Predictor Of Prognosis In Cancer Research Articles

Pathformer: a biological pathway informed transformer for disease diagnosis and prognosis using multi-omics data.

Multi-omics data provide a comprehensive view of gene regulation at multiple levels, which is helpful in achieving accurate diagnosis of complex diseases like cancer. However, conventional integration methods rarely utilize prior biological knowledge and lack interpretability. To integrate various multi-omics data of tissue and liquid biopsies for disease diagnosis and prognosis, we developed a biological pathway informed Transformer, Pathformer. It embeds multi-omics input with a compacted multi-modal vector and a pathway-based sparse neural network. Pathformer also leverages criss-cross attention mechanism to capture the crosstalk between different pathways and modalities. We first benchmarked Pathformer with 18 comparable methods on multiple cancer datasets, where Pathformer outperformed all the other methods, with an average improvement of 6.3%-14.7% in F1 score for cancer survival prediction, 5.1%-12% for cancer stage prediction, and 8.1%-13.6% for cancer drug response prediction. Subsequently, for cancer prognosis prediction based on tissue multi-omics data, we used a case study to demonstrate the biological interpretability of Pathformer by identifying key pathways and their biological crosstalk. Then, for cancer early diagnosis based on liquid biopsy data, we used plasma and platelet datasets to demonstrate Pathformer's potential of clinical applications in cancer screening. Moreover, we revealed deregulation of interesting pathways (e.g., scavenger receptor pathway) and their crosstalk in cancer patients' blood, providing potential candidate targets for cancer microenvironment study. Pathformer is implemented and freely available at https://github.com/lulab/Pathformer. Supplementary data are available at Bioinformatics online.

Prediction of ovarian cancer prognosis using statistical radiomic features of ultrasound images.

Ovarian cancer is the deadliest gynecologic malignancy worldwide. Ultrasound is the most useful non-invasive test for preoperative diagnosis of ovarian cancer. In this study, by leveraging multiple ultrasound images from the same patient to generate personalized, informative statistical radiomic features, we aimed to develop improved ultrasound image-based prognostic models for ovarian cancer. A total of 2,057 ultrasound images from 514 ovarian cancer patients, including 355 patients with epithelial ovarian cancer, from two hospitals in China were collected for this study. The models were constructed using our recently developed Frequency Appearance in Multiple Univariate pre-Screening (FAMUS) feature selection algorithm and Cox proportional hazards model. The models showed high predictive performance for overall survival (OS) and recurrence-free survival (RFS) in both epithelial and nonepithelial ovarian cancer, with concordance indices (C-index) ranging from 0.773 to 0.794. Radiomic scores predicted 2-year OS and RFS risk groups with significant survival differences (log-rank test, P<1.0e-4 for both validation cohorts). OS and RFS hazard ratios between low- and high-risk groups were 15.994 and 30.692 (internal cohort) and 19.339 and 19.760 (external cohort), respectively. The improved performance of these newly developed prognostic models was mainly attributed to the use of multiple preoperative ultrasound images from the same patient to generate statistical radiomic features, rather than simply using the largest tumor ROI among them. The models also revealed that the roundness of tumor lesion shape was positively correlated with prognosis for ovarian cancer. In summary, the newly developed prognostic models based on statistical radiomic features from ultrasound images were highly predictive of the risk of cancer-related death and possible recurrence not only for patients with epithelial ovarian cancer but also for those with nonepithelial ovarian cancer. They thereby provide reliable, non-invasive markers for individualized prognosis evaluation and clinical decision-making for patients with ovarian cancer.

A novel Natural killer cell-related gene signature for improving the prediction of prognosis and immunotherapy response in bladder cancer.

Bladder cancer (BLCA) is a commonly diagnosed cancer worldwide that exhibits high rates of recurrence and metastasis. Immunotherapy is increasingly being recognised in the clinical management of bladder cancer. In addition, the prospect of developing Natural Killer (NK) cell-related immunotherapy is promising in BLCA. We established and verified a prognostic signature based on NK cell-related gene expression. We then calculated the NKscore of BLCA samples and correlated it with the clinical outcomes, molecular subtypes of BLCA, tumour microenvironment (TME), and predicted efficacy of immune checkpoint inhibitors (ICI) and chemotherapy drugs to thoroughly explore the implications of the NKscore. Finally, the role of the NK signature gene HECTD1 in BLCA was verified by Quantitative Real-time PCR, Cell Counting Kit-8 Assay (CCK-8), Transwell Assay and Colony Formation Experiment. We analysed NK cell-associated genes and identified six genes with significant prognostic relevance. A high NK score significantly represents a worse prognosis. NKscore was significantly correlated with seven types of classical molecular subtype classifications of BLCA. In addition, NKscore positively correlates with NK-related immune checkpoints, suggesting that emerging NK cell immune checkpoint inhibitors, such as monalizumab, may have potential therapeutic promise for patients with high NKscore. The results of the T cell inflamed score (TIS) and tumour immune dysfunction exclusion (TIDE) score confirmed the suitability of immunotherapy for patients with a high NK score. Likewise, patients with a high NK score may be more suitable for several significant chemotherapeutic drugs. Functional experiments showed that the knockdown of HECTD1 significantly attenuated the proliferation, migration, and invasion ability of tumour cells. to sum up, the capability of our signature to predict prognosis and immunotherapy response was robust. Hopefully, these results will provide new insights for BLCA research and patient immunotherapy.

An Aging-Related lncRNA Signature Establishing for Breast Cancer Prognosis and Immunotherapy Responsiveness Prediction.

Emerging evidence demonstrates the vital role of aging and long non-coding RNAs (lncRNAs) in breast cancer (BC) progression. Our study intended to develop a prognostic risk model based on aging-related lncRNAs (AG-lncs) to foresee BC patients' outcomes. 307 aging-related genes (AGs) were sequenced from the TCGA project. Then, 697 AG-lncs were identified by the co-expression analysis with AGs. Using multivariate and univariate Cox regression analysis, and LASSO, 6 AG-lncs, including al136531.1, mapt-as1, al451085.2, otud6b-as1, tnfrsf14-as1, and linc01871, were validated to compute the risk score and establish a risk signature. Expression levels of al136531.1, mapt-as1, al451085.2, tnfrsf14-as1, and linc01871 were higher in low-risk BC patients, whereas otud6b-as1 expression was higher in high-risk BC patients. In the training and testing set, high-risk patients performed shorter PFI, OS, and DFS than low-risk patients. Our risk signature had the highest concordance index among other established prognostic signatures and displayed ideal predictive ability for 1-, 3- and 5-year patient OS in the nomogram. Additionally, BC patients with different risk score levels showed different immune statuses and responses to immunotherapy via GSEA, ssGSEA, ESTIMATE algorithm, and TIDE algorithm analysis. Of note, the qRT-PCR analysis validated that these 6 AG-lncs expressed quite differentially in BC tissues at various clinical stages. The risk signature of 6 AG-lncs might offer a novel prognostic biomarker and promisingly enhance BC immunotherapy's effectiveness.

Harnessing TME depicted by histological images to improve cancer prognosis through a deep learning system

Spatial transcriptomics (ST) provides insights into the tumor microenvironment (TME), which is closely associated with cancer prognosis, but ST has limited clinical availability. In this study, we provide a powerful deep learning system to augment TME information based on histological images for patients without ST data, thereby empowering precise cancer prognosis. The system provides two connections to bridge existing gaps. The first is the integrated graph and image deep learning (IGI-DL) model, which predicts ST expression based on histological images with a 0.171 increase in mean correlation across three cancer types compared with five existing methods. The second connection is the cancer prognosis prediction model, based on TME depicted by spatial gene expression. Our survival model, using graphs with predicted ST features, achieves superior accuracy with a concordance index of 0.747 and 0.725 for The Cancer Genome Atlas breast cancer and colorectal cancer cohorts, outperforming other survival models. For the external Molecular and Cellular Oncologycolorectal cancer cohort, our survival model maintains a stable advantage.

Impact of the serum creatinine and cystatin C ratio for prediction of sarcopenia and prognosis in biliary tract cancer.

Sarcopenia is a poor prognostic factor in cancer patients. In recent years, there have been reports that serum creatinine and cystatin C (Cr/CysC) ratio is associated with sarcopenia. However, the prognostic value of the Cr/CysC ratio in biliary tract cancer is unclear. We evaluated the impact of the Cr/CysC ratio on sarcopenia and prognosis in biliary tract cancer. We retrospectively reviewed the records of 190 patients with biliary tract cancer who had undergone surgical resection from January 2017 to March 2023. Frozen serum samples collected at the time of surgery were used to measure CysC. We calculated the Cr/CysC ratio and investigated the relationship with sarcopenia and the prognostic significance. We calculated the cutoff value of the Cr/CysC ratio for low skeletal muscle index (SMI) (< 42cm2/m2 for males and < 38cm2/m2 for females). The optimal cutoff value of the Cr/CysC ratio was 0.848. The low Cr/CysC ratio group was significantly associated with higher preoperative CRP and lower albumin, lower SMI, lower handgrip strength, and higher intramuscular adipose tissue content. In multivariate analysis, patients with a low Cr/CysC ratio showed poorer overall survival (hazard ratio 2.60, 95% confidence interval 1.07-6.29, p = 0.033), which was significantly worse than in those with a high Cr/CysC ratio. In patients with biliary tract cancer, the Cr/CysC ratio showed weak correlation with sarcopenic indicators. However, the Cr/CysC ratio could be strong prognostic factor in biliary tract cancer.

Bilateral efforts to improve SERS detection efficiency of exosomes by Au/Na7PMo11O39 Combined with Phospholipid Epitope Imprinting

Detection of cancer-related exosomes in body fluids has become a revolutionary strategy for early cancer diagnosis and prognosis prediction. We have developed a two-step targeting detection method, termed PS-MIPs-NELISA SERS, for rapid and highly sensitive exosomes detection. In the first step, a phospholipid polar site imprinting strategy was employed using magnetic PS-MIPs (phospholipids-molecularly imprinted polymers) to selectively isolate and enrich all exosomes from urine samples. In the second step, a nanozyme-linked immunosorbent assay (NELISA) technique was utilized. We constructed Au/Na7PMo11O39 nanoparticles (NPs) with both surface-enhanced Raman scattering (SERS) property and peroxidase catalytic activity, followed by the immobilization of CD9 antibodies on the surface of Au/Na7PMo11O39 NPs. The Au/Na7PMo11O39-CD9 antibody complexes were then used to recognize CD9 proteins on the surface of exosomes enriched by magnetic PS-MIPs. Lastly, the high sensitivity detection of exosomes was achieved indirectly via the SERS activity and peroxidase-like activity of Au/Na7PMo11O39 NPs. The quantity of exosomes in urine samples from pancreatic cancer patients obtained by the PS-MIPs-NELISA SERS technique showed a linear relationship with the SERS intensity in the range of 6.21 × 107–2.81 × 108 particles/mL, with a limit of detection (LOD) of 5.82 × 107 particles/mL. The SERS signal intensity of exosomes in urine samples from pancreatic cancer patients was higher than that of healthy volunteers. This bidirectional MIPs-NELISA-SERS approach enables noninvasive, highly sensitive, and rapid detection of cancer, facilitating the monitoring of disease progression during treatment and opening up a new avenue for rapid early cancer screening.

Extracellular Vesicles for Childhood Cancer Liquid Biopsy.

Liquid biopsy involves the utilization of minimally invasive or noninvasive techniques to detect biomarkers in biofluids for disease diagnosis, monitoring, or guiding treatments. This approach is promising for the early diagnosis of childhood cancer, especially for brain tumors, where tissue biopsies are more challenging and cause late detection. Extracellular vesicles offer several characteristics that make them ideal resources for childhood cancer liquid biopsy. Extracellular vesicles are nanosized particles, primarily secreted by all cell types into body fluids such as blood and urine, and contain molecular cargos, i.e., lipids, proteins, and nucleic acids of original cells. Notably, the lipid bilayer-enclosed structure of extracellular vesicles protects their cargos from enzymatic degradation in the extracellular milieu. Proteins and nucleic acids of extracellular vesicles represent genetic alterations and molecular profiles of childhood cancer, thus serving as promising resources for precision medicine in cancer diagnosis, treatment monitoring, and prognosis prediction. This review evaluates the recent progress of extracellular vesicles as a liquid biopsy platform for various types of childhood cancer, discusses the mechanistic roles of molecular cargos in carcinogenesis and metastasis, and provides perspectives on extracellular vesicle-guided therapeutic intervention. Extracellular vesicle-based liquid biopsy for childhood cancer may ultimately contribute to improving patient outcomes.

Conditional survival analysis and real-time prognosis prediction in stage III T3–T4 colon cancer patients after surgical resection: a SEER database analysis

BackgroundConditional survival (CS) takes into consideration the duration of survival post-surgery and can provide valuable additional insights. The aim of this study was to investigate the risk factors associated with reduced one-year postoperative conditional survival in patients diagnosed with stage III T3–T4 colon cancer and real-time prognosis prediction. Furthermore, we aim to develop pertinent nomograms and predictive models.MethodsClinical data and survival outcomes of patients diagnosed with stage III T3–T4 colon cancer were obtained from the Surveillance, Epidemiology, and End Results (SEER) database, covering the period from 2010 to 2019. Patients were divided into training and validation cohorts at a ratio of 7:3. The training set consisted of a total of 11,386 patients for conditional overall survival (cOS) and 11,800 patients for conditional cancer-specific survival (cCSS), while the validation set comprised 4876 patients for cOS and 5055 patients for cCSS. Univariate and multivariate Cox regression analyses were employed to identify independent risk factors influencing one-year postoperative cOS and cCSS. Subsequently, predictive nomograms for cOS and cCSS at 2-year, 3-year, 4-year, and 5-year intervals were constructed based on the identified prognostic factors. The performance of these nomograms was rigorously assessed through metrics including the concordance index (C-index), calibration curves, and the area under curve (AUC) derived from the receiver operating characteristic (ROC) analysis. Clinical utility was further evaluated using decision curve analysis (DCA).ResultsA total of 18,190 patients diagnosed with stage III T3–T4 colon cancer were included in this study. Independent risk factors for one-year postoperative cOS and cCSS included age, pT stage, pN stage, pretreatment carcinoembryonic antigen (CEA) levels, receipt of chemotherapy, perineural invasion (PNI), presence of tumor deposits, the number of harvested lymph nodes, and marital status. Sex and tumor site were significantly associated with one-year postoperative cOS, while radiation therapy was notably associated with one-year postoperative cCSS. In the training cohort, the developed nomogram demonstrated a C-index of 0.701 (95% CI, 0.711–0.691) for predicting one-year postoperative cOS and 0.701 (95% CI, 0.713–0.689) for one-year postoperative cCSS. Following validation, the C-index remained robust at 0.707 (95% CI, 0.721–0.693) for one-year postoperative cOS and 0.700 (95% CI, 0.716–0.684) for one-year postoperative cCSS. ROC and calibration curves provided evidence of the model's stability and reliability. Furthermore, DCA underscored the nomogram’s superior clinical utility.ConclusionsOur study developed nomograms and predictive models for postoperative stage III survival in T3–T4 colon cancer with the aim of accurately estimating conditional survival. Survival bias in our analyses may lead to overestimation of survival outcomes, which may limit the applicability of our findings.

Deep Learning in Breast Cancer Imaging: State of the Art and Recent Advancements in Early 2024.

The rapid advancement of artificial intelligence (AI) has significantly impacted various aspects of healthcare, particularly in the medical imaging field. This review focuses on recent developments in the application of deep learning (DL) techniques to breast cancer imaging. DL models, a subset of AI algorithms inspired by human brain architecture, have demonstrated remarkable success in analyzing complex medical images, enhancing diagnostic precision, and streamlining workflows. DL models have been applied to breast cancer diagnosis via mammography, ultrasonography, and magnetic resonance imaging. Furthermore, DL-based radiomic approaches may play a role in breast cancer risk assessment, prognosis prediction, and therapeutic response monitoring. Nevertheless, several challenges have limited the widespread adoption of AI techniques in clinical practice, emphasizing the importance of rigorous validation, interpretability, and technical considerations when implementing DL solutions. By examining fundamental concepts in DL techniques applied to medical imaging and synthesizing the latest advancements and trends, this narrative review aims to provide valuable and up-to-date insights for radiologists seeking to harness the power of AI in breast cancer care.

Proximity-Guaranteed DNA Machine for Accurate Identification of Breast Cancer Extracellular Vesicles.

Breast cancer is one of the most diagnosed cancers worldwide. Precise diagnosis and subtyping have important significance for targeted therapy and prognosis prediction of breast cancer. Herein, we design a proximity-guaranteed DNA machine for accurate identification of breast cancer extracellular vesicles (EVs), which is beneficial to explore the subtype features of breast cancer. In our design, two proximity probes are located close on the same EV through specific recognition of coexisting surface biomarkers, thus being ligated with the help of click chemistry. Then, the ligated product initiates the operation of a DNA machine involving catalytic hairpin assembly and clusters of regularly interspaced short palindromic repeats (CRISPR)-Cas12a-mediated trans-cleavage, which finally generates a significant response that enables the identification of EVs expressing both biomarkers. Principle-of-proof studies are performed using EVs derived from the breast cancer cell line BT474 as the models, confirming the high sensitivity and specificity of the DNA machine. When further applied to clinical samples, the DNA machine is shown to be capable of not only distinguishing breast cancer patients with special subtypes but also realizing the tumor staging regarding the disease progression. Therefore, our work may provide new insights into the subtype-based diagnosis of breast cancer as well as identification of more potential therapeutic targets in the future.

Extracellular vesicle‐based biomarker in head and neck cancer: prospects and challenges

AbstractHead and neck squamous cell carcinoma (HNSCC) represents the sixth most common malignancy worldwide. However, very few established diagnostic biomarkers for HNSCC have been universally applied in clinical practice. Recently, much attention has been paid to extracellular vesicles (EVs) regarding their roles as cancer biomarkers because EVs carry plentiful cargoes, including lipids, proteins, nucleic acids, and metabolites. In HNSCC, several molecules carried by EVs, which are derived from peripheral blood and saliva, have been implicated to be effective in cancer detection, staging, treatment planning, response monitoring, and prognosis prediction. Although several EV molecules have been identified to be significantly correlated with a set of clinical‐pathological parameters of HNSCC, several key limitations need to be resolved before the clinical application of EVs as carriers of biomarkers in HNSCC. In this review, we discuss current knowledge in the literature regarding EV‐based biomarkers in HNSCC, emphasizing current limitations of their clinical applications.

The Construction of a Multi-Gene Risk Model for Colon Cancer Prognosis and Drug Treatments Prediction.

In clinical practice, colon cancer is a prevalent malignant tumor of the digestive system, characterized by a complex and progressive process involving multiple genes and molecular pathways. Historically, research efforts have primarily focused on investigating individual genes; however, our current study aims to explore the collective impact of multiple genes on colon cancer and to identify potential therapeutic targets associated with these genes. For this research, we acquired the gene expression profiles and RNA sequencing data of colon cancer from TCGA. Subsequently, we conducted differential gene expression analysis using R, followed by GO and KEGG pathway enrichment analyses. To construct a protein-protein interaction (PPI) network, we selected survival-related genes using the log-rank test and single-factor Cox regression analysis. Additionally, we performed LASSO regression analysis, immune infiltration analysis, mutation analysis, and cMAP analysis, as well as an investigation into ferroptosis. Our differential expression and survival analyses identified 47 hub genes, and subsequent LASSO regression analysis refined the focus to 23 key genes. These genes are closely linked to cancer metastasis, proliferation, apoptosis, cell cycle regulation, signal transduction, cancer microenvironment, immunotherapy, and neurodevelopment. Overall, the hub genes discovered in our study are pivotal in colon cancer and are anticipated to serve as important biological markers for the diagnosis and treatment of the disease.

ЗАСТОСУВАННЯ ГЛИБОКИХ ШТУЧНИХ НЕЙРОННИХ МЕРЕЖ ДЛЯ КЛАСИФІКАЦІЇ МУЛЬТИМОДАЛЬНИХ ДАНИХ

Multimodal data analysis is gaining attention in recent research. Pu Liang et al. (2023) provide a comprehensive overview on multimodal machine learning, highlighting its founda-tions, challenges and achievements in recent years. More problem-oriented works propose new methods and applications for multimodal ML, such a Ngiam et al. (2011) propose to use joint audio and video data to improve speech recognition accuracy; Sun, Wand and Li (2018) describe application of multimodal classification for breast cancer prognosis prediction; Mao et al. (2014) propose an architecture of multimodal recurrent network to generate text de-scription of images and so on. However, such works usually focus on the task itself and meth-ods therein, and not on integrating multimodal data processing into other software systems. The goal of this research is to propose a way to conduct multimodal data processing, specifically as a part of a digital twin systems, thus efficiency and near-real-time operation are required. The paper presents an approach to conduct parallel multimodal data classification, adapting to available computing power. The method is modular and scalable and intended for in digital twin application as a part of analysis and modeling tools. Later, the detailed example of such a software module is discussed. It uses multimodal data from open datasets to detect and classify the behavior of pets using deep learning mod-els. Videos are processed using two artificial neural networks: YOLOv3 object detection net-work to process individual frames of the video and a relatively simple convolutional network to classify sounds based on their frequency spectra. Constructed module uses a producer-consumer parallel processing pattern and allows processing 5 frames per second of a video on available hardware, which can be sufficiently improved by using GPU acceleration or more paralleled processing threads.

Identifying a Novel Eight-NK Cell-related Gene Signature for Ovarian Cancer Prognosis Prediction.

Ovarian cancer (OVC) is the most common and costly tumor in the world with unfavorable overall survival and prognosis. This study is aimed to explore the prognostic value of natural killer cells related genes for OVC treatment. RNA-seq and clinical information were acquired from the TCGA-OVC dataset (training dataset) and the GSE51800 dataset (validation dataset). Genes linked to NK cells were obtained from the immPort dataset. Moreover, ConsensusClusterPlus facilitated the screening of molecular subtypes. Following this, the risk model was established by LASSO analysis, and immune infiltration and immunotherapy were then detected by CIBERSORT, ssGSEA, ESTIMATE, and TIDE algorithms. Based on 23 NK cell-related genes with prognosis, TCGA-OVC samples were classified into two clusters, namely C1 and C2. Of these, C1 had better survival outcomes as well as enhanced immune infiltration and tumor stem cells. Additionally, it was more suitable for immunotherapy and was also sensitive to traditional chemotherapy drugs. The eight-gene prognosis model was constructed and verified via the GSE51800 dataset. Additionally, a high infiltration level of immune cells was observed in low-risk patients. Low-risk samples also benefited from immunotherapy and chemotherapy drugs. Finally, a nomogram and ROC curves were applied to validate model accuracy. The present study identified a RiskScore signature, which could stratify patients with different infiltration levels, immunotherapy, and chemotherapy drugs. Our study provided a basis for precisely evaluating OVC therapy and prognosis.

Mechanistic insights into FEN1-mediated drug sensitivity and risk signature in colon cancer: An integrative bioinformatics study.

The overexpression of Flap endonuclease 1 (FEN1) has been implicated in drug resistance and prognosis across various cancer types. However, the precise role of FEN1 in colon cancer remains to be fully elucidated. In this study, we employed comprehensive datasets from The Cancer Genome Atlas, Gene Expression Omnibus, and Human Protein Atlas to examine FEN1 expression and assess its correlation with clinical pathology and prognosis in colon cancer. We utilized the pRRophetic algorithm to evaluate drug sensitivity and performed differential expression analysis to identify genes associated with FEN1-mediated drug sensitivity. Gene set enrichment analysis was conducted to further investigate these genes. Additionally, single-cell sequencing analysis was employed to explore the relationship between FEN1 expression and functional states. Cox regression analysis was implemented to construct a prognostic model, and a nomogram for prognosis was developed. Our analysis of The Cancer Genome Atlas and Gene Expression Omnibus datasets revealed a significant upregulation of FEN1 in colon cancer. However, while FEN1 expression showed no notable correlation with prognosis, it displayed associations with metastasis. Single-cell sequencing analysis further confirmed a positive correlation between FEN1 expression and colon cancer metastasis. Furthermore, we detected marked discrepancies in drug responsiveness between the High_FEN1 and Low_FEN1 groups, identifying 342 differentially expressed genes. Enrichment analysis showed significant suppression in processes related to DNA replication, spliceosome, and cell cycle pathways in the Low_FEN1 group, while the calcium signaling pathway, cAMP signaling pathway, and other pathways were activated. Of the 197 genes differentially expressed and strongly linked to FEN1 expression, 39 were significantly implicated in colon cancer prognosis. Finally, we constructed a risk signature consisting of 5 genes, which, when combined with drug treatment and pathological staging, significantly improved the prediction of colon cancer prognosis. This study offers novel insights into the interplay among FEN1 expression levels, colon cancer metastatic potential, and sensitivity to therapeutic agents. Furthermore, we successfully developed a multi-gene prognostic risk signature derived from FEN1.

Abstract 4939: Computational pathology approach for prognostic advancements and therapeutic benefits in gastrointestinal cancer: A multi-centric retrospective study

Abstract Background: The current cancer staging methods cannot accurately predict survival and therapeutic benefits in cancer patients. Digital pathomics is an emerging field with potential to revolutionize disease evaluation. The present study aims to develop and validate a deep learning pathomics signature from digital hematoxylin-eosin (H&amp;E) staining slides to predict the survival of pan-gastrointestinal cancer, and further investigated its associations with chemotherapy and immunotherapy response. P atients and Methods: This international multicenter study included 2,463 patients with pan-gastrointestinal cancer from twelve cohorts of seven cancer centers, among which 1653 patients were diagnosed with GC. We proposed a deep learning pathomics signature (DLPS) by integrating information on three scales from the whole slide images (WSIs) of H&amp;E, including pathomics nucleus, pathomics microenvironment, and single-cell spatial distribution features. Next, we evaluated the predictive accuracy of DLPS for prognosis, chemotherapy response, and immunotherapy response. To interpret the predictive ability of the DLPS, we analyzed multi-omics data and employed the Shapley value strategy to provide its biological insights and importance ranking. Results: The DLPS was significantly associated with overall survival in patients with GC (hazard ratio range: 1.203-2.717, all p &amp;lt; 0.001), which was further validated in other gastrointestinal cancers (all p &amp;lt; 0.02). The DLPS remained an independent predictor of prognosis in multivariable analysis (all p &amp;lt; 0.001). Furthermore, a nomogram incorporating the DLPS and TNM stage shows significantly improved accuracy in predicting cancer prognosis compared to that with TNM stage alone (all p &amp;lt; 0.05). Shapley value analysis highlighted DLPS as the strongest predictor for prognosis. Importantly, GC patients with a low-DLPS (but not those with a high-DLPS) exhibited substantial benefits from adjuvant chemotherapy (all p &amp;lt; 0.05). Furthermore, we found the objective responses of anti-PD-1 immunotherapy is significantly higher in the low-DLPS group (29.6%) than in the high-DLPS group (8.3%, p &amp;lt; 0.05). Upon analyzing multi-omics data, we found that a higher DLPS was positively correlated with tumor promoting, chemotherapy resistance, immune tumor microenvironment and metabolic signaling. Conclusion: The DLPS enabled improved assessment on prognosis, and has the potential to identify patients who will benefit from adjuvant chemotherapy and immunotherapy, which can further be extended to many gastrointestinal cancers or other solid tumors. Citation Format: Taojun Zhang, Zepang Sun, Zhe Li, M. Usman Ahmad, Md Tauhidul Islam, Fan Yang, Zhenhui Li, Yuming Jiang. Computational pathology approach for prognostic advancements and therapeutic benefits in gastrointestinal cancer: A multi-centric retrospective study [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 4939.

Deep learning on tertiary lymphoid structures in hematoxylin-eosin predicts cancer prognosis and immunotherapy response

Tertiary lymphoid structures (TLSs) have been associated with favorable immunotherapy responses and prognosis in various cancers. Despite their significance, their quantification using multiplex immunohistochemistry (mIHC) staining of T and B lymphocytes remains labor-intensive, limiting its clinical utility. To address this challenge, we curated a dataset from matched mIHC and H&E whole-slide images (WSIs) and developed a deep learning model for automated segmentation of TLSs. The model achieved Dice coefficients of 0.91 on the internal test set and 0.866 on the external validation set, along with intersection over union (IoU) scores of 0.819 and 0.787, respectively. The TLS ratio, defined as the segmented TLS area over the total tissue area, correlated with B lymphocyte levels and the expression of CXCL13, a chemokine associated with TLS formation, in 6140 patients spanning 16 tumor types from The Cancer Genome Atlas (TCGA). The prognostic models for overall survival indicated that the inclusion of the TLS ratio with TNM staging significantly enhanced the models’ discriminative ability, outperforming the traditional models that solely incorporated TNM staging, in 10 out of 15 TCGA tumor types. Furthermore, when applied to biopsied treatment-naïve tumor samples, higher TLS ratios predicted a positive immunotherapy response across multiple cohorts, including specific therapies for esophageal squamous cell carcinoma, non-small cell lung cancer, and stomach adenocarcinoma. In conclusion, our deep learning-based approach offers an automated and reproducible method for TLS segmentation and quantification, highlighting its potential in predicting immunotherapy response and informing cancer prognosis.

Identification and verification of AK4 as a protective immune-related biomarker in adipose-derived stem cells and breast cancer

BackgroundBreast cancer (BC) remains the most common cancer among women, and novel post-surgical reconstruction techniques, including autologous fat transplantation, have emerged. While Adipose-derived stem cells (ADSCs) are known to impact the viability of fat grafts, their influence on breast cancer progression remains unclear. This study aims to elucidate the genetic interplay between ADSCs and breast cancer, focusing on potential therapeutic targets. MethodsUsing the GEO and TCGA databases, we pinpointed differentially expressed (DE) mRNAs, miRNAs, lncRNAs, and pseudogenes of ADSCs and BC. We performed functional enrichment analysis and constructed protein-protein interaction (PPI), RNA binding protein (RBP)-pseudogene-mRNA, and lncRNA-miRNA-transcription factor (TF)-gene networks. Our study delved into the correlation of AK4 expression with 33 different malignancies and examined its impact on prognostic outcomes across a pan-cancer cohort. Additionally, we scrutinized immune infiltration, microsatellite instability, and tumor mutational burden, and conducted single-cell analysis to further understand the implications of AK4 expression. We identified novel sample subtypes based on hub genes using the ConsensusClusterPlus package and examined their association with immune infiltration. The random forest algorithm was used to screen DE mRNAs between subtypes to validate the powerful prognostic prediction ability of the artificial neural network. ResultsOur analysis identified 395 DE mRNAs, 3 DE miRNAs, 84 DE lncRNAs, and 26 DE pseudogenes associated with ADSCs and BC. Of these, 173 mRNAs were commonly regulated in both ADSCs and breast cancer, and 222 exhibited differential regulation. The PPI, RBP-pseudogene-mRNA, and lncRNA-miRNA-TF-gene networks suggested AK4 as a key regulator. Our findings support AK4 as a promising immune-related therapeutic target for a wide range of malignancies. We identified 14 characteristic genes based on the AK4-related cluster using the random forest algorithm. Our artificial neural network yielded excellent diagnostic performance in the testing cohort with AUC values of 0.994, 0.973, and 0.995, indicating its ability to distinguish between breast cancer and non-breast cancer cases. ConclusionsOur research sheds light on the dual role of ADSCs in BC at the genetic level and identifies AK4 as a key protective mRNA in breast cancer. We found that AK4 significantly predicts cancer prognosis and immunotherapy, indicating its potential as a therapeutic target.

Diagnostic and prognostic value of circulating exosomal glypican-1 in pancreatic cancer: a meta-analysis.

Pancreatic cancer (PC) is usually detected in the advanced stages. Liquid biopsy has become a revolutionary strategy for cancer diagnosis and prognosis prediction. This study aims to investigate the diagnostic and prognostic value of circulating exosomal glypican-1 (GPC-1) in PC. We systematically searched relevant studies. For diagnostic accuracy, pooled sensitivity and specificity and the area under the summary receiver operating characteristic curve (AUC) were calculated. Regarding prognostic value, hazard ratios (HRs) and 95% CIs for overall survival (OS) were summarized by using a random-effects model. We found 8 studies that examined the diagnostic value of circulating exosomal GPC-1 in PC, and 3 studies that investigated its prognostic value. Pooled sensitivity and specificity were 0.88 (95% CI, 0.65-0.97) and 0.86 (95% CI, 0.72-0.94). The AUC was 0.93 (95% CI, 0.90-0.95). Prognostic analysis showed that higher levels of circulating exosomal GPC-1 were associated with poorer OS in PC patients, and the combined HR for OS was 4.59 (random-effects model, 95% CI=1.17-18.03, P=.022). The results of both studies were robust and neither had publication bias. Circulating exosomal GPC-1 may be used as a diagnostic and prognostic biomarker for PC. However, this result needs to be validated by further research using a larger sample size.