway to corroborate the abovementioned changes. Results: Our multi-omics approach revealed distinct metabolic dysfunctions associated with HBV-associated HCC. Specifically, we observed upregulated steroid hormone biosynthesis, primary bile acid metabolism, and sphingolipid metabolism in HBV-associated HCC patients’ serum. Notably, metabolites involved in primary bile acid and sphingolipids can activate the MAPK/mTOR pathway. Tissue metabolomics and lipidomics analyses further validated the serum metabolic alterations, particularly alterations in lipid composition and accumulation of unsaturated fatty acids. Conclusion: Our findings emphasize the pivotal role of HBV in HCC metabolism, elucidating the activation of a unique MAPK/mTOR signaling axis by primary bile acids and sphingolipids. Moreover, the hyperactive MAPK/mTOR signaling axis transduction leads to significant reprogramming in lipid metabolism within HCC cells, further triggering the activation of the MAPK/mTOR pathway in turn, thereby establishing a self-feeding circle driven by primary bile acids and sphingolipids.
- Top Cited Papers
- 16 April 2024
- Published by American Chemical Society (ACS) in ACS Applied Materials & Interfaces
- Vol. 16 (17) , 21546-21556
Oxygen-Independent Radiodynamic Therapy: Radiation-Boosted Chemodynamics for Reprogramming the Tumor Immune Environment and Enhancing Antitumor Immune Response
Radiodynamic therapy (RDT) has emerged as a promising modality for cancer treatment, offering notable advantages such as deep tissue penetration and radiocatalytic generation of oxygen free radicals. However, the oxygen-dependent nature of RDT imposes limitations on its efficacy in hypoxic conditions, particularly in modulating and eliminating radioresistant immune suppression cells. A novel approach involving the creation of a “super” tetrahedron polyoxometalate (POM) cluster, Fe12-POM, has been developed for radiation boosted chemodynamic catalysis to enable oxygen-independent RDT in hypoxic conditions. This nanoscale cluster comprises four P2W15 units functioning as energy antennas, while the Fe3 core serves as an electron receptor and catalytic center. Under X-ray radiation, a metal-to-metal charge transfer phenomenon occurs between P2W15 and the Fe3 core, resulting in the valence transition of Fe3+ to Fe2+ and a remarkable 139-fold increase in hydroxyl radical generation compared to Fe12-POM alone. The rapid generation of hydroxyl radicals, in combination with PD-1 therapy, induces a reprogramming of the immune environment within tumors. This reprogramming is characterized by upregulation of CD80/86, downregulation of CD163 and FAP, as well as the release of interferon-γ and tumor necrosis factor-α. Consequently, the occurrence of abscopal effects is facilitated, leading to significant regression of both local and distant tumors in mice. The development of oxygen-independent RDT represents a promising approach to address cancer recurrence and improve treatment outcomes.
- 16 April 2024
- Published by American Chemical Society (ACS) in ACS Applied Materials & Interfaces
- Vol. 16 (17) , 21546-21556
Top Cited PapersOxygen-Independent Radiodynamic Therapy: Radiation-Boosted Chemodynamics for Reprogramming the Tumor Immune Environment and Enhancing Antitumor Immune Response
Radiodynamic therapy (RDT) has emerged as a promising modality for cancer treatment, offering notable advantages such as deep tissue penetration and radiocatalytic generation of oxygen free radicals. However, the oxygen-dependent nature of RDT imposes limitations on its efficacy in hypoxic conditions, particularly in modulating and eliminating radioresistant immune suppression cells. A novel approach involving the creation of a “super” tetrahedron polyoxometalate (POM) cluster, Fe12-POM, has been developed for radiation boosted chemodynamic catalysis to enable oxygen-independent RDT in hypoxic conditions. This nanoscale cluster comprises four P2W15 units functioning as energy antennas, while the Fe3 core serves as an electron receptor and catalytic center. Under X-ray radiation, a metal-to-metal charge transfer phenomenon occurs between P2W15 and the Fe3 core, resulting in the valence transition of Fe3+ to Fe2+ and a remarkable 139-fold increase in hydroxyl radical generation compared to Fe12-POM alone. The rapid generation of hydroxyl radicals, in combination with PD-1 therapy, induces a reprogramming of the immune environment within tumors. This reprogramming is characterized by upregulation of CD80/86, downregulation of CD163 and FAP, as well as the release of interferon-γ and tumor necrosis factor-α. Consequently, the occurrence of abscopal effects is facilitated, leading to significant regression of both local and distant tumors in mice. The development of oxygen-independent RDT represents a promising approach to address cancer recurrence and improve treatment outcomes.
- Open Access
- 26 February 2024
- Published by Springer Nature in Annals of Nuclear Medicine
- Vol. 38 (5) , 360-368
Properties of [18F]FAPI monitoring of acute radiation pneumonia versus [18F]FDG in mouse models
Objective: In this study, the uptake characteristics of [18F]fibroblast activation protein inhibitor (FAPI) molecular imaging probe were investigated in acute radiation pneumonia and lung cancer xenografted mice before and after radiation to assess the future applicability of [18F]FAPI positron emission tomography/computed tomography (PET/CT) imaging in early radiotherapy response. Methods: Initially, the biodistribution of [18F]FAPI tracer in vivo were studied in healthy mice at each time-point. A comparison of [18F]FAPI and [18F]fluorodeoxyglucose (FDG) PET/CT imaging efficacy in normal ICR, LLC tumor-bearing mice was evaluated. A radiation pneumonia model was then investigated using a gamma counter, small animal PET/CT, and autoradiography. The uptake properties of [18F]FAPI in lung cancer and acute radiation pneumonia were investigated using autoradiography and PET/CT imaging in mice. Results: The tumor area was visible in [18F]FAPI imaging and the tracer was swiftly eliminated from normal tissues and organs. There was a significant increase of [18F]FDG absorption in lung tissue after radiotherapy compared to before radiotherapy, but no significant difference of [18F]FAPI uptake under the same condition. Furthermore, both the LLC tumor volume and the expression of FAP-ɑ decreased after thorax irradiation. Correspondingly, there was no notable [18F]FAPI uptake after irradiation, but there was an increase of [18F]FDG uptake in malignancies and lungs. Conclusions: The background uptake of [18F]FAPI is negligible. Moreover, the uptake of [18F]FAPI may not be affected by acute radiation pneumonitis compared to [18F]FDG, which may be used to more accurately evaluate early radiotherapy response of lung cancer with acute radiation pneumonia.
- 26 February 2024
- Published by Springer Nature in Annals of Nuclear Medicine
- Vol. 38 (5) , 360-368
Open AccessProperties of [18F]FAPI monitoring of acute radiation pneumonia versus [18F]FDG in mouse models
Objective: In this study, the uptake characteristics of [18F]fibroblast activation protein inhibitor (FAPI) molecular imaging probe were investigated in acute radiation pneumonia and lung cancer xenografted mice before and after radiation to assess the future applicability of [18F]FAPI positron emission tomography/computed tomography (PET/CT) imaging in early radiotherapy response. Methods: Initially, the biodistribution of [18F]FAPI tracer in vivo were studied in healthy mice at each time-point. A comparison of [18F]FAPI and [18F]fluorodeoxyglucose (FDG) PET/CT imaging efficacy in normal ICR, LLC tumor-bearing mice was evaluated. A radiation pneumonia model was then investigated using a gamma counter, small animal PET/CT, and autoradiography. The uptake properties of [18F]FAPI in lung cancer and acute radiation pneumonia were investigated using autoradiography and PET/CT imaging in mice. Results: The tumor area was visible in [18F]FAPI imaging and the tracer was swiftly eliminated from normal tissues and organs. There was a significant increase of [18F]FDG absorption in lung tissue after radiotherapy compared to before radiotherapy, but no significant difference of [18F]FAPI uptake under the same condition. Furthermore, both the LLC tumor volume and the expression of FAP-ɑ decreased after thorax irradiation. Correspondingly, there was no notable [18F]FAPI uptake after irradiation, but there was an increase of [18F]FDG uptake in malignancies and lungs. Conclusions: The background uptake of [18F]FAPI is negligible. Moreover, the uptake of [18F]FAPI may not be affected by acute radiation pneumonitis compared to [18F]FDG, which may be used to more accurately evaluate early radiotherapy response of lung cancer with acute radiation pneumonia.
- Open Access
- 5 January 2024
- Published by Frontiers Media SA in Frontiers in Bioengineering and Biotechnology
- Vol. 11, 1357668
Editorial: Novel biomaterials to improve the biological effects of electromagnetic and particle radiation
Editorial: Novel biomaterials to improve the biological effects of electromagnetic and particle radiation
- 5 January 2024
- Published by Frontiers Media SA in Frontiers in Bioengineering and Biotechnology
- Vol. 11, 1357668
Open AccessEditorial: Novel biomaterials to improve the biological effects of electromagnetic and particle radiation
Editorial: Novel biomaterials to improve the biological effects of electromagnetic and particle radiation
- Open Access
- 1 January 2024
- Published by Wiley in Cancer Medicine
- Vol. 13 (2) , e6950
Carbon ion irradiation exerts antitumor activity by inducing cGAS–STING activation and immune response in prostate cancer‐bearing mice
Background and Purpose: As an advanced radiotherapy technique, carbon ion radiotherapy has demonstrated good efficacy and low toxicity for prostate cancer patients, but the radiobiological mechanism of killing tumor cells has not been fully elucidated. This study aims to explore the antitumor effects of carbon ion irradiation (CIR) through investigating the immune response induced by CIR in prostate cancer‐bearing mice and the underlying molecular mechanism.Materials and Methods: We established subcutaneous transplantation tumor models of prostate cancer to evaluate the tumor inhibition effect of CIR. Investigation of immunophenotype alterations were assessed by flow cytometry. Immunofluorescence, western blot, and real‐time quantitative PCR was employed to analyze the activation of cGAS–STING pathway.Results: CIR showed more powerful tumor growth control than photon irradiation in immunocompetent syngeneic C57BL/6 mice. CIR exerts antitumor effect by triggering immune response, characterized by increased CD4+ T cells and macrophages in tumor, enhanced CD8+ T cells and T effector memory cells in spleen, improved IFN‐γ production of CD8+ tumor‐infiltrating lymphocytes, and reduced exhausted T cells in tumor and spleen. Additionally, production of cytoplasmic double‐stranded DNA, protein levels of p‐TBK1 and p‐IRF3 in the cGAS–STING pathway, and gene expression levels of downstream interferon‐stimulated genes were significantly increased after CIR in a dose‐dependent manner. Treatment of RM1 tumor‐bearing mice with the STING inhibitor C‐176 impaired the antitumor effect of CIR.Conclusion: The excellent antitumor activity of CIR in immunocompetent prostate cancer‐bearing C57BL/6 mice may be attributed to stronger induction of antitumor immune response and higher activation of cGAS–STING pathway.
- 1 January 2024
- Published by Wiley in Cancer Medicine
- Vol. 13 (2) , e6950
Open AccessCarbon ion irradiation exerts antitumor activity by inducing cGAS–STING activation and immune response in prostate cancer‐bearing mice
Background and Purpose: As an advanced radiotherapy technique, carbon ion radiotherapy has demonstrated good efficacy and low toxicity for prostate cancer patients, but the radiobiological mechanism of killing tumor cells has not been fully elucidated. This study aims to explore the antitumor effects of carbon ion irradiation (CIR) through investigating the immune response induced by CIR in prostate cancer‐bearing mice and the underlying molecular mechanism.Materials and Methods: We established subcutaneous transplantation tumor models of prostate cancer to evaluate the tumor inhibition effect of CIR. Investigation of immunophenotype alterations were assessed by flow cytometry. Immunofluorescence, western blot, and real‐time quantitative PCR was employed to analyze the activation of cGAS–STING pathway.Results: CIR showed more powerful tumor growth control than photon irradiation in immunocompetent syngeneic C57BL/6 mice. CIR exerts antitumor effect by triggering immune response, characterized by increased CD4+ T cells and macrophages in tumor, enhanced CD8+ T cells and T effector memory cells in spleen, improved IFN‐γ production of CD8+ tumor‐infiltrating lymphocytes, and reduced exhausted T cells in tumor and spleen. Additionally, production of cytoplasmic double‐stranded DNA, protein levels of p‐TBK1 and p‐IRF3 in the cGAS–STING pathway, and gene expression levels of downstream interferon‐stimulated genes were significantly increased after CIR in a dose‐dependent manner. Treatment of RM1 tumor‐bearing mice with the STING inhibitor C‐176 impaired the antitumor effect of CIR.Conclusion: The excellent antitumor activity of CIR in immunocompetent prostate cancer‐bearing C57BL/6 mice may be attributed to stronger induction of antitumor immune response and higher activation of cGAS–STING pathway.
- Open Access
- 2 December 2023
- Published by Springer Nature in European Journal of Medical Research
- Vol. 28 (1) , 1-13
Clinico-biological-radiomics (CBR) based machine learning for improving the diagnostic accuracy of FDG-PET false-positive lymph nodes in lung cancer
Background: The main problem of positron emission tomography/computed tomography (PET/CT) for lymph node (LN) staging is the high false positive rate (FPR). Thus, we aimed to explore a clinico-biological-radiomics (CBR) model via machine learning (ML) to reduce FPR and improve the accuracy for predicting the hypermetabolic mediastinal–hilar LNs status in lung cancer than conventional PET/CT. Methods: A total of 260 lung cancer patients with hypermetabolic mediastinal–hilar LNs (SUVmax ≥ 2.5) were retrospectively reviewed. Patients were treated with surgery with systematic LN resection and pathologically divided into the LN negative (LN-) and positive (LN +) groups, and randomly assigned into the training (n = 182) and test (n = 78) sets. Preoperative CBR dataset containing 1738 multi-scale features was constructed for all patients. Prediction models for hypermetabolic LNs status were developed using the features selected by the supervised ML algorithms, and evaluated using the classical diagnostic indicators. Then, a nomogram was developed based on the model with the highest area under the curve (AUC) and the lowest FPR, and validated by the calibration plots. Results: In total, 109 LN− and 151 LN + patients were enrolled in this study. 6 independent prediction models were developed to differentiate LN− from LN + patients using the selected features from clinico-biological-image dataset, radiomics dataset, and their combined CBR dataset, respectively. The DeLong test showed that the CBR Model containing all-scale features held the highest predictive efficiency and the lowest FPR among all of established models (p < 0.05) in both the training and test sets (AUCs of 0.90 and 0.89, FPRs of 12.82% and 6.45%, respectively) (p < 0.05). The quantitative nomogram based on CBR Model was validated to have a good consistency with actual observations. Conclusion: This study presents an integrated CBR nomogram that can further reduce the FPR and improve the accuracy of hypermetabolic mediastinal–hilar LNs evaluation than conventional PET/CT in lung cancer, thereby greatly reducing the risk of overestimation and assisting for precision treatment.
- 2 December 2023
- Published by Springer Nature in European Journal of Medical Research
- Vol. 28 (1) , 1-13
Open AccessClinico-biological-radiomics (CBR) based machine learning for improving the diagnostic accuracy of FDG-PET false-positive lymph nodes in lung cancer
Background: The main problem of positron emission tomography/computed tomography (PET/CT) for lymph node (LN) staging is the high false positive rate (FPR). Thus, we aimed to explore a clinico-biological-radiomics (CBR) model via machine learning (ML) to reduce FPR and improve the accuracy for predicting the hypermetabolic mediastinal–hilar LNs status in lung cancer than conventional PET/CT. Methods: A total of 260 lung cancer patients with hypermetabolic mediastinal–hilar LNs (SUVmax ≥ 2.5) were retrospectively reviewed. Patients were treated with surgery with systematic LN resection and pathologically divided into the LN negative (LN-) and positive (LN +) groups, and randomly assigned into the training (n = 182) and test (n = 78) sets. Preoperative CBR dataset containing 1738 multi-scale features was constructed for all patients. Prediction models for hypermetabolic LNs status were developed using the features selected by the supervised ML algorithms, and evaluated using the classical diagnostic indicators. Then, a nomogram was developed based on the model with the highest area under the curve (AUC) and the lowest FPR, and validated by the calibration plots. Results: In total, 109 LN− and 151 LN + patients were enrolled in this study. 6 independent prediction models were developed to differentiate LN− from LN + patients using the selected features from clinico-biological-image dataset, radiomics dataset, and their combined CBR dataset, respectively. The DeLong test showed that the CBR Model containing all-scale features held the highest predictive efficiency and the lowest FPR among all of established models (p < 0.05) in both the training and test sets (AUCs of 0.90 and 0.89, FPRs of 12.82% and 6.45%, respectively) (p < 0.05). The quantitative nomogram based on CBR Model was validated to have a good consistency with actual observations. Conclusion: This study presents an integrated CBR nomogram that can further reduce the FPR and improve the accuracy of hypermetabolic mediastinal–hilar LNs evaluation than conventional PET/CT in lung cancer, thereby greatly reducing the risk of overestimation and assisting for precision treatment.