Improving quantitative prediction of protein subcellular locations in fluorescence images through deep generative models.

Machine learning has been employed in recognizing protein localization at the subcellular level, which highly facilitates the protein function studies, especially for those multi-label proteins that localize in more than one organelle. However, existing works mostly study the qualitative classification of protein subcellular locations, ignoring fraction of one multi-label protein in different locations. Actually, about 50 % proteins are multi-label proteins, and the ignorance of quantitative information highly restricts the understanding of their spatial distribution and functional mechanism. One reason of the lack of quantitative study is the insufficiency of quantitative annotations. To address the data shortage problem, here we proposed a generative model, PLocGAN, which could generate cell images with conditional quantitative annotation of the fluorescence distribution. The model was a conditional generative adversarial network, in which the condition learning utilized partial label learning to overcome the lack of training labels and allowed training with only qualitative labels. Meanwhile, it used contrastive learning to enhance diversity of the generated images. We assessed the PLocGAN on four pixel-fused synthetic datasets and one real dataset, and demonstrated that the model could generate images with good fidelity and diversity, outperforming existing state-of-the-art generative methods. To verify the utility of PLocGAN in the quantitative prediction of protein subcellular locations, we replaced the training images with generated quantitative images and built prediction models, and found that they had a boosting effect on the quantitative estimation. This work demonstrates the effectiveness of deep generative models in bioimage analysis, and provides a new solution for quantitative subcellular proteomics.

A comprehensive analysis of GAS2 family members identifies that GAS2L1 is a novel biomarker and promotes the proliferation of hepatocellular carcinoma.

Hepatocellular carcinoma (HCC) is a common primary liver cancer with a high incidence and mortality. Members of the growth-arresting-specific 2 (GAS2) family are involved in various biological processes in human malignancies. To date, there is only a limited amount of information available about the expression profile and clinical importance of GAS2 family in HCC. In this study, we found that GAS2L1 and GAS2L3 were distinctly upregulated in HCC specimens compared to non-tumor specimens. Pan-cancer assays indicated that GAS2L1 and GAS2L3 were highly expressed in most cancers. The Pearson's correlation revealed that the expressions of GAS2, GAS2L1 and GAS2L2 were negatively associated with methylation levels. Survival assays indicated that GAS2L1 and GAS2L3 were independent prognostic factors for HCC patients. Immune cell infiltration analysis revealed that GAS2, GAS2L1 and GAS2L3 were associated with several immune cells. Finally, we confirmed that GAS2L1 was highly expressed in HCC cells and its knockdown suppressed the proliferation of HCC cells. Taken together, our findings suggested the expression patterns and prognostic values of GAS2 members in HCC, providing insights for further study of the GAS2 family as sensitive diagnostic and prognostic markers for HCC.

The role of autophagy regulated by the PI3K/AKT/mTOR pathway and innate lymphoid cells in eosinophilic chronic sinusitis with nasal polyps.

BACKGROUND: The PI3K/Akt/mTOR pathway and autophagy are important physiological processes. But their roles in eCRSwNP remains controversial. METHODS: In this study, we used the eCRSwNP mouse model, PI3K/Akt/mTOR pathway inhibitors, and autophagy inhibitors and activators to investigate the regulatory effects of the PI3K/Akt/mTOR pathway on autophagy, and their effects on eosinophilic inflammation, and tissue remodeling. The role of ILC2s in eCRSwNP was also studied, and the relationship between ILC2s and autophagy was preliminarily determined. RESULTS: Our results show that eosinophilic inflammation in eCRSwNP mice could be inhibited by promoting the autophagy; otherwise, eosinophilic inflammation could be promoted. Meanwhile, inhibition of the PI3K/Akt/mTOR pathway can further promote autophagy and inhibit eosinophilic inflammation. Meanwhile, inhibiting the PI3K/Akt/mTOR pathway and promoting autophagy can reduce the number of ILC2s and the severity of tissue remodeling in the nasal polyps of eCRSwNP mice. CONCLUSIONS: We conclude that the PI3K/Akt/mTOR pathway plays roles in eosinophilic inflammation and tissue remodeling of eCRSwNP, in part by regulating the level of autophagy. The downregulation of autophagy is a pathogenesis of eCRSwNP; therefore, the recovery of normal autophagy levels might be a new target for eCRSwNP therapy. Furthermore, autophagy might inhibit eosinophilic inflammation and tissue remodeling, in part by reducing the number of ILC2s.

Association analyses between the variants of SNAP25, SV2C and ST3GAL2 and the efficacy of botulinum toxin A in the treatment of the primary Meige syndrome.

Objective: Individual differences were observed in the clinical efficacy of Botulinum toxin A (BoNT-A) in the treatment of the primary Meige syndrome. Our study aimed to explore the potential associations between the clinical efficacy of BoNT-A in the treatment of the primary Meige syndrome and variants of SNAP25, SV2C and ST3GAL2, which are involving in the translocation of the BoNT-A in vivo. Methods: Patients with the primary Meige syndrome treated with BoNT-A were enrolled. Clinical efficacy was evaluated by the maximum improvement rate of motor symptoms and the duration of efficacy. Variants of SNAP25, SV2C and ST3GAL2 were obtained by Sanger sequencing. Another cohort diagnosed with primary cervical dystonia was also enrolled in the replication stage. Results: Among the 104 primary Meige syndrome patients, 80 patients (76.9%) had a good efficacy (the maximum improvement rate of motor symptoms ≥30%) and 24 (23. 1%) had a poor (the maximum improvement rate of motor symptoms <30%). As to the duration of efficacy, 52 patients (50.0%) had a long duration of efficacy (≥4 months), and 52 (50.0%) had a short (<4 months). In terms of primary Meige syndrome, SNAP25 rs6104571 was found associating with the maximum improvement rate of motor symptoms (Genotype: P = 0.02, OR = 0.26; Allele: P = 0.013, OR = 0.29), and SV2C rs31244 was found associating with the duration of efficacy (Genotype: P = 0.024, OR = 0.13; Allele: P = 0.012, OR = 0.13). Besides, we also conducted the association analyses between the variants and BoNT-A-related adverse reactions. Although, there was no statistical difference between the allele of SV2C rs31244 and BoNT-A-related adverse reactions, there was a trend (P = 0.077, OR = 2.56). In the replication stage, we included 39 patients with primary cervical dystonia to further expanding the samples' size. Among the 39 primary cervical dystonia patients, 25 patients (64.1%) had a good efficacy (the maximum improvement rate of motor symptoms ≥50%) and 14 (35.9%) had a poor (the maximum improvement rate of motor symptoms <50%). As to the duration of efficacy, 32 patients (82.1%) had a long duration of efficacy (≥6 months), and 7 (17.9%) had a short (<6 months). Integrating primary Meige syndrome and primary cervical dystonia, SV2C rs31244 was still found associating with the duration of efficacy (Genotype: P = 0.002, OR = 0. 23; Allele: P = 0.001, OR = 0. 25). Conclusion: In our study, SNAP25 rs6104571 was associated with the maximum improvement rate of motor symptoms in patients with primary Meige syndrome treated with BoNT-A, and patients carrying this variant had a lower improvement rate of motor symptoms. SV2C rs31244 was associated with duration of treatment in patients with primary Meige syndrome treated with BoNT-A and patients carrying this variant had a shorter duration of treatment. Patients with primary Meige syndrome carrying SV2C rs31244 G allele have an increase likelihood of BoNT-A-related adverse reactions. Involving 39 patients with primary cervical dystonia, the results further verify that SV2C rs31244 was associated with duration of treatment and patients carrying this variant had a shorter duration of treatment.

Effectiveness and tolerability of programmed cell death protein-1 inhibitor + chemotherapy compared to chemotherapy for upper gastrointestinal tract cancers.

BACKGROUND: The combination of programmed cell death protein-1 (PD-1) inhibitor and chemotherapy is approved as a standard first- or second-line treatment in patients with advanced oesophageal or gastric cancer. However, it is unclear whether this combination is superior to chemotherapy alone. AIM: To assess the comparative effectiveness and tolerability of combining PD-1 inhibitors with chemotherapy vs chemotherapy alone in patients with advanced gastric cancer, gastroesophageal junction (GEJ) cancer, or oesophageal carcinoma. METHODS: We searched the PubMed and Embase databases for studies that compared the efficacy and tolerance of PD-1 inhibitors in combination with chemotherapy vs chemotherapy alone in patients with advanced oesophageal or gastric cancer. We employed either random or fixed models to analyze the outcomes of each clinical trial, encompassing data on overall survival (OS), progression-free survival (PFS), objective response rate, and adverse events (AEs). RESULTS: Nine phase 3 clinical trials (7016 advanced oesophageal and gastric cancer patients) met the inclusion criteria. Our meta-analysis demonstrated that the pooled PD-1 inhibitor + chemotherapy group had a significantly longer OS than the chemotherapy-alone group [hazard ratio (HR) = 0.76, 95% confidence interval (CI): 0.71-0.81]; the pooled PFS result was consistent with that of OS (HR = 0.67, 95%CI: 0.61-0.74). The count of patients achieving an objective response in the PD-1 inhibitor + chemotherapy group surpassed that of the chemotherapy-alone group [odds ratio (OR) = 1.86, 95%CI: 1.59-2.18]. AE incidence was also higher in the combination-therapy group than in the chemotherapy-alone group, regardless of whether ≥ grade 3 only (OR = 1.30, 95%CI: 1.07-1.57) or all AE grades (OR = 1.88, 95%CI: 1.39-2.54) were examined. We performed a subgroup analysis based on the programmed death-ligand 1 (PD-L1) combined positive score (CPS) and noted extended OS and PFS durations within the CPS ≥ 1, CPS ≥ 5, and CPS ≥ 10 subgroups of the PD-1 inhibitor + chemotherapy group. CONCLUSION: In contrast to chemotherapy alone, the combination of PD-1 inhibitor and chemotherapy appears to present a more favorable option for initial or subsequent treatment in patients with gastric cancer, GEJ tumor, or oesophageal cancer. This holds true particularly for individuals with PD-L1 CPS scores of ≥ 5 and ≥ 10.

Automated identification of protein expression intensity and classification of protein cellular locations in mouse brain regions from immunofluorescence images.

Knowledge of protein expression in mammalian brains at regional and cellular levels can facilitate understanding of protein functions and associated diseases. As the mouse brain is a typical mammalian brain considering cell type and structure, several studies have been conducted to analyze protein expression in mouse brains. However, labeling protein expression using biotechnology is costly and time-consuming. Therefore, automated models that can accurately recognize protein expression are needed. Here, we constructed machine learning models to automatically annotate the protein expression intensity and cellular location in different mouse brain regions from immunofluorescence images. The brain regions and sub-regions were segmented through learning image features using an autoencoder and then performing K-means clustering and registration to align with the anatomical references. The protein expression intensities for those segmented structures were computed on the basis of the statistics of the image pixels, and patch-based weakly supervised methods and multi-instance learning were used to classify the cellular locations. Results demonstrated that the models achieved high accuracy in the expression intensity estimation, and the F1 score of the cellular location prediction was 74.5%. This work established an automated pipeline for analyzing mouse brain images and provided a foundation for further study of protein expression and functions.

Evaluation of therapeutic benefits of botulinum toxin for foot dystonia associated with Parkinson's disease.

BACKGROUND: Foot dystonia occurs in patients with Parkinson's disease (PD) and leads to pain, malformation, and difficulty with walking. Botulinum toxin injections may be effective for foot dystonia, but the extent of improvement and effects on motor function are unclear. METHODS: In this study, we performed botulinum toxin injections for foot dystonia in 25 patients with PD. At 3 weeks and 3 months post-infection, we assessed changes in plantar pressure distribution utilizing the Pressure Plate system; dystonia using the Modified Ashworth Spasm score; pain using the visual analog scale (VAS) score; and lower extremity function using the Calf-raise Senior (CRS) test, Timed Up and Go (TUG) test, and gait parameters (eg, stride length, step length). RESULTS: We found improved Modified Ashworth Spasm score (p < 0.01) and VAS score (p < 0.01) post-injection. CRS test score (3 weeks, p = 0.006; 3 months, p = 0.068), stride length (3 weeks, p = 0.012; 3 months, p = 0.715), and step length (3 weeks, p = 0.011; 3 months, p = 0.803) also improved. Plantar pressure distribution improved after botulinum toxin injection (metatarsal 1, 3 weeks, p = 0.031; 3 months, p = 0.144; metatarsal 2, 3 weeks, p = 0.049; 3 months, p = 0.065; metatarsal 3, 3 weeks, p = 0.002; 3 months, p = 0.017; metatarsal 4, 3 weeks, p = 0.017; 3 months, p = 0.144; medial heel, 3 weeks, p = 0.01; 3 months, p = 0.395; lateral heel, 3 weeks, p = 0.035; 3 months, p = 0.109). CONCLUSION: Botulinum toxin injection for foot dystonia in patients with PD can reduce spasms and pain and normalize plantar pressure distribution, which improves balance and lower extremity function.

6-Aza-2-Thiothymine-Capped Gold Nanoclusters as Robust Antimicrobial Nanoagents for Eradicating Multidrug-Resistant Escherichia coli Infection.

Multidrug-resistant bacterial infections, especially those caused by multidrug-resistant Escherichia coli (E. coli) bacteria, are an ever-growing threat because of the shrinking arsenal of efficacious antibiotics. Therefore, it is urgently needed to develop a kind of novel, long-term antibacterial agent effectively overcome resistant bacteria. Herein, we present a novel designed antibacterial agent-6-Aza-2-thiothymine-capped gold nanoclusters (ATT-AuNCs), which show excellent antibacterial activity against multidrug-resistant E. coli bacteria. The prepared AuNCs could permeabilize into the bacterial cell membrane via binding with a bivalent cation (e.g., Ca2+), followed by the generation of reactive oxygen species (e.g., •OH and •O2-), ultimately resulting in protein leakage from compromised cell membranes, inducing DNA damage and upregulating pro-oxidative genes intracellular. The AuNCs also speed up the wound healing process without noticeable hemolytic activity or cytotoxicity to erythrocytes and mammalian tissue. Altogether, the results indicate the great promise of ATT-AuNCs for treating multidrug-resistant E. coli bacterial infection.

Development and validation of a nomogram for predicting internal mammary sentinel node metastasis in breast cancer patients.

OBJECTIVE: Internal mammary nodes are important in breast cancer prognosis, but their diagnosis is often missed in clinical practice, leading to inaccurate staging and treatment. We developed a validated nomogram to predict the presence of internal mammary sentinel nodes (IMSN) metastasis. METHODS: A total of 864 sequential IMSN biopsy procedures from a prospective studies database of 1505 cases were used for model development and validation. Multivariable logistic regression was performed on 519 sequential IMSN biopsy procedures from multi-center data between August 2018 and July 2022 to predict the presence of IMSN metastasis. A nomogram was developed based on the logistic regression model and subsequently applied to 345 sequential IMSN biopsy procedures from single-center data between November 2011 and July 2018. The model's discrimination was assessed using the area under the receiver operating characteristic curve. RESULTS: The overall frequency of IMSN metastasis was 17.0% in our study. A predictive model for IMSN metastasis was constructed using tumor size, tumor location, lymphovascular invasion, the number of positive axillary nodes (P < 0.05 for all variables in multivariate analysis), and histological grade (P < 0.05 only in univariate analysis). The nomogram was accurate, with a concordance index of 0.84 in the bootstrapping analysis and an area under the receiver operating characteristic curve of 0.80 in the validation population. CONCLUSION: Our nomogram provides an accurate and validated multivariable predictive model for estimating the individual likelihood of having IMSN metastasis. This may be useful for personalized treatment decisions regarding internal mammary radiotherapy in breast cancer patients.

Effect of CICARE communication nursing model combined with motivational psychological intervention in patients with post-intensive care unit syndrome.

BACKGROUND: Post-intensive care syndrome (PICS) is a term used to describe a constellation of new or worsened dysfunctions in the physical, cognitive, or mental health status of critically ill patients after their discharge from the intensive care unit (ICU). These dysfunctions persist beyond the acute phase of illness and have a significant impact on both the patient and their family. Connect, Introduce, Communicate, Ask, Respond, Exit (CICARE) communication advocates that patients should be respected and accepted when receiving medical services. Clinicians should attach importance to the communication mode of feelings, including connection, introduction, communication, ask, response, and exit 6 steps. AIM: To assess the impact of CICARE communication on the reduction of anxiety, depression, and post-traumatic stress disorder (PTSD) symptoms in patients transitioning from the ICU to other care settings. METHODS: This prospective, randomized, controlled study was performed between October 2021 and March 2023. Intensive Care Unit Memory Tool was used to evaluate patients' ICU memory. The Hospital Anxiety and Depression Scale was employed to determine the presence of anxiety or depression symptoms. Impact of Event Scale-Revised was utilized to assess the presence of PTSD. All data were processed and analyzed using R language software version 4.1.0. The measurement data were expressed as mean ± SD, and the t test was used. The count data were analyzed by the χ2 test and expressed as [n (%)]. RESULTS: In total, 248 subjects were included in this study. Among them, 206 were successfully followed up for three months after transfer from the ICU, and 42 cases were lost to follow-up. There was no significant difference in the composition of ICU memory between the two groups. The application of the CICARE communication nursing model combined with the motivational psychological intervention nursing model, as well as the adoption of only the motivational psychological intervention nursing model, demonstrated favorable effects on PICS. Both groups of patients showed a reduction in anxiety scores, depression scores, and PTSD scores following the implementation of these two nursing models. However, it is noteworthy that the experimental group exhibited greater improvements compared to the control group. CONCLUSION: Our findings suggest that CICARE communication nursing mode may have good influence on relieving PICS.

BaLa5V2O3N7: a novel anti-perovskite oxynitride for electrode applications.

The exploration of transition metal oxynitrides has garnered significant interest due to their intriguing property diversity. Herein, we present a promising new transition metal oxynitride BaLa5V2O3N7, which features an anti-perovskite structure type. This unique structural configuration endows the material with remarkable conductivity, particularly at low temperatures, paving the way for the material to be used in a wide range of technological applications.

Amelioratory effects of astragaloside IV on hepatocarcinogenesis via Nrf2-mediated pSmad3C/3L transformation.

BACKGROUND: Phosphorylated Smad3 isoforms are reversible and antagonistic, and the tumour-suppressive pSmad3C can shift to an oncogenic pSmad3L signal. In addition, Nrf2 has a two-way regulatory effect on tumours, protecting normal cells from carcinogens and promoting tumour cell survival in chemotherapeutics. Accordingly, we hypothesised that the transformation of pSmad3C/3L is the basis for Nrf2 to produce both pro- and/or anti-tumourigenic effects in hepatocarcinogenesis. Astragaloside IV (AS-IV), the major component of Astragalus membranaceus, exerts anti-fibrogenic and carcinogenic actions. Lately, AS-IV administration could delay the occurrence of primary liver cancer by persistently inhibiting the fibrogenesis and regulating pSmad3C/3 L and Nrf2/HO-1 pathways synchronously. However, effect of AS-IV on hepatocarcinogenesis implicated in the bidirectional cross-talking of pSmad3C/3 L and Nrf2/HO-1 signalling, especially which one contributes palpably than the other still remains unclear. PURPOSE: This study aims to settle the above questions by using in vivo (pSmad3C+/- and Nrf2-/- mice) and in vitro (plasmid- or lentivirus- transfected HepG2 cells) models of HCC. STUDY DESIGN AND METHODS: The correlation of Nrf2 to pSmad3C/pSmad3L in HepG2 cells was analysed by Co-immunoprecipitation and dual-luciferase reporter assay. Pathological changes of Nrf2, pSmad3C, and pSmad3L in human HCC patients, pSmad3C+/- mice, and Nrf2-/- mice were gauged by immunohistochemical, haematoxylin and eosin staining, Masson, and immunofluorescence assays. Finally, western blot and qPCR were used to verify the bidirectional cross-talking of pSmad3C/3L and Nrf2/HO-1 signalling protein and mRNA in vivo and in vitro models of HCC. RESULTS: Histopathological manifestations and biochemical indicators revealed that pSmad3C+/- could abate the ameliorative effects of AS-IV on fibrogenic/carcinogenic mice with Nrf2/HO-1 deactivation and pSmad3C/p21 transform to pSmad3L/PAI-1//c-Myc. As expected, cell experiments confirmed that upregulating pSmad3C boosts the inhibitory activity of AS-IV on phenotypes (cell proliferation, migration and invasion), followed by a shift of pSmad3L to pSmad3C and activation of Nrf2/HO-1. Synchronously, experiments in Nrf2-/- mice and lentivirus-carried Nrf2shRNA cell echoed the results of pSmad3C knockdown. Complementarily, Nrf2 overexpression resulted in the opposite result. Furthermore, Nrf2/HO-1 contributes to AS-IV's anti-HCC effect palpably compared with pSmad3C/3L. CONCLUSION: These studies highlight that harnessing the bidirectional crosstalk pSmad3C/3 L and Nrf2/HO-1, especially Nrf2/HO-1 signalling, acts more effectively in AS-IV's anti-hepatocarcinogenesis, which may provide an important theoretical foundation for the use of AS-IV against HCC.

Primary malignant melanoma of the esophagus: A case report.

BACKGROUND: Primary malignant melanoma of the esophagus (PMME) is a rare malignant disease whose clinical and molecular pathological features, origin and pathogenesis, diagnosis and treatment have not been elucidated. CASE SUMMARY: In this paper, we report a case of a 73-year-old male with PMME. The patient complained of progressive dysphagia accompanied by substantial weight loss. Gastroscopy revealed a purple black bulging-type mass in the lower esophagus with easy bleeding on contact and scattered satellite lesions in the stomach. Histopathological biopsy revealed melanocytes in the esophageal mucosa. Physical examination and multidisciplinary consultation led to diagnostic exclusion of melanoma originating in other organs, such as the skin. Through this case report and literature review, we aimed to describe the clinical and molecular pathological features of PMME and summarize possible pathways of pathogenesis as well as cutting-edge therapeutic advances. CONCLUSION: PMME is a rare malignancy of the esophagus with a poor prognosis. Clinicians should raise their awareness and be able to identify early lesions.

Combined Single-Cell and Spatial Transcriptomics Reveal the Metabolic Evolvement of Breast Cancer during Early Dissemination.

Breast cancer is now the most frequently diagnosed malignancy, and metastasis remains the leading cause of death in breast cancer. However, little is known about the dynamic changes during the evolvement of dissemination. In this study, 65 968 cells from four patients with breast cancer and paired metastatic axillary lymph nodes are profiled using single-cell RNA sequencing (scRNA-seq) and spatial transcriptomics. A disseminated cancer cell cluster with high levels of oxidative phosphorylation (OXPHOS), including the upregulation of cytochrome C oxidase subunit 6C and dehydrogenase/reductase 2, is identified. The transition between glycolysis and OXPHOS when dissemination initiates is noticed. Furthermore, this distinct cell cluster is distributed along the tumor's leading edge. The findings here are verified in three different cohorts of breast cancer patients and an external scRNA-seq dataset, which includes eight patients with breast cancer and paired metastatic axillary lymph nodes. This work describes the dynamic metabolic evolvement of early disseminated breast cancer and reveals a switch between glycolysis and OXPHOS in breast cancer cells as the early event during lymph node metastasis.

Modulating tumour vascular normalisation using triptolide-loaded NGR-functionalized liposomes for enhanced cancer radiotherapy.

Radiotherapy is an effective therapy in tumour treatment. However, the characteristics of the tumour microenvironment, including hypoxia, low pH, and interstitial fluid pressure bring about radioresistance. To improve the anti-tumour effect of radiotherapy, it has been demonstrated that antiangiogenic therapy can be employed to repair the structural and functional defects of tumour angiogenic vessels, thereby preventing radioresistance or poor therapeutic drug delivery. In this study, we prepared triptolide (TP)-loaded Asn-Gly-Arg (NGR) peptide conjugated mPEG2000-DSPE-targeted liposomes (NGR-PEG-TP-LPs) to induce tumour blood vessel normalisation, to the end of increasing the sensitivity of tumour cells to radiotherapy. Further, to quantify the tumour vessel normalisation window, the structure and functionality of tumour blood vessels post NGR-PEG-TP-LPs treatment were evaluated. Thereafter, the anti-tumour effect of radiotherapy following these treatments was evaluated using HCT116 xenograft-bearing mouse models based on the tumour vessel normalisation period window. The results obtained showed that NGR-PEG-TP-LPs could modulate tumour vascular normalisation to increase the oxygen content of the tumour microenvironment and enhance the efficacy of radiotherapy. Further, liver and kidney toxicity tests indicated that NGR-PEG-TP-LPs are safe for application in cancer treatment.

Resistance to antibody-drug conjugates in breast cancer: mechanisms and solutions.

Antibody-drug conjugates (ADCs) are a rapidly developing therapeutic approach in cancer treatment that has shown remarkable activity in breast cancer. Currently, there are two ADCs approved for the treatment of human epidermal growth factor receptor 2-positive breast cancer, one for triple-negative breast cancer, and multiple investigational ADCs in clinical trials. However, drug resistance has been noticed in clinical use, especially in trastuzumab emtansine. Here, the mechanisms of ADC resistance are summarized into four categories: antibody-mediated resistance, impaired drug trafficking, disrupted lysosomal function, and payload-related resistance. To overcome or prevent resistance to ADCs, innovative development strategies and combination therapy options are being investigated. Analyzing predictive biomarkers for optimal therapy selection may also help to prevent drug resistance.

Automatic recognition of protein subcellular location patterns in single cells from immunofluorescence images based on deep learning.

With the improvement of single-cell measurement techniques, there is a growing awareness that individual differences exist among cells, and protein expression distribution can vary across cells in the same tissue or cell line. Pinpointing the protein subcellular locations in single cells is crucial for mapping functional specificity of proteins and studying related diseases. Currently, research about single-cell protein location is still in its infancy, and most studies and databases do not annotate proteins at the cell level. For example, in the human protein atlas database, an immunofluorescence image stained for a particular protein shows multiple cells, but the subcellular location annotation is for the whole image, ignoring intercellular difference. In this study, we used large-scale immunofluorescence images and image-level subcellular locations to develop a deep-learning-based pipeline that could accurately recognize protein localizations in single cells. The pipeline consisted of two deep learning models, i.e. an image-based model and a cell-based model. The former used a multi-instance learning framework to comprehensively model protein distribution in multiple cells in each image, and could give both image-level and cell-level predictions. The latter firstly used clustering and heuristics algorithms to assign pseudo-labels of subcellular locations to the segmented cell images, and then used the pseudo-labels to train a classification model. Finally, the image-based model was fused with the cell-based model at the decision level to obtain the final ensemble model for single-cell prediction. Our experimental results showed that the ensemble model could achieve higher accuracy and robustness on independent test sets than state-of-the-art methods.

Automated classification of protein expression levels in immunohistochemistry images to improve the detection of cancer biomarkers.

BACKGROUND: The expression changes of some proteins are associated with cancer progression, and can be used as biomarkers in cancer diagnosis. Automated systems have been frequently applied in the large-scale detection of protein biomarkers and have provided a valuable complement for wet-laboratory experiments. For example, our previous work used an immunohistochemical image-based machine learning classifier of protein subcellular locations to screen biomarker proteins that change locations in colon cancer tissues. The tool could recognize the location of biomarkers but did not consider the effect of protein expression level changes on the screening process. RESULTS: In this study, we built an automated classification model that recognizes protein expression levels in immunohistochemical images, and used the protein expression levels in combination with subcellular locations to screen cancer biomarkers. To minimize the effect of non-informative sections on the immunohistochemical images, we employed the representative image patches as input and applied a Wasserstein distance method to determine the number of patches. For the patches and the whole images, we compared the ability of color features, characteristic curve features, and deep convolutional neural network features to distinguish different levels of protein expression and employed deep learning and conventional classification models. Experimental results showed that the best classifier can achieve an accuracy of 73.72% and an F1-score of 0.6343. In the screening of protein biomarkers, the detection accuracy improved from 63.64 to 95.45% upon the incorporation of the protein expression changes. CONCLUSIONS: Machine learning can distinguish different protein expression levels and speed up their annotation in the future. Combining information on the expression patterns and subcellular locations of protein can improve the accuracy of automatic cancer biomarker screening. This work could be useful in discovering new cancer biomarkers for clinical diagnosis and research.

GraphLoc: a graph neural network model for predicting protein subcellular localization from immunohistochemistry images.

MOTIVATION: Recognition of protein subcellular distribution patterns and identification of location biomarker proteins in cancer tissues are important for understanding protein functions and related diseases. Immunohistochemical (IHC) images enable visualizing the distribution of proteins at the tissue level, providing an important resource for the protein localization studies. In the past decades, several image-based protein subcellular location prediction methods have been developed, but the prediction accuracies still have much space to improve due to the complexity of protein patterns resulting from multi-label proteins and the variation of location patterns across cell types or states. RESULTS: Here, we propose a multi-label multi-instance model based on deep graph convolutional neural networks, GraphLoc, to recognize protein subcellular location patterns. GraphLoc builds a graph of multiple IHC images for one protein, learns protein-level representations by graph convolutions and predicts multi-label information by a dynamic threshold method. Our results show that GraphLoc is a promising model for image-based protein subcellular location prediction with model interpretability. Furthermore, we apply GraphLoc to the identification of candidate location biomarkers and potential members for protein networks. A large portion of the predicted results have supporting evidence from the existing literatures and the new candidates also provide guidance for further experimental screening. AVAILABILITY AND IMPLEMENTATION: The dataset and code are available at: www.csbio.sjtu.edu.cn/bioinf/GraphLoc. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Individualized risk estimation for postoperative pulmonary complications after hepatectomy based on perioperative variables.

BACKGROUND: At present, there is no perfect system to evaluate pulmonary complications of liver surgery using perioperative variables. AIM: To design and verify a risk assessment system for predicting postoperative pulmonary complications (PPCs) after hepatectomy based on perioperative variables. METHODS: A retrospective analysis was performed on 1633 patients who underwent liver surgery. The variables were screened using univariate and multivariate analyses, and graded scores were assigned to the selected variables. Logistic regression was used to develop the liver operation pulmonary complication scoring system (LOPCSS) for the prediction of PPCs. The LOPCSS was verified using the receiver operating characteristic curve. RESULTS: According to the multivariate correlation analysis, the independent factors which influenced PPCs of liver surgery were age [≥ 65 years old/< 65 years old, odds ratio (OR) = 1.926, P = 0.011], medical diseases requiring drug treatment (yes/no, OR = 3.523, P < 0.001), number of liver segments to be removed (≥ 3/≤ 2, OR = 1.683, P = 0.002), operation duration (≥ 180 min/< 180 min, OR = 1.896, P = 0.004), and blood transfusion (yes/no, OR = 1.836, P = 0.003). The area under the curve (AUC) of the LOPCSS was 0.742. The cut-off value of the expected score for complications was 5. The incidence of complications in the group with ≤ 4 points was significantly lower than that in the group with ≥ 6 points (2.95% vs 33.40%, P < 0.001). Furthermore, in the validation dataset, the corresponding AUC of LOPCSS was 0.767. CONCLUSION: As a novel and simplified assessment system, the LOPCSS can effectively predict PPCs of liver surgery through perioperative variables.