Probiotics are natural systems bridging synthetic biology, physical biotechnology, and immunology, initiating innate and adaptive anti-tumor immune activity. We previously constructed an all-in-one engineered food-grade probiotic Lactococcus lactis (FOLactis) which could boost the crosstalk among different immune cells such as dendritic cells (DCs), natural killer cells, and T cells. Herein, considering the limited clinical efficacy of naked personalized neoantigen peptide vaccines, we decorate FOLactis with tumor antigens by employing a Plug-and-Display system comprising membrane-inserted peptides. Intranodal injection of FOLactis coated with neoantigen peptides (Ag-FOLactis) induces robust DCs presentation and neoantigen-specific cellular immunity. Notably, Ag-FOLactis not only triggers a 45-fold rise in the quantity of locally reactive neoantigen-specific T cells but also induces epitope spreading in both subcutaneous and metastatic tumor-bearing models, leading to potent inhibition of tumor growth. These findings imply that Ag-FOLactis represents a powerful platform to rapidly and easily display antigens, facilitating the development of a bio-activated platform for personalized therapy.

Objective:To develop and assess a three-class machine learning model for predicting wild-type, 19 del, and 21 L858R mutations of the epidermal growth factor receptor (EGFR) in lung adenocarcinoma using 18F-FDG PET/CT radiomic features and clinical features. Methods:The retrospective data was collected from 703 patients (346 males, 357 females; age (64.3±9.0) years) with lung adenocarcinoma at the First People′s Hospital of Changzhou from January 2018 to June 2023. Patients were divided into the training set (563 cases) and test set (140 cases) at the ratio of 8∶2. Clinical features were selected using recursive feature elimination (RFE). Radiomic features were extracted from PET and CT images, and the optimal feature sets were selected using minimum redundancy maximum relevance (mRMR) and least absolute shrinkage and selection operator (LASSO) methods. Base models were constructed by using random forest (RF), logistic regression (LR), support vector machine (SVM), K-nearest neighbors (KNN), and multi-layer perceptron (MLP), and the stacking method was applied to establish the CT and PET ensemble models. Delong test was used to compare the AUC differences between the PET/CT combined model and the clinical + PET/CT integrated model.Results:Among 703 patients, 273 were with EGFR wild-type, 202 were with 19 del mutation, and 228 were with 21 L858R mutation. In the single-modal analysis, the AUCs of CT ensemble model in the training and test sets were 0.893 and 0.667, respectively, while the AUCs of PET ensemble model were 0.692 and 0.660. The AUC of PET/CT combined model were 0.897 in training set and 0.672 in test set. The AUC of clinical + PET/CT integrated model showed further improvement, with AUCs of 0.902 and 0.721 in training and test sets, respectively. Notably, the clinical + PET/CT integrated model outperformed PET/CT combined model in predicting wild-type EGFR (test set AUC: 0.784 vs 0.707; Z=3.28, P=0.001). Conclusion:The three-class model (clinical + PET/CT integrated model) based on 18F-FDG PET/CT radiomics and clinical features effectively predicts EGFR mutation subtypes in lung adenocarcinoma.

With the rapid development of generative adversarial networks (GAN), learning the mapping between CT and PET images enables cross-modality generation. This not only integrates anatomical and functional information to improve image quality, but also helps reduce the radiation exposure to some extent. Based on a review of representative GAN architectures such as conditional GAN and CycleGAN, this paper discusses their research progress and limitations in various application scenarios, including initial tumor diagnosis and staging, treatment evaluation and follow-up, as well as methods for reducing PET/CT radiation dose. Additionally, challenges related to small-sample learning, model interpretability, and cross-institutional standardization are highlighted, and the clinical application prospects of GAN-based cross-modality generation technology are explored.

Objective:To explore the value of 18F-fluorodeoxyglucose ( 18F-FDG) PET-CT dual-modality habitat imaging technology in predicting the epidermal growth factor receptor (EGFR) mutation status in lung adenocarcinoma. Methods:This study was designed as a cross-sectional study. Clinical and imaging data of 403 patients with lung adenocarcinoma who underwent 18F-FDG PET-CT imaging with definitive EGFR results from January 2018 to April 2022 at the Third Affiliated Hospital of Soochow University were retrospectively analyzed.The patients were divided into a development set (282 cases) and a validation set (121 cases) using a stratified random sampling method at a 7∶3 ratio. An adaptive clustering algorithm was used to segment the regions of interest, forming different habitats and obtaining derived parameters. Independent samples t-test or Mann-Whitney U test were used to compare clinical, imaging indicators, and habitat-derived parameters between EGFR mutant and wild-type patient. The clinical, imaging indicators, and habitat-derived parameters that showed statistically significant differences in univariate analysis were included in multivariate logistic regression to construct clinical and clinical-habitat combined models, respectively. The receiver operating characteristic curve and area under the curve (AUC) were used to evaluate the model′s ability to predict EGFR mutations in lung adenocarcinoma. Additionally, the net reclassification index (NRI) was employed to assess the model′s classification improvement capability. Results:There were 249 cases of EGFR mutation and 154 cases of wild type. The optimal number of habitats was two, namely Habitat 1 and Habitat 2. The parameters included in the clinical model were smoking history, bronchial sign, pleural indentation sign, and tumor diameter. The parameters incorporated into the clinical-habitat combined model were smoking history, bronchial sign, pleural indentation sign, Habitat 2, and Habitat 1 voxel count. In the development set, the AUCs for predicting EGFR mutations in lung adenocarcinoma using the clinical model and the clinical-habitat combined model were 0.723 and 0.733, respectively, with no statistically significant difference ( Z=0.60, P=0.549); In the validation set, the AUCs were 0.684 and 0.715, respectively, with no statistically significant difference ( Z=1.32, P=0.186). The accuracy (0.694) and specificity (0.609) of the clinical-habitat combined model in the validation set were slightly higher than those of the clinical model (0.686 and 0.565, respectively). NRI analysis confirmed that the clinical-habitat combined model improved the correct classification of EGFR wild-type lung adenocarcinoma by 10.9% compared to the clinical model ( P=0.018). Conclusion:18F-FDG PET-CT dual-modality habitat imaging technology can be used to analyze the microenvironment of lung adenocarcinoma and has the potential in non-invasively predicting EGFR mutation status, providing an important basis for personalized and accurate treatment of patients with lung adenocarcinoma.

With the rapid development of generative adversarial networks (GAN), learning the mapping between CT and PET images enables cross-modality generation. This not only integrates anatomical and functional information to improve image quality, but also helps reduce the radiation exposure to some extent. Based on a review of representative GAN architectures such as conditional GAN and CycleGAN, this paper discusses their research progress and limitations in various application scenarios, including initial tumor diagnosis and staging, treatment evaluation and follow-up, as well as methods for reducing PET/CT radiation dose. Additionally, challenges related to small-sample learning, model interpretability, and cross-institutional standardization are highlighted, and the clinical application prospects of GAN-based cross-modality generation technology are explored.

Objective:To develop and assess a three-class machine learning model for predicting wild-type, 19 del, and 21 L858R mutations of the epidermal growth factor receptor (EGFR) in lung adenocarcinoma using 18F-FDG PET/CT radiomic features and clinical features. Methods:The retrospective data was collected from 703 patients (346 males, 357 females; age (64.3±9.0) years) with lung adenocarcinoma at the First People′s Hospital of Changzhou from January 2018 to June 2023. Patients were divided into the training set (563 cases) and test set (140 cases) at the ratio of 8∶2. Clinical features were selected using recursive feature elimination (RFE). Radiomic features were extracted from PET and CT images, and the optimal feature sets were selected using minimum redundancy maximum relevance (mRMR) and least absolute shrinkage and selection operator (LASSO) methods. Base models were constructed by using random forest (RF), logistic regression (LR), support vector machine (SVM), K-nearest neighbors (KNN), and multi-layer perceptron (MLP), and the stacking method was applied to establish the CT and PET ensemble models. Delong test was used to compare the AUC differences between the PET/CT combined model and the clinical + PET/CT integrated model.Results:Among 703 patients, 273 were with EGFR wild-type, 202 were with 19 del mutation, and 228 were with 21 L858R mutation. In the single-modal analysis, the AUCs of CT ensemble model in the training and test sets were 0.893 and 0.667, respectively, while the AUCs of PET ensemble model were 0.692 and 0.660. The AUC of PET/CT combined model were 0.897 in training set and 0.672 in test set. The AUC of clinical + PET/CT integrated model showed further improvement, with AUCs of 0.902 and 0.721 in training and test sets, respectively. Notably, the clinical + PET/CT integrated model outperformed PET/CT combined model in predicting wild-type EGFR (test set AUC: 0.784 vs 0.707; Z=3.28, P=0.001). Conclusion:The three-class model (clinical + PET/CT integrated model) based on 18F-FDG PET/CT radiomics and clinical features effectively predicts EGFR mutation subtypes in lung adenocarcinoma.

Objective:To explore the value of 18F-fluorodeoxyglucose ( 18F-FDG) PET-CT dual-modality habitat imaging technology in predicting the epidermal growth factor receptor (EGFR) mutation status in lung adenocarcinoma. Methods:This study was designed as a cross-sectional study. Clinical and imaging data of 403 patients with lung adenocarcinoma who underwent 18F-FDG PET-CT imaging with definitive EGFR results from January 2018 to April 2022 at the Third Affiliated Hospital of Soochow University were retrospectively analyzed.The patients were divided into a development set (282 cases) and a validation set (121 cases) using a stratified random sampling method at a 7∶3 ratio. An adaptive clustering algorithm was used to segment the regions of interest, forming different habitats and obtaining derived parameters. Independent samples t-test or Mann-Whitney U test were used to compare clinical, imaging indicators, and habitat-derived parameters between EGFR mutant and wild-type patient. The clinical, imaging indicators, and habitat-derived parameters that showed statistically significant differences in univariate analysis were included in multivariate logistic regression to construct clinical and clinical-habitat combined models, respectively. The receiver operating characteristic curve and area under the curve (AUC) were used to evaluate the model′s ability to predict EGFR mutations in lung adenocarcinoma. Additionally, the net reclassification index (NRI) was employed to assess the model′s classification improvement capability. Results:There were 249 cases of EGFR mutation and 154 cases of wild type. The optimal number of habitats was two, namely Habitat 1 and Habitat 2. The parameters included in the clinical model were smoking history, bronchial sign, pleural indentation sign, and tumor diameter. The parameters incorporated into the clinical-habitat combined model were smoking history, bronchial sign, pleural indentation sign, Habitat 2, and Habitat 1 voxel count. In the development set, the AUCs for predicting EGFR mutations in lung adenocarcinoma using the clinical model and the clinical-habitat combined model were 0.723 and 0.733, respectively, with no statistically significant difference ( Z=0.60, P=0.549); In the validation set, the AUCs were 0.684 and 0.715, respectively, with no statistically significant difference ( Z=1.32, P=0.186). The accuracy (0.694) and specificity (0.609) of the clinical-habitat combined model in the validation set were slightly higher than those of the clinical model (0.686 and 0.565, respectively). NRI analysis confirmed that the clinical-habitat combined model improved the correct classification of EGFR wild-type lung adenocarcinoma by 10.9% compared to the clinical model ( P=0.018). Conclusion:18F-FDG PET-CT dual-modality habitat imaging technology can be used to analyze the microenvironment of lung adenocarcinoma and has the potential in non-invasively predicting EGFR mutation status, providing an important basis for personalized and accurate treatment of patients with lung adenocarcinoma.

Objective:To investigate the value of intratumoral and peritumoral radiomics models based on 18F-FDG PET-CT in predicting epidermal growth factor receptor (EGFR) mutation status in lung adenocarcinoma and interpret peritumoral radiomics features. Methods:This study was a cross-sectional study. Patients with lung adenocarcinoma who underwent 18F-FDG PET-CT at the Third Affiliated Hospital of Soochow University between January 2018 and April 2022 were retrospectively collected and samplied into a training set (309 cases) and a test set (206 cases) in a 6∶4 ratio randomly. Radiomics features were extracted from the intratumoral and peritumoral regions of interest based on PET and CT images, respectively, and the optimal feature sets were selected. Radiomics models were established using the XGBoost algorithm, and radiomics scores (intratumoral CT label, peritumoral CT label, intratumoral PET label, peritumoral PET label) were calculated. Logistic regression analysis was used to construct a clinical model and a combined model (incorporating PET-CT intratumoral and peritumoral radiomics, clinical features, and CT semantic features). The predictive performance of the models was evaluated using receiver operating characteristic curves and the area under the curve (AUC). Unsupervised clustering, Spearman correlation analysis, and visualization methods were used for the interpretability of peritumoral radiomics features. Results:In both the training and test sets, the AUC value of CT peritumoral labels was greater than that of CT intratumoral labels for predicting EGFR mutation status in lung adenocarcinoma (training set: Z=3.84, P<0.001; test set: Z=1.99, P=0.046). In the test set, the AUC value of PET intratumoral labels (0.684) was slightly higher than that of PET peritumoral labels (0.672) for predicting EGFR mutation status, but the difference was not statistically significant ( P>0.05). The combined model had the highest AUC value for predicting EGFR mutation status of lung adenocarcinoma in both the training and test sets and was significantly better than the clinical model (training set: Z=6.52, P<0.001; test set: Z=2.31, P=0.021). Interpretability analysis revealed that CT peritumoral radiomics features were correlated with CT shape features, and there were significant differences in CT peritumoral features between different EGFR mutation statuses. Conclusions:The value of CT peritumoral labels is superior to that of CT intratumoral labels in predicting EGFR mutation status in lung adenocarcinoma. The predictive performance of the model can be improved by combining PET-CT intratumoral and peritumoral radiomics, clinical features, and CT semantic features.

Objective:To investigate the correlation between the SUV index (SUV max of the lesion/SUV mean of the liver) in 18F-FDG PET/CT imaging and the invasiveness of early lung adenocarcinoma presenting as ground-glass nodule (GGN). Methods:From January 2012 to March 2020, 167 GGN patients (49 males, 118 females; age: (61.5±9.0) years) with early lung adenocarcinoma who underwent PET/CT imaging in Changzhou First People′s Hospital were retrospectively enrolled. The image parameters including the GGN number, location, type, edge, shape, abnormal bronchus sign, vacuole sign, pleural depression, vessel convergence sign, GGN diameter ( DGGN), solid component diameter ( Dsolid), consolidation to tumor ratio (CTR, Dsolid/ DGGN), CT values (CT value of ground-glass opacity (CT GGO), CT value of lung parenchyma (CT LP), ΔCT GGO-LP (CT GGO-CT LP)) and SUV index were analyzed. Single and multivariate logistic regressions were used to analyze the correlation between SUV index and infiltration. The generalized additive model was used for curve fitting, and the piece-wise regression model was used to further explain the nonlinearity. Results:In 189 GGNs, invasive adenocarcinoma accounted for 85.2% (161/189). Single logistic regression showed that the GGN number, type, shape, edge, abnormal bronchus sign, pleural depression, vessel convergence sign, DGGN, Dsolid, CTR, CT GGO, ΔCT GGO-LP and SUV index were related factors of infiltration (odds ratio ( OR) values: 0.396-224.083, P<0.001 or P<0.05). After fully adjusting for confounding factors, SUV index was significantly correlated with increased risk of invasion ( OR=2.162 (95% CI: 1.191-3.923), P=0.011). Curve fitting showed that the SUV index was non-linearly related to the risk of infiltration, and the risk of infiltration increased significantly only when the SUV index was greater than 0.43 ( OR=3.509 (95% CI: 1.429-8.620), P=0.006). The correlation between SUV index and infiltration had no interaction between age, vacuoles, pleural depression and CTR subgroups (all P>0.05). Conclusions:SUV index is an independent factor related to the invasiveness of early lung adenocarcinoma. The higher the SUV index, the greater the risk of invasion; but the two are not simply linearly correlated.

Purpose: Neoadjuvant therapy modality can increase the operability rate and mitigate pathological risks in locally advanced cervical cancer, but treatment response varies widely. It remains unclear whether genetic alterations correlate with the response to neoadjuvant therapy and disease-free survival (DFS) in locally advanced cervical cancer. Materials and Methods: A total of 62 locally advanced cervical cancer (stage IB-IIA) patients who received neoadjuvant chemoradiation plus radical hysterectomy were retrospectively analyzed. Patients’ tumor biopsy samples were comprehensively profiled using targeted next generation sequencing. Pathologic response to neoadjuvant treatment and DFS were evaluated against the association with genomic traits. Results: Genetic alterations of PIK3CA were most frequent (37%), comparable to that of Caucasian populations from The Cancer Genome Atlas. The mutation frequency of genes including TERT, POLD1, NOS2, and FGFR3 was significantly higher in Chinese patients whereas RPTOR, EGFR, and TP53 were underrepresented in comparison to Caucasians. Germline mutations were identified in 21% (13/62) of the cohort and more than half (57%) had mutations in DNA damage repair genes, including BRCA1/2, TP53 and PALB2. Importantly, high tumor mutation burden, TP53 polymorphism (rs1042522), and KEAP1 mutations were found to be associated with poor pathologic response to neoadjuvant chemoradiation treatment. KEAP1 mutations, PIK3CA-SOX2 co-amplification, TERC copy number gain, and TYMS polymorphism correlated with an increased risk of disease relapse. Conclusion We report the genomic profile of locally advanced cervical cancer patients and the distinction between Asian and Caucasian cohorts. Our findings highlight genomic traits associated with unfavorable neoadjuvant chemoradiation response and a higher risk of early disease recurrence.