With the rapid development of artificial intelligence, computational pathology has been seamlessly integrated into the entire clinical workflow, which encompasses diagnosis, treatment, prognosis, and biomarker discovery. This integration has significantly enhanced clinical accuracy and efficiency while reducing the workload for clinicians. Traditionally, research in this field has depended on the collection and labeling of large datasets for specific tasks, followed by the development of task-specific computational pathology models. However, this approach is labor intensive and does not scale efficiently for open-set identification or rare diseases. Given the diversity of clinical tasks, training individual models from scratch to address the whole spectrum of clinical tasks in the pathology workflow is impractical, which highlights the urgent need to transition from task-specific models to foundation models (FMs). In recent years, pathological FMs have proliferated. These FMs can be classified into three categories, namely, pathology image FMs, pathology image-text FMs, and pathology image-gene FMs, each of which results in distinct functionalities and application scenarios. This review provides an overview of the latest research advancements in pathological FMs, with a particular emphasis on their applications in oncology. The key challenges and opportunities presented by pathological FMs in precision oncology are also explored.
Humans ; Precision Medicine/methods* ; Medical Oncology/methods* ; Artificial Intelligence ; Neoplasms/pathology* ; Computational Biology/methods*

Objective To analyze dynamic contrast-enhanced MRI (DCE-MRI) radiomic features using machine learning algorithms, and to develop and validate a predictive model for HER-2 status in breast cancer. Methods The DCE-MRI images of 272 treatment-naive female patients with breast cancer between 2020 and 2022 were included in this study. Regions of interest (ROIs) were manually segmented using 3d-Slicer software, and radiomic features were extracted. All patients were randomly divided into training sets or validation sets at a ratio of 4∶1. The least absolute shrinkage and selection operator (LASSO) algorithm was used for feature screening on the training set, followed by the development of predictive models using six machine learning algorithms. Internal cross-validation was performed to compare the performance differences between the models. The best-performing model was selected, trained on the training set, and evaluated on the validation set. Evaluation metrics included area under the curve (AUC), sensitivity, specificity, precision, and recall rate. Results The clinical data of patients in the training set and validation set showed no significant differences. Five features were identified by the LASSO algorithm. With these features, six machine learning models were developed on the training set, and their predictive performance was internally cross-validated using the bagging method. XGBoost model had the highest mean AUC (0.696), followed by RF model (0.690); XGBoost model had the highest mean precision (0.756), followed by LR and RF models. Therefore, XGBoost was the optimal model. An HER-2 predictive model was built using the XGBoost algorithm on the training set and applied to the validation set. The AUC, precision, sensitivity, and specificity of the predictive model on the validation set were calculated, and ROC curves, precision-recall curves, calibration curves, and decision-making curves were plotted. Conclusion This study constructed and evaluated different DCE-MRI radiomics-based machine learning models for predicting HER-2 status in breast cancer. Among them, XGBoost algorithm performed the best and has the potential to become a new non-invasive method for preoperative prediction of HER-2 status, providing reliable evidence for personalized clinical diagnosis and treatment.

Objective:To understand the genetic characteristics of varicella-zoster virus（VZV）prevalent in Qinghai province，China since 2020.Methods:A total of 54 pharyngeal swab specimens were collected from sporadic suspected varicella cases in Qinghai province in 2020，2023，and 2024. Real-time fluorescence quantitative polymerase chain reaction was used for etiological screening of the specimens. Sequencing of three genes，namely ORF22，ORF38，and ORF62，and single-nucleotide polymorphism（SNP）analysis were performed on VZV nucleic acid-positive specimens.Results:All 54 suspected varicella cases were diagnosed with VZV infection，and three gene sequences were successfully obtained from 53 specimens. The results of genotype identification showed that all VZV infection case specimens obtained in this study in Qinghai province were wild strains. Among them，4 specimens in 2020 were of clade 2 type；among 14 specimens in 2023，7 were of clade 2 type and the remaining 7 were of clade 5 type；among 35 specimens in 2024，27 were of clade5 type，5 were of clade 2 type，and 3 were of clade 4 type. The SNP results showed that in 2023 and 2024，one specimen each had an A→G base mutation at position 37 990，and in 2024，3 specimens had a T→C base mutation at position 37946. Among them，the sequences containing the former mutation have been prevalent and spread in multiple regions of China，and the latter has not been reported in other regions of China.Conclusion:From 2020 to 2024 in Qinghai province，at least three genotypes of VZV，namely clade 2 type，clade 5 type，and clade 4 type，co-prevailed，and the clade 5 genotype of VZV may become the dominant prevalent strain.

Objective:To explore the clinical features and molecular characteristics of myeloproliferative neoplasms (MPNs) patients with NFE2 gene mutations.Methods:Gene targeted sequencing was used to detect NFE2 gene mutation in 723 patients diagnosed with MPNs who were admitted to Institute of Hematology & Blood Diseases Hospital, Chinese Academy of Medical Sciences & Peking Union Medical College between April 2021 and June 2023. The association between NFE2 gene mutations and clinical features and molecular characteristics of MPNs patients were retrospectively analyzed.Results:Among 723 patients with MPNs, NFE2 gene mutations were found in 41 cases (5.7%) . NFE2 gene mutations were predominantly frameshift mutations (44.4%) , followed by nonsense mutations (33.3%) . The median number of mutations in patients with NFE2 gene mutations (4 [2,5]) was higher compared to the group without NFE2 gene mutations (2, [1,3]) ( P<0.001) . NFE2 gene mutations frequently co-occurred with mutations in MPL, ATM, PPM1D, and TET1. NFE2 gene mutations were mostly sub-clonal events, with 80.5% occurring after MPNs driver mutations (JAK2, CALR, or MPL) . NFE2 mutations were correlated with older age [median age: 60 (54, 67) years vs 54 (41, 63) years, P=0.001]. Patients with NFE2 gene mutations had a higher incidence of pre-diagnosis thrombosis (39.0% vs 22.0%, P=0.012) and pre-diagnosis arterial thrombosis (36.6% vs 20.4%, P=0.014) . Using a logistic regression analysis model adjusting for age and comorbidities (including chronic infections, malignancies, and autoimmune diseases) , NFE2 gene mutation was identified as an independent determinant of elevated tumor necrosis factor-alpha (TNF-α) ( OR=2.747, 95% CI: 1.143-6.605, P=0.024) , interferon-gamma (IFN-γ) ( OR=2.689, 95% CI: 1.191-6.076, P=0.017) , IL-10 ( OR=3.219, 95% CI: 1.343-7.717, P=0.009) , IL-12P70 ( OR=3.397, 95% CI:1.003-11.508, P=0.049) , IL-17 ( OR=2.284, 95% CI: 1.017-5.127, P=0.045) . In polycythaemia vera (PV) patients with the NFE2 gene mutation, the proportion of those classified as high-risk is notably higher in both the IWG-PV and mutation-enhanced international prognostic systems for PV (MIPSS-PV) (66.7% vs 25.3% for IWG-PV, P=0.033; 22.2% vs 2.0% for MIPSS-PV, P=0.013) . Similarly, for essential thrombocythaemia (ET) patients, the proportion in the high-risk group of the mutation-enhanced international prognostic systems for ET (MIPSS-ET) is significantly higher (15.4% vs 6.1%, P=0.021) . No statistically significant differences were observed in overall survival or cumulative incidence of thrombosis between NFE2-mutated (38 cases) and non-mutated MPNs patients (671 cases, P>0.05) . Conclusion:NFE2 gene mutations in MPNs were predominantly frameshift mutations. NFE2 gene mutations were correlated with older age, elevated levels of several inflammatory factors (including TNF-α、IFN-γ、IL-10、IL-12P70、IL-17) , and they mostly occurred in late-stage of MPNs.

Objective To compare the node strength in white matter networks between depressive disorder and bipolar depression patients,analyze structural connectivity impairments across brain regions,and assess their diagnostic utility.Methods This longitudinal study initially enrolled 91 patients with a baseline diagnosis of depressive episode.All subjects underwent diffusion tensor imaging at recruitment and white matter structural weighted networks were constructed using deterministic fiber tracking.After≥9 years of naturalistic follow-up,23 patients who maintained a diagnosis of major depressive disorder(MDD group)and 18 patients who maintained a diagnosis of bipolar disorder(BD group),while 30 demographically matched healthy controls(HC group)were included for comparison.The differences in nodal connection strength within the brain white matter networks among the three groups were compared.Furthermore,the receiver operator characteristic(ROC)curve was utilized to evaluate the diagnostic value of the differential brain regions in distinguishing between MDD and BD.Results In the left anterior cingulate gyrus,the node strength was lower in the BD than in the MDD group(3.89±0.76 vs.4.74±0.60).However,the node strength in the right caudate nucleus(4.94±1.26 vs.3.46±0.99)and right globus pallidus(1.98±0.67 vs.1.25±0.29)was higher in the BD than in the MDD group(P<0.01,FWE-corrected).The combined connectivity strengths of three brain regions—the left anterior cingulate gyrus,right caudate nucleus and right globus pallidus—were used to differentiate between MDD and BD,and an ROC curve was plotted,with an area under the curve(AUC)of 0.95(95%CI:0.91～0.99;P<0.001).The sensitivity and the specificity were 0.89 and 0.87,respectively.Conclusion The differences in node strength between patient groups may serve as a potential neuroimaging biomarker.Integration of node connectivity strengths from these differential brain regions can achieve superior discrimination accuracy.

Objective:To explore behavioral and electrophysiological differences in risky decision-making between depressed patients with and without suicidal ideation.Methods:A total of 61 patients with first-episode untreated depression were enrolled in the depression clinic of Nanjing Brain Hospital from September 2023 to January 2024, which were divided into the suicidal ideation group( n=32) and the non-suicidal ideation group ( n=29).At the same time, healthy controls matched with sex, age and years of education were recruited from the community( n=36).The event-related potentials (ERP) of the participants were detected, and the amplitude and latency of feedback related negative waves (FRN) and P300 during the feedback phase under Iowa gambling task (IGT) were recorded. Statistical analysis was performed using SPSS 26.0 software.The inter-and intra-group differences of ERP indexes were compared using two-way ANOVA, and Spearman correlation analysis was conducted to examine the relationship between ERP indexes and scores of the Beck scale for suicidal ideation. Results:(1)Compared with healthy controls, depressed patients with and without suicidal ideation had both lower net scores in IGT (both P<0.05).(2)When comparing the mean FRN amplitude under different feedback types among the three groups, the main effect of feedback type ( F=8.799, P=0.004), the main effect of group ( F=6.396, P=0.002) and the interaction effect ( F=4.200, P=0.018)were all significant. Under gain feedback conditions, the mean FRN amplitude was lower in both depressed groups compared with healthy controls (both P<0.05). (3)The comparison of the mean P300 amplitude under different feedback types among the three groups showed that the main effect of group ( F=15.719, P<0.001) and the main effect of feedback type ( F=15.949, P=0.001) were both significant, while the interaction effect between group and feedback type was not significant ( F=1.573, P=0.213). The group with suicidal ideation ((0.85±0.21) μV) had a smaller amplitude than both the non-suicidal ideation group ((1.61±0.22) μV) and healthy controls ((2.46±0.20) μV) (both P<0.05). (4)In depressed patients, P300 mean amplitude under both loss and gain feedback conditions were both negatively correlated with suicidal ideation (loss: r=-0.435, P=0.001; gain: r=-0.318, P=0.013). Conclusion:Depressed patients with and without suicidal ideation both exhibit impaired risk decision-making. The decrease of P300 mean amplitude is more significant in depressed patients with suicidal ideation than those without suicidal ideation.P300 mean amplitude may serve as an electrophysiological marker to differentiate depressed patients with suicidal ideation and those without suicidal ideation.

Objective:To analyze the characteristics of electroencephalogram microstate parameters in first-episode drug-naive patients with major depressive disorder (MDD), so as to provide electrophysiological evidence for the pathogenesis and early diagnosis of MDD.Methods:Eighty-four first-episode, drug-naive outpatients diagnosed with MDD(MDD group) and 82 healthy controls(healthy group) participated in this study. Resting-state EEG data (5-6 min, with eyes closed) were recorded for all participants. Data preprocessing and microstate analysis were performed using MATLAB and EEGLAB software. Temporal parameters of resting-state brain network microstates were compared using SPSS 26.0.Results:This study identified four typical microstates: Class A microstate(auditory network), Class B microstate(visual network), Class C microstate(salient network), and Class D microstate(attention and control network). The coverage rate (0.16±0.06, 0.21±0.06), duration (67.72±7.07, 72.28±8.59), and incidence rate (2.38±0.68, 2.82±0.67) of microstate A in MDD group were significantly lower than those in healthy group ( F=22.115, 13.368, 18.779, all P<0.001), while the above indexes of microstate B in MDD group were significantly higher than those in healthy group(coverage rate: 0.24±0.07 vs 0.18±0.06, duration: 76.35±11.28 vs 69.46±8.52, incidence rat: 3.16±0.52 vs 2.52±0.57) ( F=41.287, 18.999, 52.245, all P<0.001). Additionally, the microstate D in MDD group showed significantly lower coverage rate(0.33±0.08, 0.36±0.08) and duration (89.66±15.38, 95.46±16.79)compared with healthy group( F=3.932, 4.215, both P<0.05). Notably, significant differences were observed in the transition probabilities between the following microstates: A→B, A→D, B→A, C→A, C→B, D→A and D→B (all P<0.05). Conclusion:First-episode drug-naive depressive patients are characterized by alterations in microstate A, microstate B, and microstate D, which may be the potential pathogenesis of MDD and may serve as electrophysiological indicators for early diagnosis of MDD.

Objective:To analyze the iodine nutrition status of children and adolescents and influencing factors in Zhejiang Province, providing scientific basis for optimizing iodine deficiency disorders (IDD) prevention and control strategies.Methods:In June 2022, a multistage stratified sampling method was used to divide 16 counties (cities, districts, abbreviated as counties) in Zhejiang Province into three categories based on their geographical locations (average distance from the coastline): coastal areas (including Dinghai District, Jiaojiang District, Sanmen County, Cixi City and Lucheng District), sub-coastal areas (including Wuxing District, Haining City, Linping District, Fuyang District and Fenghua District), and inland areas(including Suichang County, Changshan County, Shengzhou City, Jindong District, Dongyang City and Yongjia County). One county was selected from each category, and one township (street) was selected from each county. Two administrative villages (neighborhood committees) were selected from each township (street). Ten households including all children and adolescents aged 6-17 in each household were selected from each administrative village (neighborhood committee). Demographic information and personal dietary characteristics were collected via questionnaires, while household salt and a random urine sample were tested for iodine level. Trend analysis was conducted using a χ 2trend test, and a multivariate logistic stepwise regression model was used to analyze the influencing factors of urinary iodine levels. Results:A total of 755 children and adolescents aged 6-17 were selected, including 387 males (51.26%) and 368 females (48.74%), with an age of (11.24 ± 3.32) years. There were 269 children and adolescents in coastal areas (35.63%) and 409 children and adolescents in urban areas (54.17%). A total of 755 household salt samples were collected, with a median salt iodine concentration of 21.80 mg/kg. These included 263 non-iodized salt samples, 38 unqualified iodized salt samples, and 454 qualified iodized salt samples. The coverage rate of iodized salt was 65.17% (492/755), and the consumption rate of qualified iodized salt was 60.13% (454/755). The distribution of salt iodine quality among children and adolescents in different geographical locations showed statistically significant differences (χ 2 = 111.95, P < 0.001), with the proportion of non-iodized salt gradually decreasing from coastal areas to inland areas (χ 2trend = 90.17, P < 0.001). A total of 755 urine samples were collected, with a median urinary iodine concentration of 186.60 μg/L. The proportions of urinary iodine < 100, 100-199, 200-299, and ≥300 μg/L were 16.95% (128/755), 37.62% (284/755), 24.37% (184/755), and 21.06% (159/755), respectively. The χ 2trend test revealed a nonlinear positive correlation between salt iodine level and urinary iodine level (χ 2regression = 21.98, P < 0.001; χ 2partial = 6.96, P < 0.001). The frequency distribution of urinary iodine in children and adolescents from different geographical locations and between urban and rural areas showed statistically significant differences (χ 2 = 29.63, 16.56, P < 0.001). Among them, the proportion of children and adolescents with urinary iodine < 100 μg/L gradually decreasing from coastal areas to inland areas (χ 2trend = 6.15, P = 0.013). The results of multivariate logistic regression analysis revealed that sub-coastal regions, inland regions, and urban-rural regions ( OR = 1.57, 1.53, 1.64, 95% CI: 1.11-2.24, 1.03-2.27, 1.17-2.32, P < 0.05) were significantly associated with urinary iodine levels in children and adolescents aged 6-17. Conclusions:In 2022, the iodine nutrition of children and adolescents in Zhejiang Province is generally suitable, but there is a risk of iodine deficiency among coastal children and adolescents. Geographic location and urban/rural areas are influencing factors on iodine nutrition status of children and adolescents in Zhejiang Province.

Objective:To investigate the effect of childhood trauma on non-suicidal self-injury (NSSI) behavior in patients with depression.Method:Embase, PubMed, Cochrane Library, PsycINFO, China National Knowledge Infrastructure, Wanfang Data and China Biology Medicine dis were searched from inception to March 2024 for cross-sectional, case-control and cohort studies on childhood trauma and NSSI in patients with depression. Two researchers independently screened studies, extracted data, and assessed quality. The effect indicators were the odds ratio ( OR) of childhood trauma and school bullying to NSSI in the depressed population and the mean difference ( MD) of the childhood trauma subscale scores between the depressed population with and without NSSI. Meta-analysis was performed using Review Manager 5.3 and Stata17 software. Results:A total of 29 articles with 5 095 depressed patients were included. Childhood trauma was significantly associated with NSSI in patients with depression ( OR=2.91, 95% CI=2.01-4.21). Various forms of childhood trauma were related to NSSI behaviors in depressive patients: physical abuse ( MD=0.77, 95% CI=0.47-1.06), emotional abuse ( MD=2.99, 95% CI=2.10-3.88), physical neglect ( MD=1.17, 95% CI=0.47-1.87), emotional neglect ( MD=2.59, 95% CI=1.82-3.36), and sexual abuse ( MD=0.35, 95% CI=0.19-0.51). Additionally, school bullying among extra-family factors was identified as a risk factor for NSSI ( OR=2.16, 95% CI=1.46-3.18). Conclusion:Childhood trauma is a risk factor for NSSI behaviors in patients with depression. Different types of childhood trauma within the family, including various forms of abuse and neglect, and school bullying outside the family are related to NSSI behaviors in this population.

Objective:This study aims to analyze the gamma band effective connectivity characteristics of the prefrontal-striatal circuitry in bipolar disorder patients with and without a history of manic episodes, as well as in major depressive disorder patients, during the recognition of positive emotional faces, this study aims to identify unique neurophysiological features that may aid in the early detection of bipolar disorder.Methods:This retrospective study collected clinical data and magnetoencephalography (MEG) imaging data from patients performing a positive emotional face recognition task at the Affiliated Brain Hospital of Nanjing Medical University from May 2009 to December 2019. The study included 75 patients with major depressive disorder and 29 patients with bipolar disorder in a depressive episode (rBD group). Concurrently, 39 age-and gender-matched healthy controls (HC group) were recruited. After a follow-up period of at least 5 years, 23 out of the 75 patients with major depressive disorder converted to bipolar disorder (ctBD group), while the remaining 52 who did not convert maintained a diagnosis of major depressive disorder.Results:There were statistically significant differences in gamma-band effective connectivity in the prefrontal-striatal circuit when recognizing positive emotional faces among the converted to bipolar disorder (ctBD), raw bipolar disorder, major depressive disorder, and HC groups ( H=9.04, 10.30, 8.30, 13.43, 14.38, 12.62, 9.82, 8.94, 24.62, 7.89, 18.53, 9.97, 9.58, 12.79, P<0.05). The ctBD group, rBD group, and major depressive group all showed reduction in effective connectivity from the right orbital inferior frontal gyrus (ORBinf.R) to the left orbital inferior frontal gyrus (ORBinf.L) [ Z=-1.98, -3.38, -2.88], from the right orbital inferior frontal gyrus to the right ventral striatum (VS.R) ( Z=-2.05, -2.76, -2.11; P<0.05) and from the left ventral striatum (VS.L) to the left orbital middle frontal gyrus (ORBmid.L) ( Z=-2.76, -1.98, -2.43; P<0.05). Among the disease groups, the ctBD group showed significantly enhanced effective connectivity strength compared to the major depressive group from the right amygdala (AMYG.R) to the left orbital inferior frontal gyrus(0.04(0.03, 0.08)), from the right amygdala to the left ventral striatum(0.05(0.03, 0.09)), and from the right ventral striatum to the right anterior cingulate and paracingulate gyri (ACG.R) (0.04(0.02, 0.08)) ( Z=4.17, 3.70, 3.35; P<0.001).The ctBD group also exhibited enhanced effective connectivity compared to the rBD group from ORBinf.R to the ACG.R, fron the AMYG.R to the ORBinf.L, from the AMYG.R to the VS.L, and from the VS.R to the ACG.R ( Z=2.05, 4.61, 3.60, 3.04; P<0.05).The rBD group demonstrated reduced effective connectivity compared to the major depressive disorder group from the right orbital middle frontal gyrus(ORBmid.R) to the left anterior cingulate and paracingulate gyri (ACG.L), ORBinf.R to the ACG.R and from the ORBinf.R to the AMYG.R ( Z=-2.12, -2.40, -2.22; P<0.05). Conclusion:There are significant differences in the gamma-band effective connectivity characteristics of the prefrontal-striatal pathway when recognizing positive emotional faces between patients with bipolar disorder in depressive episodes and those with depression, as well as differences between bipolar depressed patients with and without a history of manic episodes.