Subtype Classification Research Articles

Advances in the application of computational pathology in diagnosis, immunomicroenvironment recognition, and immunotherapy evaluation of breast cancer: a narrative review.

Breast cancer (BC) is a prevalent and highly lethal malignancy affecting women worldwide. Immunotherapy has emerged as a promising therapeutic strategy for BC, offering potential improvements in patient survival. Neoadjuvant therapy (NAT) has also gained significant clinical traction. With the advancement of computer technology, Artificial Intelligence (AI) has been increasingly applied in pathology research, expanding and redefining the scope of the field. This narrative review aims to provide a comprehensive overview of the current literature on the application of computational pathology in BC, specifically focusing on diagnosis, immune microenvironment recognition, and the evaluation of immunotherapy and NAT response. A thorough examination of relevant literature was conducted, focusing on studies investigating the role of computational pathology in BC diagnosis, immune microenvironment recognition, and immunotherapy and NAT assessment. The application of computational pathology has shown significant potential in BC management. AI-based techniques enable improved diagnosis and classification of BC subtypes, enhance the identification and characterization of the immune microenvironment, and facilitate the evaluation of immunotherapy and NAT response. However, challenges related to data quality, standardization, and algorithm development still need to be addressed. The integration of computational pathology and AI has transformative implications for BC patient care. By leveraging AI-based technologies, clinicians can make more informed decisions in diagnosis, treatment planning, and therapeutic response assessment. Future research should focus on refining AI algorithms, addressing technical challenges, and conducting large-scale clinical validation studies to facilitate the translation of computational pathology into routine clinical practice for BC patients.

Development of Symbolic Expressions Ensemble for Breast Cancer Type Classification Using Genetic Programming Symbolic Classifier and Decision Tree Classifier

Breast cancer is a type of cancer with several sub-types. It occurs when cells in breast tissue grow out of control. The accurate sub-type classification of a patient diagnosed with breast cancer is mandatory for the application of proper treatment. Breast cancer classification based on gene expression is challenging even for artificial intelligence (AI) due to the large number of gene expressions. The idea in this paper is to utilize the genetic programming symbolic classifier (GPSC) on the publicly available dataset to obtain a set of symbolic expressions (SEs) that can classify the breast cancer sub-type using gene expressions with high classification accuracy. The initial problem with the used dataset is a large number of input variables (54,676 gene expressions), a small number of dataset samples (151 samples), and six classes of breast cancer sub-types that are highly imbalanced. The large number of input variables is solved with principal component analysis (PCA), while the small number of samples and the large imbalance between class samples are solved with the application of different oversampling methods generating different dataset variations. On each oversampled dataset, the GPSC with random hyperparameter values search (RHVS) method is trained using 5-fold cross validation (5CV) to obtain a set of SEs. The best set of SEs is chosen based on mean values of accuracy (ACC), the area under the receiving operating characteristic curve (AUC), precision, recall, and F1-score values. In this case, the highest classification accuracy is equal to 0.992 across all evaluation metric methods. The best set of SEs is additionally combined with a decision tree classifier, which slightly improves ACC to 0.994.

The proteomic landscape of soft tissue sarcomas

Soft tissue sarcomas (STS) are rare and diverse mesenchymal cancers with limited treatment options. Here we undertake comprehensive proteomic profiling of tumour specimens from 321 STS patients representing 11 histological subtypes. Within leiomyosarcomas, we identify three proteomic subtypes with distinct myogenesis and immune features, anatomical site distribution and survival outcomes. Characterisation of undifferentiated pleomorphic sarcomas and dedifferentiated liposarcomas with low infiltrating CD3 + T-lymphocyte levels nominates the complement cascade as a candidate immunotherapeutic target. Comparative analysis of proteomic and transcriptomic profiles highlights the proteomic-specific features for optimal risk stratification in angiosarcomas. Finally, we define functional signatures termed Sarcoma Proteomic Modules which transcend histological subtype classification and show that a vesicle transport protein signature is an independent prognostic factor for distant metastasis. Our study highlights the utility of proteomics for identifying molecular subgroups with implications for risk stratification and therapy selection and provides a rich resource for future sarcoma research.

Monocytes in the Characterization of Pain in Palliative Patients with Severe Dementia-A Pilot Study.

In assessing and managing pain, when obtaining a self-report is impossible, therapeutic decision-making becomes more challenging. This study aimed to investigate whether monocytes and some membrane monocyte proteins, identified as a cluster of differentiation (CD), could be potential non-invasive peripheral biomarkers in identifying and characterizing pain in patients with severe dementia. We used 53 blood samples from non-oncological palliative patients, 44 patients with pain (38 of whom had dementia) and 0 without pain or dementia (controls). We evaluated the levels of monocytes and their subtypes, including classic, intermediate, and non-classic, and characterized the levels of specific phenotypic markers, namely CD11c, CD86, CD163, and CD206. We found that the relative concentrations of monocytes, particularly the percentage of classic monocytes, may be a helpful pain biomarker. Furthermore, the CD11c expression levels were significantly higher in patients with mixed pain, while CD163 and CD206 expression levels were significantly higher in patients with nociceptive pain. These findings suggest that the levels of monocytes, particularly the classic subtype, and their phenotype markers CD11c, CD163, and CD206 could serve as pain biomarkers in patients with severe dementia.

Explainable CAD System for Classification of Acute Lymphoblastic Leukemia Based on a Robust White Blood Cell Segmentation

Leukemia is a significant health challenge, with high incidence and mortality rates. Computer-aided diagnosis (CAD) has emerged as a promising approach. However, deep-learning methods suffer from the “black box problem”, leading to unreliable diagnoses. This research proposes an Explainable AI (XAI) Leukemia classification method that addresses this issue by incorporating a robust White Blood Cell (WBC) nuclei segmentation as a hard attention mechanism. The segmentation of WBC is achieved by combining image processing and U-Net techniques, resulting in improved overall performance. The segmented images are fed into modified ResNet-50 models, where the MLP classifier, activation functions, and training scheme have been tested for leukemia subtype classification. Additionally, we add visual explainability and feature space analysis techniques to offer an interpretable classification. Our segmentation algorithm achieves an Intersection over Union (IoU) of 0.91, in six databases. Furthermore, the deep-learning classifier achieves an accuracy of 99.9% on testing. The Grad CAM methods and clustering space analysis confirm improved network focus when classifying segmented images compared to non-segmented images. Overall, the proposed visual explainable CAD system has the potential to assist physicians in diagnosing leukemia and improving patient outcomes.

Identification of TIMP1 as an inflammatory biomarker associated with temporal lobe epilepsy based on integrated bioinformatics and experimental analyses

BackgroundEpilepsy is the second most prevalent neurological disease. Although there are many antiseizure drugs, approximately 30% of cases are refractory to treatment. Temporal lobe epilepsy (TLE) is the most common epilepsy subtype, and previous studies have reported that hippocampal inflammation is an important mechanism associated with the occurrence and development of TLE. However, the inflammatory biomarkers associated with TLE are not well defined.MethodsIn our study, we merged human hippocampus datasets (GSE48350 and GSE63808) through batch correction and generally verified the diagnostic roles of inflammation-related genes (IRGs) and subtype classification according to IRGs in epilepsy through differential expression, random forest, support vector machine, nomogram, subtype classification, enrichment, protein‒protein interaction, immune cell infiltration, and immune function analyses. Finally, we detected the location and expression of inhibitor of metalloproteinase-1 (TIMP1) in epileptic patients and kainic acid-induced epileptic mice.ResultsAccording to the bioinformatics analysis, we identified TIMP1 as the most significant IRG associated with TLE, and we found that TIMP1 was mainly located in cortical neurons and scantly expressed in cortical gliocytes by immunofluorescence staining. We detected decreased expression of TIMP1 by quantitative real-time polymerase chain reaction and western blotting.ConclusionTIMP1, the most significant IRG associated with TLE, might be a novel and promising biomarker to study the mechanism of epilepsy and guide the discovery of new drugs for its treatment.

Analyses of m6A regulatory genes and subtype classification in atrial fibrillation.

To explore the role of m6A regulatory genes in atrial fibrillation (AF), we classified atrial fibrillation patients into subtypes by two genotyping methods associated with m6A regulatory genes and explored their clinical significance. We downloaded datasets from the Gene Expression Omnibus (GEO) database. The m6A regulatory gene expression levels were extracted. We constructed and compared random forest (RF) and support vector machine (SVM) models. Feature genes were selected to develop a nomogram model with the superior model. We identified m6A subtypes based on significantly differentially expressed m6A regulatory genes and identified m6A gene subtypes based on m6A-related differentially expressed genes (DEGs). Comprehensive evaluation of the two m6A modification patterns was performed. The data of 107 samples from three datasets, GSE115574, GSE14975 and GSE41177, were acquired from the GEO database for training models, comprising 65 AF samples and 42 sinus rhythm (SR) samples. The data of 26 samples from dataset GSE79768 comprising 14 AF samples and 12 SR samples were acquired from the GEO database for external validation. The expression levels of 23 regulatory genes of m6A were extracted. There were correlations among the m6A readers, erasers, and writers. Five feature m6A regulatory genes, ZC3H13, YTHDF1, HNRNPA2B1, IGFBP2, and IGFBP3, were determined (p < 0.05) to establish a nomogram model that can predict the incidence of atrial fibrillation with the RF model. We identified two m6A subtypes based on the five significant m6A regulatory genes (p < 0.05). Cluster B had a lower immune infiltration of immature dendritic cells than cluster A (p < 0.05). On the basis of six m6A-related DEGs between m6A subtypes (p < 0.05), two m6A gene subtypes were identified. Both cluster A and gene cluster A scored higher than the other clusters in terms of m6A score computed by principal component analysis (PCA) algorithms (p < 0.05). The m6A subtypes and m6A gene subtypes were highly consistent. The m6A regulatory genes play non-negligible roles in atrial fibrillation. A nomogram model developed by five feature m6A regulatory genes could be used to predict the incidence of atrial fibrillation. Two m6A modification patterns were identified and evaluated comprehensively, which may provide insights into the classification of atrial fibrillation patients and guide treatment.

DP19 Two patients with Kaposi sarcoma from the 2-week-wait clinic

Abstract We present two cases of Kaposi sarcoma (KS) in HIV-negative men who have sex with men (MSM) with no coexisting immunosuppressive treatment or conditions. They both presented with suspicious lesions on lower extremities, with no other mucocutaneous involvement. The initial clinical presentation of both patients was atypical, with uncharacteristic aspects of lesions on clinical examination and dermoscopy, which prompted further investigations. Skin biopsies were performed for both patients. Both cases had similar histological findings in keeping with KS: slit-like vascular spaces within the dermis, proliferation of capillary blood vessels with associated inflammatory infiltrate with lymphocytes and plasma cells and focal aggregates of spindle cells. Immunohistochemistry revealed endothelial cells displaying positive staining for human herpesvirus (HHV)-8. Immunodeficiency screening revealed no abnormalities, and repeat HIV screening was negative in both patients. These cases present an interesting occurrence of an increasingly encountered new subtype of KS seen in HIV-seronegative immunocompetent MSM. In addition to the four established variants of KS, this is an emerging variant that arises mostly in middle-aged MSM. HHV-8 has been identified as the aetiological agent in KS; indeed, the viral infection precedes the development of the disease. The fifth subtype is increasingly recognized: nonepidemic KS in MSM who are HIV-negative and have no other causes for immunodeficiency, as seen in our two cases. This subtype has been reported in the literature since the 1980s in this specific population. The clinical presentation resembles classic KS, with predominantly cutaneous involvement of the lower limbs, running a similar indolent course. There is a strong connection between HHV-8 infection and the development of KS; however, in countries where there is a low prevalence of the virus, in the general population, there is increased detection in the MSM group. Correlating the high detection rate of HHV-8 in MSM with its role in the development in KS, immunocompetent MSM are at risk of developing the condition. However, this subtype carries an overall good prognosis, with an indolent course and good response to similar therapeutic options used in the classic subtype.

Migraine with aura detection and subtype classification using machine learning algorithms and morphometric magnetic resonance imaging data.

Migraine with aura (MwA) is a neurological condition manifested in moderate to severe headaches associated with transient visual and somatosensory symptoms, as well as higher cortical dysfunctions. Considering that about 5% of the world's population suffers from this condition and manifestation could be abundant and characterized by various symptoms, it is of great importance to focus on finding new and advanced techniques for the detection of different phenotypes, which in turn, can allow better diagnosis, classification, and biomarker validation, resulting in tailored treatments of MwA patients. This research aimed to test different machine learning techniques to distinguish healthy people from those suffering from MwA, as well as people with simple MwA and those experiencing complex MwA. Magnetic resonance imaging (MRI) post-processed data (cortical thickness, cortical surface area, cortical volume, cortical mean Gaussian curvature, and cortical folding index) was collected from 78 subjects [46 MwA patients (22 simple MwA and 24 complex MwA) and 32 healthy controls] with 340 different features used for the algorithm training. The results show that an algorithm based on post-processed MRI data yields a high classification accuracy (97%) of MwA patients and precise distinction between simple MwA and complex MwA with an accuracy of 98%. Additionally, the sets of features relevant to the classification were identified. The feature importance ranking indicates the thickness of the left temporal pole, right lingual gyrus, and left pars opercularis as the most prominent markers for MwA classification, while the thickness of left pericalcarine gyrus and left pars opercularis are proposed as the two most important features for the simple and complex MwA classification. This method shows significant potential in the validation of MwA diagnosis and subtype classification, which can tackle and challenge the current treatments of MwA.

Avascular necrosis of humeral head after proximal humerus fracture: comparison between classification systems in predicting necrosis risk.

Avascular necrosis (AVN) of the humeral head is a relatively frequent complication after proximal humerus fractures (PHF). There are many factors related to the risk of developing AVN, with fracture complexity being one of the most relevant. Aim of the present study is to evaluate the correlation between different classification systems and the risk of post-operative AVN in patients treated with locking plate fixation for Neer 3 and 4-part PHFs. The study population included 44 patients (F:M 2:1, mean age 62) treated between December 2014 and April 2019. Fractures were classified according to Neer, Russo and Lego classifications and Hertel's criteria. AVN was established on postoperative radiographs after a minimum follow-up of 6 months. The odds ratio (OR) for AVN for each classification subtype was calculated. The incidence of necrosis was 11%. Significant association with AVN was found for Neer 4 part (OR=8), Russo IVB (OR=4), Lego pattern 12 (OR=8) and in presence of 3 mayor Hertel's criteria (OR=15,5). There is a significant association between AVN and Neer 4 part, Russo type IVB and Lego pattern 12 fractures. AVN risk is also significantly higher in presence of 3 major Hertel's criteria.

Specific inhibitory effects of guanosine on breast cancer cell proliferation

Breast cancer is the most commonly diagnosed cancer and the leading cause of cancer-related death. Drug therapy for breast cancer is currently selected based on the subtype classification; however, many anticancer drugs are highly cytotoxic. Since intracellular levels of GTP are elevated in many cancer cells that undergo a specific cell proliferation cycle, GTP has potential as a target for cancer therapy. The present study focused on nucleosides and nucleotides and examined intracellular GTP-dependent changes in cell proliferation rates in normal (MCF-12A) and cancer (MCF-7) breast cell lines. Decreased cell proliferation due to a reduction in intracellular GTP levels by mycophenolic acid (MPA), an inosine monophosphate dehydrogenase inhibitor, was observed in both cell lines. The inhibitory effects of MPA on cell proliferation were suppressed when it was applied in combination with Guanosine (Guo), a substrate for GTP salvage synthesis, while the single exposure to Guo suppressed the proliferation of MCF-7 cells only. Although the underlying mechanisms remain unclear, since the inhibitory effects of Guo on cell proliferation did not correlate with GTP or ATP intracellular levels or the GTP/ATP ratio, there may be another cause besides GTP metabolism. Guo inhibited the proliferation of MCF-7, a human breast cancer cell line, but not MCF-12A, a human normal breast cell line. Further studies are needed to investigate the potential of applying Guo as a target for the development of a novel cancer treatment system.

Inflammatory Bowel Disease in Children.

Inflammatory bowel disease is an important cause of chronic diarrhea in children, with a rising incidence, globally. The two main subtypes include Crohn's disease and ulcerative colitis. The clinical features are variable, and diagnosis requires initial first-line investigations followed by the involvement of specialist input for targeted imaging and endoscopy with biopsy, to confirm the diagnosis. Despite detailed investigation, inflammatory bowel disease may be clinically indistinguishable from chronic infections such as intestinal tuberculosis, and anti-tuberculosis treatment may be considered prior to further management considerations. The medical management of inflammatory bowel disease depends on subtype classification and severity, and may involve a step-wise approach to immunosuppressive therapies. In children, the consequences of poorly controlled disease are wide ranging, from psychosocial impacts and school non-attendance, to growth impairment and pubertal delay with subsequent impacts on bone health. In addition, an increased need for hospitalization and surgical intervention, and ultimately risk of cancer long-term. A multi-disciplinary team with expertise in inflammatory bowel disease is recommended to mitigate these risks and help to achieve the goal of sustained remission with endoscopic healing. This review focuses on updates on best clinical practice on the diagnosis and management of inflammatory bowel disease in children.

MorphAttnNet: An Attention-based morphology framework for lung cancer subtype classification

Lung cancer is recognized as the most life-threatening cancer among other type of cancers all over the world. Early stage recognition and proper diagnosis can increase the five-year survival rate and also save the patient life. Accurate identification of lung cancer subtype from histopathological images plays an important role and help doctors to take necessary decisions for lung cancer treatment. Therefore, in this work, a new deep learning (DL) framework based on image morphology is developed for lung cancer subtype classification. The proposed Morphology-based Attention Network, (MorphAttnNet) can classify lung benign, adenocarcinoma (ADC), and squamous cell carcinoma (SCC) from histopathology images. The framework is designed based on convolution and morphological operations. Attention-based mechanism is incorporated to select important features from histopathology images. The framework with its morphology-based path combined with attention blocks is able to capture morphological variations of lung cancer subtypes accurately and effectively. Finally, the extracted deep features from convolution and morphological paths are combined and used for lung cancer subtype classification. The performance of the proposed framework is analyzed on publicly available LC25000 dataset and achieved a sensitivity, specificity, average accuracy, precision, and f1-score of 98.33%, 97.76%, 98.96%, 99.12%, and 98.72% respectively for lung cancer subtype classification. The proposed system is also compared with existing state-of-the-art systems and achieved considerable performance indices for lung cancer subtype classification.

MMDAE-HGSOC: A novel method for high-grade serous ovarian cancer molecular subtypes classification based on multi-modal deep autoencoder

High-grade serous ovarian cancer (HGSOC) is a type of ovarian cancer developed from serous tubal intraepithelial carcinoma. The intrinsic differences among molecular subtypes are closely associated with prognosis and pathological characteristics. At present, multi-omics data integration methods include early integration and late integration. Most existing HGSOC molecular subtypes classification methods are based on the early integration of multi-omics data. The mutual interference among multi-omics data is ignored, which affects the effectiveness of feature learning. High-dimensional multi-omics data contains genes unassociated with HGSOC molecular subtypes, resulting in redundant information, which is not conducive to model training. In this paper, we propose a multi-modal deep autoencoder learning method, MMDAE-HGSOC. MiRNA expression, DNA methylation, and copy number variation (CNV) are integrated with mRNA expression data to construct a multi-omics feature space. The multi-modal deep autoencoder network is used to learn the high-level feature representation of multi-omics data. The superposition LASSO (S-LASSO) regression algorithm is proposed to fully obtain the associated genes of HGSOC molecular subtypes. The experimental results show that MMDAE-HGSOC is superior to the existing classification methods. Finally, we analyze the enrichment gene ontology (GO) terms and biological pathways of these significant genes, which are discovered during the gene selection process.

Significance of Oxidative Stress in the Diagnosis and Subtype Classification of Intervertebral Disc Degeneration.

Intervertebral disc degeneration (IVDD) is a common illness of aging, and its pathophysiological process is mainly manifested by cell aging and apoptosis, an imbalance in the production and catabolism of extracellular matrix, and an inflammatory response. Oxidative stress (OS) is an imbalance that decreases the body's intrinsic antioxidant defense system and/or raises the formation of reactive oxygen species and performs multiple biological functions in the body. However, our current knowledge of the effect of OS on the progression and treatment of IVDD is still extremely limited. In this study, we obtained 35 DEGs by differential expression analysis of 437 OS-related genes (OSRGs) between IVDD patients and healthy individuals from GSE124272 and GSE150408. Then, we identified six hub OSRGs (ATP7A, MELK, NCF1, NOX1, RHOB, and SP1) from 35 DEGs, and the high accuracy of these hub genes was confirmed by constructing ROC curves. In addition, to forecast the risk of IVDD patients, we developed a nomogram. We obtained two OSRG clusters (clusters A and B) by consensus clustering based on the six hub genes. Then, 3147 DEGs were obtained by differential expression analysis in the two clusters, and all samples were further divided into two gene clusters (A and B). We investigated differences in immune cell infiltration levels between different clusters and found that most immune cells had higher infiltration levels in OSRG cluster B or gene cluster B. In conclusion, OS is important in the formation and progression of IVDD, and we believe that our work will help guide future research on OS in IVDD.

ACSL4: biomarker, mediator and target in quadruple negative breast cancer

Breast cancer is a heterogeneous disease for which effective treatment depends on correct categorization of its molecular subtype. For the last several decades this determination has relied on hormone receptor status for estrogen, progesterone and HER2. More recently, gene expression data have been generated that further stratify both receptor-positive and receptor-negative cancers. The fatty acid-activating enzyme, ACSL4, has been demonstrated to play a role in the malignant phenotype of a variety of cancers, including breast. This lipid metabolic enzyme is differentially expressed as a function of subtype in breast tumors, with highest expression observed in the mesenchymal (claudin low) and basal-like subtypes. Here we review data that support the potential of utilizing ACSL4 status as both a biomarker of molecular subtype and a predictor of response to a variety of targeted and non-targeted treatment regimens. Based on these findings, we suggest 3 expanded roles for ACSL4: 1. as a biomarker for classification of breast cancer subtypes; 2. as a predictor of sensitivity to hormone-based and certain other therapies; and 3. as a target for the development of new treatment modalities.

Prognostic value of central blood pressure on the outcomes of embolic stroke of undetermined source

We investigated the prognostic impact of central blood pressure (BP) on outcomes in patients with embolic stroke of undetermined source (ESUS). The prognostic value of central BP according to ESUS subtype was also evaluated. We recruited patients with ESUS and data on their central BP parameters (central systolic BP [SBP], central diastolic BP [DBP], central pulse pressure [PP], augmentation pressure [AP], and augmentation index [AIx]) during admission. ESUS subtype classification was arteriogenic embolism, minor cardioembolism, two or more causes, and no cause. Major adverse cardiovascular event (MACE) was defined as recurrent stroke, acute coronary syndrome, hospitalization for heart failure, or death. Over a median of 45.8 months, 746 patients with ESUS were enrolled and followed up. Patients had a mean age of 62.8 years, and 62.2% were male. Multivariable Cox regression analysis showed that central SBP and PP were associated with MACE. All-cause mortality was independently associated with AIx. In patients with no cause ESUS, central SBP and PP, AP, and AIx were independently associated with MACE. AP and AIx were independently associated with all-cause mortality (all p < 0.05). We demonstrated that central BP can predict poor long-term prognosis in patients with ESUS, especially those with the no cause ESUS subtype.

Vegetation Subtype Classification of Evergreen Broad-Leaved Forests in Mountainous Areas Using a Hierarchy-Based Classifier

Evergreen broad-leaved forests with rich biodiversity play a key role in stabilizing global vegetation productivity and maintaining land carbon sinks. However, quantitative and accurate classification results for humid, evergreen, broad-leaved forests (HEBF) and semi-humid evergreen broad-leaved forests (SEBF) with different vegetation productivity and significant differences in species composition are lacking. Remote sensing technology brings the possibility of vegetation subtype classification. Taking the mountainous evergreen broad-leaved forests distributed in Sichuan Province as an example, this study proposed a hierarchy-based classifier combined with environmental variables to quantitatively classify the two vegetation subtypes with different ecological characteristics but similar image features. Additionally, we applied Sun–Canopy–Sensor and C parameter(SCS + C) topographic correction to preprocess the images, effectively correcting the radiometric distortion and enhancing the accuracy of vegetation classification. Finally, achieving an overall accuracy (OA) of 87.91% and a Kappa coefficient of 0.76, which is higher than that of directly using the classifier to classify the two vegetation subtypes. The study revealed the widespread distribution of evergreen broad-leaved forests in Sichuan, with a clear boundary between the distribution areas of HEBF and SEBF. The HEBF in the east is located in the basin and the low marginal mountains; the SEBF is located in the southwest dry valley. The methods employed in this study offer an effective approach to vegetation classification in mountainous areas. The findings can provide guidance for ecological engineering construction, ecological protection, and agricultural and livestock development.

Classification of colorectal cancer consensus molecular subtypes using attention-based multi-instance learning network on whole-slide images

Colorectal cancer (CRC) is the third most common and second most lethal cancer globally. It is highly heterogeneous with different clinical-pathological characteristics, prognostic status, and therapy responses. Thus, the precise diagnosis of CRC subtypes is of great significance for improving the prognosis and survival of CRC patients. Nowadays, the most commonly used molecular-level CRC classification system is the Consensus Molecular Subtypes (CMSs). In this study, we applied a weakly supervised deep learning method, named attention-based multi-instance learning (MIL), on formalin-fixed paraffin-embedded (FFPE) whole-slide images (WSIs) to distinguish CMS1 subtype from CMS2, CMS3, and CMS4 subtypes, as well as distinguish CMS4 from CMS1, CMS2, and CMS3 subtypes. The advantage of MIL is training a bag of the tiled instance with bag-level labels only. Our experiment was performed on 1218 WSIs obtained from The Cancer Genome Atlas (TCGA). We constructed three convolutional neural network-based structures for model training and evaluated the ability of the max-pooling operator and mean-pooling operator on aggregating bag-level scores. The results showed that the 3-layer model achieved the best performance in both comparison groups. When compared CMS1 with CMS234, max-pooling reached the ACC of 83.86 % and the mean-pooling operator reached the AUC of 0.731. While comparing CMS4 with CMS123, mean-pooling reached the ACC of 74.26 % and max-pooling reached the AUC of 0.609. Our results implied that WSIs could be utilized to classify CMSs, and manual pixel-level annotation is not a necessity for computational pathology imaging analysis.

Characterization of stemness features and construction of a stemness subtype classifier to predict survival and treatment responses in lung squamous cell carcinoma

BackgroundCancer stemness has been proven to affect tumorigenesis, metastasis, and drug resistance in various cancers, including lung squamous cell carcinoma (LUSC). We intended to develop a clinically applicable stemness subtype classifier that could assist physicians in predicting patient prognosis and treatment response.MethodsThis study collected RNA-seq data from TCGA and GEO databases to calculate transcriptional stemness indices (mRNAsi) using the one-class logistic regression machine learning algorithm. Unsupervised consensus clustering was conducted to identify a stemness-based classification. Immune infiltration analysis (ESTIMATE and ssGSEA algorithms) methods were used to investigate the immune infiltration status of different subtypes. Tumor Immune Dysfunction and Exclusion (TIDE) and Immunophenotype Score (IPS) were used to evaluate the immunotherapy response. The pRRophetic algorithm was used to estimate the efficiency of chemotherapeutic and targeted agents. Two machine learning algorithms (LASSO and RF) and multivariate logistic regression analysis were performed to construct a novel stemness-related classifier.ResultsWe observed that patients in the high-mRNAsi group had a better prognosis than those in the low-mRNAsi group. Next, we identified 190 stemness-related differentially expressed genes (DEGs) that could categorize LUSC patients into two stemness subtypes. Patients in the stemness subtype B group with higher mRNAsi scores exhibited better overall survival (OS) than those in the stemness subtype A group. Immunotherapy prediction demonstrated that stemness subtype A has a better response to immune checkpoint inhibitors (ICIs). Furthermore, the drug response prediction indicated that stemness subtype A had a better response to chemotherapy but was more resistant to epidermal growth factor receptor tyrosine kinase inhibitors (EGFR-TKIs). Finally, we constructed a nine-gene-based classifier to predict patients’ stemness subtype and validated it in independent GEO validation sets. The expression levels of these genes were also validated in clinical tumor specimens.ConclusionThe stemness-related classifier could serve as a potential prognostic and treatment predictor and assist physicians in selecting effective treatment strategies for patients with LUSC in clinical practice.