In recent years, with the rapid development of artificial intelligence technology, the application of deep learning in the field of pathology has been continuously expanding. Particularly, the rise of multimodal data fusion methods has opened up new technical paths for the precise diagnosis, prognosis assessment, and individualized treatment of tumors. By integrating multi-level and multi-source data such as clinical information, pathological omics, molecular omics, and imaging omics, deep learning models can identify potential associated features and key biological mechanisms that are difficult to reveal by a single modality, thereby significantly improving the accuracy of disease classification and the scientific nature of risk stratification. This article systematically reviews the research progress of multimodal data fusion methods based on deep learning in the field of pathology in recent years, focuses on sorting out different types of fusion strategies, evaluates their advantages and challenges in practical clinical applications, and looks forward to future development trends.

Histological staining is the basis of pathological analysis,but the traditional staining method relies on chemical reagents,which not only consumes a lot of resources,but also causes harm to the environment and human health.In recent years,with the rapid development of deep learning technology,virtual staining technology,as a new method,is expected to effectively replace and supplement the traditional histological staining methods.It uses neural networks to analyze unstained tissue images,generate digital images that are highly similar to chemical staining effects,and even realize the mutual conversion between different staining modes,reducing the laboratory's dependence on chemical reagents and providing sustainable research programs.In this paper,the basic principles of virtual staining and its potential applications in histopathology are introduced in detail,and the current challenges and future research directions are discussed.

Histological staining is the basis of pathological analysis,but the traditional staining method relies on chemical reagents,which not only consumes a lot of resources,but also causes harm to the environment and human health.In recent years,with the rapid development of deep learning technology,virtual staining technology,as a new method,is expected to effectively replace and supplement the traditional histological staining methods.It uses neural networks to analyze unstained tissue images,generate digital images that are highly similar to chemical staining effects,and even realize the mutual conversion between different staining modes,reducing the laboratory's dependence on chemical reagents and providing sustainable research programs.In this paper,the basic principles of virtual staining and its potential applications in histopathology are introduced in detail,and the current challenges and future research directions are discussed.

In recent years, with the rapid development of artificial intelligence technology, the application of deep learning in the field of pathology has been continuously expanding. Particularly, the rise of multimodal data fusion methods has opened up new technical paths for the precise diagnosis, prognosis assessment, and individualized treatment of tumors. By integrating multi-level and multi-source data such as clinical information, pathological omics, molecular omics, and imaging omics, deep learning models can identify potential associated features and key biological mechanisms that are difficult to reveal by a single modality, thereby significantly improving the accuracy of disease classification and the scientific nature of risk stratification. This article systematically reviews the research progress of multimodal data fusion methods based on deep learning in the field of pathology in recent years, focuses on sorting out different types of fusion strategies, evaluates their advantages and challenges in practical clinical applications, and looks forward to future development trends.

Background::Programmed death 1 (PD-1) blockade plus chemotherapy has become the new first-line standard of care for patients with advanced non-small-cell lung cancer (NSCLC). Yet not all NSCLC patients benefit from this regimen. This study aimed to investigate the predictors of PD-1 blockade plus chemotherapy in untreated advanced NSCLC.Methods::We integrated clinical, genomic, and survival data from 287 patients with untreated advanced NSCLC who were enrolled in one of five registered phase 3 trials and received PD-1 blockade plus chemotherapy or chemotherapy alone. We randomly assigned these patients into a discovery cohort ( n = 125), a validation cohort ( n = 82), and a control cohort ( n = 80). The candidate genes that could predict the response to PD-1 blockade plus chemotherapy were identified using data from the discovery cohort and their predictive values were then evaluated in the three cohorts. Immune deconvolution was conducted using transcriptome data of 1014 NSCLC patients from The Cancer Genome Atlas dataset. Results::A genomic variation signature, in which one or more of the 15 candidate genes were altered, was correlated with significantly inferior response rates and survival outcomes in patients treated with first-line PD-1 blockade plus chemotherapy in both discovery and validation cohorts. Its predictive value held in multivariate analyses when adjusted for baseline parameters, programmed cell death ligand 1 (PD-L1) expression level, and tumor mutation burden. Moreover, applying both the 15-gene panel and PD-L1 expression level produced better performance than either alone in predicting benefit from this treatment combination. Immune landscape analyses revealed that tumors with one or more variation in the 15-gene panel were associated with few immune infiltrates, indicating an immune-desert tumor microenvironment.Conclusion::These findings indicate that a 15-gene panel can serve as a negative prediction biomarker for first-line PD-1 blockade plus chemotherapy in patients with advanced NSCLC.

Objective:To investigate the effects of miR-125b on radiosensitivity of cervical cancer cells and its possible downstream mechanism.Methods:The expression of miR-125b and Foxp3 in cervical cancer tissues and cells was detected by RT-qPCR. The HeLa cells were irradiated with 0, 2, 4 and 6 Gy of X-rays. The expression of miR-125b and Foxp3 in each group was detected by RT-qPCR. After downregulation of miR-125b expression and 6 Gy X-ray irradiation, the proliferation ability of HeLa cells was detected by MTT assay, and the expression of Bax and Bcl-2 proteins were detected by Western blot. The relationship between miR-125b and Foxp3 was detected by Targetscan and Dual luciferin reporter assay. After downregulation of Foxp3 expression and 6 Gy X-ray irradiation, the proliferation ability of HeLa cells was detected by MTT assay, and the expression of Bax and Bcl-2 proteins were detected by Western blot. The effects of miR-125b on radiosensitivity of HeLa cells through Foxp3 were detected. After down-regulation of Foxp3, the contents of IL-10 and TGF-β in supernatant were detected by ELISA.Results:The expression of miR-125b in the tissues and cells of cervical cancer was significantly decreased, while the expression of Foxp3 was significantly increased. The expression of miR-125b in HeLa cells was increased after radiation in a dose dependent manner. The expression of Foxp3 in HeLa cells was decreased after radiation in a dose dependent manner. After 6 Gy X-ray irradiation of HeLa cells, down-regulation of miR-125b increased the cell proliferation capacity, significantly reduced the expression of Bax and increased the expression of Bcl-2. miR-125b targets Foxp3 and negatively regulates Foxp3 expression. After 6 Gy X-ray irradiation of HeLa cells, down-regulation of Foxp3 significantly reduced the proliferation capacity of HeLa cells, increased the expression of Bax and decreased the expression of Bcl-2. Overexpression of miR-125b can enhance radiosensitivity of HeLa cells through Foxp3.After 6 Gy X-ray irradiation, down-regulation of Foxp3 reduced the expression of IL-10 and TGF-β in cells.Conclusions:Upregulation of miR-125b enhances the radiosensitivity of cervical cancer cells by targeting and negatively regulating Foxp3, and the mechanism of that may be related to the down-regulation of Foxp3 to reduce the expression of IL-10 and TGF-β in the cells.

Objective:To investigate the expression and diagnostic value of SS18-SSX fusion-specific antibody and SSX C-terminal antibody in synovial sarcoma (SS).Methods:Immunohistochemical (IHC) EnVision method was used to detect the expression of SS18-SSX fusion-specific antibody and SSX C-terminal antibody in 51 genetically confirmed cases of SS and 94 non-SS tumors diagnosed at Nanjing Jinling Hospital from August 2013 to December 2020.Results:IHC staining for SS18-SSX fusion-specific antibody revealed strongly diffuse nuclear staining in 48 of 51 (48/51, 94.1%) SS cases, whereas none of the 94 non-SS tumors showed any staining. IHC staining for SSX C-terminal antibody showed strongly diffuse nuclear staining in all 51 (51/51, 100%) SS cases; six of the 94 (6/94, 6.4%) non-SS tumors showed variable staining, including two cases each of leiomyosarcoma and fibrosarcoma, and one case each of malignant peripheral nerve sheath tumor and embryonal rhabdomyosarcoma. The sensitivity and specificity of SS18-SSX fusion-specific antibody in diagnosing SS were 94.1% and 100% and these of SSX C-terminal antibody were 100% and 93.6%, respectively.Conclusions:SS18-SSX fusion-specific antibody and SSX C-terminal antibody are highly sensitive and specific markers for SS. Immunohistochemistry using these antibodies may replace FISH or molecular genetic testing in most cases.

Objective: To investigate the expression of H3.3 G34W mutant-specific antibody in giant cell tumors of bone (GCTB), and its value in the diagnosis of GCTB. Methods: Immunohistochemical (IHC) EnVision method was used to detect the expression of H3.3 G34W mutant-specific antibody and p63 in 83 GCTBs, 18 aneurysmal bone cysts, 23 chondroblastomas and 28 osteosarcomas diagnosed at Nanjing Jinling Hospital from June 2001 to April 2019. Results: Among the 83 cases of GCTB, 69 cases (69/83, 83.1%) expressed H3.3 G34W. H3.3 G34W expression was found exclusively in the mononuclear cell population with strong and diffuse nuclear staining. H3.3 G34W was expressed in 55 of 57 (96.5%) cases of GCTB in long bones, but only 14 of 26 (53.8%) cases of non-long bone GCTB. All recurrent (9/9)/metastatic GCTB (2/2), post-denosumab GCTB (3/3), primary malignant GCTB (3/3) and secondary malignant GCTB (5/5) also expressed H3.3 G34W. H3.3 G34W was negative in all aneurysmal bone cysts and chondroblastomas. H3.3 G34W was positive in 3 of 28(10.7%) cases of osteosarcomas, and giant cell-rich osteosarcoma(GCRO) was the only histological subtype of osteosarcoma that expressed H3.3 G34W. p63 was expressed in 71.1%(59/83) of GCTB, while the positive rates of p63 in aneurysmal bone cysts,chondroblastomas and osteosarcomas were 3/18, 43.5% (10/23) and 21.4% (6/28) respectively. The sensitivity and specificity of H3.3 G34W mutant-specific antibody in the diagnosis of GCTB were 83.1% and 95.7%. Conclusions H3.3 G34W mutant-specific antibody is a highly sensitive and specific marker for GCTB and helpful for the diagnosis of GCTB and its variants. The limitation of this antibody is that as a mall number of GCTB harbor G34 mutation other than G34W, and thus that cannot be detected. The incidental expression of H3.3 G34W mutant protein in osteosarcoma could be a potential diagnostic dilemma, and the results of H3.3 G34W IHC staining needs careful interpretation.

Objective: To study the clinicopathological features, immunohistochemical phenotype, molecular changes, differential diagnosis and prognosis of eosinophilic solid and cystic renal cell carcinoma (ESC RCC). Methods: A total of 15 cases were selected from 2005 to 2019 at Nanjing Jinling Hospital,Nanjing University School of Medicine for clinicopathological and immunohistochemical analysis, 10 of which were subject to cancer-associated mutation analysis using targeted next-generation sequencing (NGS) panel. A literature review was also performed. Results: The patients′ ages ranged from 15 to 68 years (mean, 33 years). The male-to-female ratio was 1.1∶1.0. During a mean follow-up of 22 months, none of the patients developed tumor recurrence, progression or metastasis. Histologically, the tumors typically demonstrated solid and cystic architectures and the neoplastic cells contained voluminous eosinophilic cytoplasm with prominent granular cytoplasmic stippling. Immunohistochemically, tumor cells in all cases were immunoreactive for CK20. Signal pathway related protein mTOR and S6 were positive in 14/15 and 6/15 cases, respectively. Cathepsin K, Melan A and HMB45 were at least focally positive in 12/15, 6/15 and 2/15 cases, respectively. CK7 and CD10 showed focal immunostain positivity in some cases, while TFE3, TFEB, CA9 and CD117 were negative in all cases. NGS demonstrated TSC1/TSC2 mutations in all tested cases (10/10). Conclusions ESC RCC is a rare tumor that tends to occur in young patients with an indolent behavior. Diagnosis can be established by its distinct clinical and histopathologic findings, immunohistochemical phenotype and molecular genetics. The tumor may be considered as a new subtype of RCC.

study the clinicopathological features, immunohistochemical phenotype, molecular changes, differential diagnosis and prognosis of eosinophilic solid and cystic renal cell carcinoma (ESC RCC). Methods A total of 15 cases were selected from 2005 to 2019 at Nanjing Jinling Hospital,Nanjing University School of Medicine for clinicopathological and immunohistochemical analysis, 10 of which were subject to cancer?associated mutation analysis using targeted next?generation sequencing (NGS) panel. A literature review was also performed. Results The patients′ ages ranged from 15 to 68 years (mean, 33 years). The male?to?female ratio was 1.1∶1.0. During a mean follow?up of 22 months, none of the patients developed tumor recurrence, progression or metastasis. Histologically, the tumors typically demonstrated solid and cystic architectures and the neoplastic cells contained voluminous eosinophilic cytoplasm with prominent granular cytoplasmic stippling. Immunohistochemically, tumor cells in all cases were immunoreactive for CK20. Signal pathway related protein Mtor and S6 were positive in 14/15 and 6/15 cases, respectively. Cathepsin K, Melan A and HMB45 were at least focally positive in 12/15, 6/15 and 2/15 cases, respectively. CK7 and CD10 showed focal immunostain positivity in some cases, while TFE3, TFEB, CA9 and CD117 were negative in all cases. NGS demonstrated TSC1/TSC2 mutations in all tested cases (10/10). Conclusions ESC RCC is a rare tumor that tends to occur in young patients with an indolent behavior. Diagnosis can be established by its distinct clinical and histopathologic findings, immunohistochemical phenotype and molecular genetics. The tumor may be considered as a new subtype of RCC.