Background and purpose:Diffuse large B-cell lymphoma(DLBCL)is subclassified into germinal center B-cell-like(GCB)and non-GCB subtypes,which differ in prognosis and treatment response.However,current distinction still relies on invasive pathological assays.This study developed radiomics and deep-learning models based on multiparametric magnetic resonance imaging(MRI)to non-invasively differentiate the two subtypes preoperatively,thereby reducing dependence on histopathological examination.Methods:This study retrospectively included patients with pathologically confirmed DLBCL diagnosed at Huashan Hospital,Fudan University,and other institutions between March 2013 and December 2024.Using multiparametric MRI data,we developed DLBCL-subtype classification models that combined 4 radiomics-based machine-learning algorithms:support vector machine(SVM),logistic regression(LR),Gaussian process(GP)and Naive Bayes(NB),with 3 deep-learning architectures[densely-connected convolutional networks 121(DenseNet121),residual network 101(ResNet101)and EfficientNet-b5].Additionally,two radiologists with different experience levels independently classified DLBCL on MRI in a blinded fashion.Model and radiologist performance were quantified using the area under the receiver operating characteristic curve(AUC),accuracy(ACC),and F1-score to evaluate their ability to distinguish GCB from non-GCB subtypes.This study was approved by the Ethics Committee of Huashan Hospital of Fudan University(No.KY2024-663),and all patients signed informed consents.Results:A total of 173 patients were enrolled(55 with GCB subtype and 118 with non-GCB subtype).Radiomics and deep learning methods effectively distinguished DLBCL subtypes.Among these,the GP radiomics model(based on T1-CE+T2-FLAIR+ADC sequences)and DenseNet121 deep learning model(based on T1-CE+T2-FLAIR+ADC sequences)demonstrated optimal performance.Both achieved excellent results on the internal validation set(GP:AUC=0.900,ACC=0.896,F1=0.840;DenseNet121:AUC=0.846,ACC=0.854,F1=0.774)and maintained robustness on the external validation set.Furthermore,the classification efficacy of the optimal AI model surpassed that of experienced radiologists(highest physician AUC=0.678).Conclusion:Radiomics and deep-learning models based on multiparametric MRI features can effectively differentiate GCB from non-GCB subtypes of DLBCL.Among them,GP and DenseNet121 exhibit outstanding performance,especially when integrating multi-sequence feature sets for classifying DLBCL subtypes on complex imaging data.

Background and purpose:Diffuse large B-cell lymphoma(DLBCL)is subclassified into germinal center B-cell-like(GCB)and non-GCB subtypes,which differ in prognosis and treatment response.However,current distinction still relies on invasive pathological assays.This study developed radiomics and deep-learning models based on multiparametric magnetic resonance imaging(MRI)to non-invasively differentiate the two subtypes preoperatively,thereby reducing dependence on histopathological examination.Methods:This study retrospectively included patients with pathologically confirmed DLBCL diagnosed at Huashan Hospital,Fudan University,and other institutions between March 2013 and December 2024.Using multiparametric MRI data,we developed DLBCL-subtype classification models that combined 4 radiomics-based machine-learning algorithms:support vector machine(SVM),logistic regression(LR),Gaussian process(GP)and Naive Bayes(NB),with 3 deep-learning architectures[densely-connected convolutional networks 121(DenseNet121),residual network 101(ResNet101)and EfficientNet-b5].Additionally,two radiologists with different experience levels independently classified DLBCL on MRI in a blinded fashion.Model and radiologist performance were quantified using the area under the receiver operating characteristic curve(AUC),accuracy(ACC),and F1-score to evaluate their ability to distinguish GCB from non-GCB subtypes.This study was approved by the Ethics Committee of Huashan Hospital of Fudan University(No.KY2024-663),and all patients signed informed consents.Results:A total of 173 patients were enrolled(55 with GCB subtype and 118 with non-GCB subtype).Radiomics and deep learning methods effectively distinguished DLBCL subtypes.Among these,the GP radiomics model(based on T1-CE+T2-FLAIR+ADC sequences)and DenseNet121 deep learning model(based on T1-CE+T2-FLAIR+ADC sequences)demonstrated optimal performance.Both achieved excellent results on the internal validation set(GP:AUC=0.900,ACC=0.896,F1=0.840;DenseNet121:AUC=0.846,ACC=0.854,F1=0.774)and maintained robustness on the external validation set.Furthermore,the classification efficacy of the optimal AI model surpassed that of experienced radiologists(highest physician AUC=0.678).Conclusion:Radiomics and deep-learning models based on multiparametric MRI features can effectively differentiate GCB from non-GCB subtypes of DLBCL.Among them,GP and DenseNet121 exhibit outstanding performance,especially when integrating multi-sequence feature sets for classifying DLBCL subtypes on complex imaging data.

Objective    To summarize the early outcomes of totally thoracoscopic minimally invasive aortic valve replacement (AVR) and double valve replacement (DVR). Methods    The clinical data of patients who underwent totally thoracoscopic minimally invasive AVR or DVR in Guangdong Provincial People’s Hospital from April 2020 to January 2021 were retrospectively analyzed. The patients were divided into an AVR group and a DVR group according to the surgical method, and the clinical data of the two groups were compared. Results    Finally 22 patients were enrolled, including 14 males and 8 females with an average age of 50.0±11.2 years at operation. Eight patients were degenerative disease, 8 were rheumatic heart disease combined with valvular disease, and 6 were bicuspid aortic valve. Out of the 22 patients, 16 underwent AVR alone, and 6 underwent DVR. All patients completed the operation successfully, and there was no death. Perivalvular leakage during surgery occurred in 2 patients. The average cardiopulmonary bypass time was 187.0±39.9 minutes, and aortic cross-clamping time was 117.0 (99.0, 158.0) minutes. Duration of mechanical ventilation and intensive care unit stay was 9.5 (4.8, 18.3) hours and 41.0 (34.0, 64.0) hours, respectively. The volume of chest drainage at the first 24 hours after surgery was 214.0±124.6 mL, and the postoperative hospital stay was 5.5 (4.0, 8.3) days. The cardiopulmonary bypass time and aortic cross-clamping time in the DVR group were longer than those in the AVR group, and the volume of chest drainage at 24 hours after surgery was more than that in the AVR group, with a statistical difference (P<0.05). Echocardiography before hospital discharge showed paravalvular leakage in 1 patient. There was no death during follow-up of 5.9±3.0 months. Conclusion    The early outcome of totally thoracoscopic minimally invasive AVR and DVR is satisfactory, and the approach of surgery is worth exploring.

We reported a 26-year-old male who was diagnosed with apical hypertrophic cardiomyopathy with left ventricular aneurysm. The location of the hypertrophic myocardium and the extent of resection were accurately assessed preoperatively using 3D modeling and printing technology. Myectomy was performed via transapical approach, and the intraoperative exploration was consistent with the description of the preoperative 3D modeling. The patient underwent the surgery successfully without any complications during the hospitalization, and the cardiopulmonary bypass time was 117 min, the aortic cross-clamping time was 57 min, and the hospital stay time was 7 d. The postoperative echocardiography demonstrated left ventricular cavity flow patency. This case provides a reference for the management of patients with apical hypertrophic cardiomyopathy.

Objective     To evaluate the early clinical outcomes of the Renatus® balloon-expandable valve in the treatment of severe aortic stenosis. Methods    From November 2021 to April 2022, a total of 38 patients who received Renatus® balloon-expandable valve for severe aortic stenosis in Guangdong Provincial People's Hospital were included. There were 22 males and 16 females, with an average age of 73.7±5.3 years. Mean aortic gradient and peak aortic jet velocity at baseline, post-procedure, and follow-up were compared. Clinical outcomes including all-cause mortality, perivalvular leakage, serious adverse cardiovascular events and the occurrence of permanent pacemaker implantation were assessed. Results    All patients completed the procedure successfully without conversion to thoracotomy or perioperative death. The post-implant mean aortic pressure gradient was decreased from 41.5 (27.8, 58.8) mm Hg to 6.0 (3.0, 8.0) mm Hg, and the peak aortic jet velocity was also decreased from 4.1±0.9 m/s to 1.7±0.4 m/s (P<0.001). Pacemakers were required in 2 (5.3%) patients. The median follow-up time was 27.5 (23.0, 87.5) d, with a follow-up rate of 100.0%. The mean aortic gradient was 8.0 (7.0, 10.8) mm Hg and peak aortic jet velocity was 2.0±0.3 m/s, showing significant improvement compared with those in the preoperative period (P<0.001). No severe aortic regurgitation or paravalvular leak was observed. There was no serious cardiovascular adverse event or reoperative event during the study period. Conclusion    Transcatheter aortic valve replacement with the domestic Renatus® balloon-expandable valve system is a safe and effective procedure for selected patients with severe aortic stenosis who are at high risk or not candidates for surgical aortic valve replacement.

Mandibular defects caused by injuries, tumors, and infections are common and can severely affect mandibular function and the patient's appearance. However, mandible reconstruction with a mandibular bionic structure remains challenging. Inspired by the process of intramembranous ossification in mandibular development, a hierarchical vascularized engineered bone consisting of angiogenesis and osteogenesis modules has been produced. Moreover, the hierarchical vascular network and bone structure generated by these hierarchical vascularized engineered bone modules match the particular anatomical structure of the mandible. The ultra-tough polyion complex has been used as the basic scaffold for hierarchical vascularized engineered bone for ensuring better reconstruction of mandible function. According to the results of in vivo experiments, the bone regenerated using hierarchical vascularized engineered bone is similar to the natural mandibular bone in terms of morphology and genomics. The sonic hedgehog signaling pathway is specifically activated in hierarchical vascularized engineered bone, indicating that the new bone in hierarchical vascularized engineered bone underwent a process of intramembranous ossification identical to that of mandible development. Thus, hierarchical vascularized engineered bone has a high potential for clinical application in mandibular defect reconstruction. Moreover, the concept based on developmental processes and bionic structures provides an effective strategy for tissue regeneration.
Bone Regeneration ; Bone Transplantation/methods* ; Hedgehog Proteins ; Humans ; Mandible/surgery* ; Osteogenesis

Objective:To evaluate the effect of preventive application of PEG-rhG-CSF on the prevention of neutropenia during concurrent chemoradiotherapy in patients with lung cancer.Methods:A total of 149 patients with lung cancer who received concurrent chemoradiotherapy at Peking University Cancer Hospital from April 2020 to April 2021 were retrospectively analyzed. There were 79 cases in the prevention group, including 48 cases of primary prevention group (preventive application of PEG-rhG-CSF in all concurrent chemoradiotherapy cycles) and 31 cases of secondary prevention group (preventive application of PEG-rhG-CSF in the concurrent chemoradiotherapy cycles after neutropenia occurred). There were 70 cases in non-prevention group. The incidence of grade 3-4 neutropenia, the completion rate of concurrent chemoradiotherapy, the rate of chemoradiotherapy dose reduction and treatment delay, and the rate of hematological toxicities related hospitalization were compared between the prevention group and the non-prevention group.Results:The incidence of grade 3-4 neutropenia in the whole group was 32.2% (48/149), including 6.3% (3/48) in the primary prevention group, 9.7% (3/31) in the secondary prevention group, and 35.7% (25/70) in the non-prevention group. The difference was statistically significant ( χ2=17.81, P<0.001) in the incidence of grade 3-4 neutropenia. The incidence of febrile neutropenia was 3.4% (5/149) in the whole group, but none of them occurred in the primary prevention group. The full completion rate of concurrent chemotherapy was 96.2% (76/79) in the prevention group, which was significantly higher than 82.9% (58/70) in the non-prevention group ( χ2=7.30, P=0.007). The incidence of treatment delayed and dose reduction of chemoradiotherapy was 19.0% (15/79) in the prevention group and 40.0% (28/70) in the non-prevention group, and the difference was statistically significant ( χ2=7.98, P=0.005). Conclusions:The preventive application of PEG-rhG-CSF can effectively reduce the incidence of neutropenia and better ensure the concurrent chemoradiotherapy in lung cancer patients on schedule.

In vitro studies have established the prevalent theory that the mitochondrial kinase PINK1 protects neurodegeneration by removing damaged mitochondria in Parkinson's disease (PD). However, difficulty in detecting endogenous PINK1 protein in rodent brains and cell lines has prevented the rigorous investigation of the in vivo role of PINK1. Here we report that PINK1 kinase form is selectively expressed in the human and monkey brains. CRISPR/Cas9-mediated deficiency of PINK1 causes similar neurodegeneration in the brains of fetal and adult monkeys as well as cultured monkey neurons without affecting mitochondrial protein expression and morphology. Importantly, PINK1 mutations in the primate brain and human cells reduce protein phosphorylation that is important for neuronal function and survival. Our findings suggest that PINK1 kinase activity rather than its mitochondrial function is essential for the neuronal survival in the primate brains and that its kinase dysfunction could be involved in the pathogenesis of PD.

Objective:The standard treatment for inoperable locally advanced esophageal cancer is concurrent chemoradiotherapy, but the survival was not satisfied. Nituzumab is a humanized IgG monoclonal antibody against EGFR. The purpose of this study is to investigate the toxicity and efficacy of concurrent chemoradiotherapy combined with nituzumab for locally advanced esophageal cancer.Methods:We retrospectively reviewed the clinical data of locally advanced esophageal cancer who were treated with concurrent chemoradiotherapy combined with nituzumab in Peking University Cancer Hospital from June 2015 to June 2020. Kaplan- Meier method was used for analysis. Results:Thirty Patients were enrolled this study.After a median follow-up of 22.5 months, The objective response rate was 93%. The 1-year, 2-year, 3-year overall survival rates were 83%, 57% and 41%, with the progression-free survival rates 75%, 47% and 32%, with the local-recurrence free survival rates 83%, 53% and 37%, with the metastasis-free survival rates 75%, 51% and 36%, respectively.The incidence of grade≥3 hematological toxicity was 32%. There were 16% patients experiencing grade≥3 esophagitis.Conclusion:The preliminary result of concurrent chemoradiotherapy combined with nituzumab is effective and safe for patients with locally advanced esophageal cancer.

Objective:To explore the performance of the commonly used whole blood C-reactive protein (CRP) detection systems and give related recommendation on the performance requirements of detection systems.Methods:A total of 7 540 venous blood samples from 26 maternal, child and children′s hospitals were collected to conduct this multi-center study on the analytical performance of 5 commonly used whole blood CRP detection systems from March to April in 2019. The blank check, carryover, repeatability, intermediate precision, linearity, sample stability, influence of hematocrit/triglyceride/bilirubin, comparison with SIEMENS specific protein analyzer and trueness were evaluated. The 5 systems included BC-5390CRP autohematology analyzer, AstepPLUS specific protein analyzer, Ottoman-1000 Automated Specific Protein POCT Workstation, i-CHROMA Immunofluorometer equipment Reader and Orion QuikRead go detecting instrument. The 5 systems were labeled as a, b, c, d and e randomly.Results:Within the 5 systems, all values of blank check were less than 1.00 mg/L, the carryovers were lower than 1.00%. The repeatability of different ranges of CRP concentrations including 3.00-10.00, 10.00-30.00 and>30.00 mg/L were less than 10.00%, 6.00% and 5.00%, respectively, and the intermediate precision was less than 10.00%. The linearity correlation coefficients of the 5 systems were all above 0.975, while the slope was within 0.950-1.050. Whole blood samples were stable within 72 hours both at room temperature (18-25 ℃) and refrigerated temperature (2-8 ℃). The CRP results were rarely influenced by high triglyceride or bilirubin, except for the immmunoturbidimetric test based on microparticles coated with anti-human CRP F(ab) 2 fragments. When triglyceride was less than 15.46 mmol/L, the deviation of CRP was less than 10.00%. When bilirubin was less than 345.47 μmol/L, the deviation of CRP was less than 10.00%. CRP was more susceptible to Hct on the systems without Hct correction. The deviation of CRP between different Hct dilution concentration and 40% dilution concentration can reach as high as 67.48%. The correlation coefficients ( r) of 5 systems were all more than 0.975 in the range of 0-300.00 mg/L compared with Siemens specific protein analyzer. All systems passed the trueness verification using the samples with specified values of 12.89 and 30.60 mg/L. Conclusion:The performance of 5 systems can basically meet the clinical needs, but it is suggested that the whole blood CRP detection system without automatic Hct correction should be modified manually.