Robust deep learning model for prognostic stratification of pancreatic ductal adenocarcinoma patients.
Ontology highlight
ABSTRACT:
SUBMITTER: Ju J
PROVIDER: S-EPMC8637475 | biostudies-literature |
REPOSITORIES: biostudies-literature
Similar Datasets
Project description:Histopathological diagnosis of pancreatic ductal adenocarcinoma (PDAC) on endoscopic ultrasonography-guided fine-needle biopsy (EUS-FNB) specimens has become the mainstay of preoperative pathological diagnosis. However, on EUS-FNB specimens, accurate histopathological evaluation is difficult due to low specimen volume with isolated cancer cells and high contamination of blood, inflammatory and digestive tract cells. In this study, we performed annotations for training sets by expert pancreatic pathologists and trained a deep learning model to assess PDAC on EUS-FNB of the pancreas in histopathological whole-slide images. We obtained a high receiver operator curve area under the curve of 0.984, accuracy of 0.9417, sensitivity of 0.9302 and specificity of 0.9706. Our model was able to accurately detect difficult cases of isolated and low volume cancer cells. If adopted as a supportive system in routine diagnosis of pancreatic EUS-FNB specimens, our model has the potential to aid pathologists diagnose difficult cases.
| S-EPMC8055968 | biostudies-literature
Project description:Background and aimsEndoscopic ultrasonography-guided fine-needle aspiration/biopsy (EUS-FNA/B) is considered to be a first-line procedure for the pathological diagnosis of pancreatic cancer owing to its high accuracy and low complication rate. The number of new cases of pancreatic ductal adenocarcinoma (PDAC) is increasing, and its accurate pathological diagnosis poses a challenge for cytopathologists. Our aim was to develop a hyperspectral imaging (HSI)-based convolution neural network (CNN) algorithm to aid in the diagnosis of pancreatic EUS-FNA cytology specimens.MethodsHSI images were captured of pancreatic EUS-FNA cytological specimens from benign pancreatic tissues (n = 33) and PDAC (n = 39) prepared using a liquid-based cytology method. A CNN was established to test the diagnostic performance, and Attribution Guided Factorization Visualization (AGF-Visualization) was used to visualize the regions of important classification features identified by the model.ResultsA total of 1913 HSI images were obtained. Our ResNet18-SimSiam model achieved an accuracy of 0.9204, sensitivity of 0.9310 and specificity of 0.9123 (area under the curve of 0.9625) when trained on HSI images for the differentiation of PDAC cytological specimens from benign pancreatic cells. AGF-Visualization confirmed that the diagnoses were based on the features of tumor cell nuclei.ConclusionsAn HSI-based model was developed to diagnose cytological PDAC specimens obtained using EUS-guided sampling. Under the supervision of experienced cytopathologists, we performed multi-staged consecutive in-depth learning of the model. Its superior diagnostic performance could be of value for cytologists when diagnosing PDAC.
| S-EPMC10501295 | biostudies-literature
Project description:BackgroundAdvances in genomics have greatly improved the survival rate in cancer patients. However, due to genetic heterogeneity, pancreatic ductal adenocarcinoma (PDAC) is still difficult to diagnose early, and its survival rate is extremely low. Therefore, we identified biomarkers that predict the prognosis of PDAC patients using independent cohort data.Materials and methodsTo develop a novel prognostic biomarker, we used the gene expression and clinical data from The Cancer Genome Atlas (TCGA) and Gene Expression Omnibus (GEO). Kaplan-Meier survival curve using median values of genes as cutoff showed that EIF4G1 was the only statistically significant gene in the 3 cohorts. We analyzed the prognostic significance of EIF4G1 using the time-dependent area under the curve (AUC) of Uno's C-index, the AUC value of the receiver operating characteristics (ROC) at 3 years, and multivariate Cox analysis. We also compared EIF4G1 levels between tumors and matched non-tumor tissues.ResultsEIF4G1 is the only prognostic gene in patients with PDAC, which was selected by Kaplan-Meier survival analysis. The survival curve showed that high expression of EIF4G1 was associated with poor prognosis of PDAC with a good discriminative ability in 3 independent cohorts. The risk stratifying ability of EIF4G1 was demonstrated by analyzing C-indices and AUC values. Multivariate Cox regression confirmed its prognostic significance. EIF4G1 expression was significantly higher in PDAC tissues than in the matched normal tissues.ConclusionEIF4G1 could be used as a novel prognostic marker for PDAC and to determine suitable treatment options.
| S-EPMC6472433 | biostudies-literature
Project description:Identification of pancreatic ductal adenocarcinoma (PDAC) and precursor lesions in histological tissue slides can be challenging and elaborate, especially due to tumor heterogeneity. Thus, supportive tools for the identification of anatomical and pathological tissue structures are desired. Deep learning methods recently emerged, which classify histological structures into image categories with high accuracy. However, to date, only a limited number of classes and patients have been included in histopathological studies. In this study, scanned histopathological tissue slides from tissue microarrays of PDAC patients (n = 201, image patches n = 81.165) were extracted and assigned to a training, validation, and test set. With these patches, we implemented a convolutional neuronal network, established quality control measures and a method to interpret the model, and implemented a workflow for whole tissue slides. An optimized EfficientNet algorithm achieved high accuracies that allowed automatically localizing and quantifying tissue categories including pancreatic intraepithelial neoplasia and PDAC in whole tissue slides. SmoothGrad heatmaps allowed explaining image classification results. This is the first study that utilizes deep learning for automatic identification of different anatomical tissue structures and diseases on histopathological images of pancreatic tissue specimens. The proposed approach is a valuable tool to support routine diagnostic review and pancreatic cancer research.
| S-EPMC8160892 | biostudies-literature
Project description:Pancreatic adenocarcinomas (PAADs) often remain undiagnosed until later stages, limiting treatment options and leading to poor survival. The lack of robust biomarkers complicates PAAD prognosis, and patient risk stratification remains a major challenge. To address this issue, we established a panel constructed by four miRNAs (miR-4444-2, miR-934, miR-1301 and miR-3655) based on The Cancer Genome Atlas (TCGA) and Human Cancer Metastasis Database (HCMDB) to predicted the prognosis of PAAD patients. Then, a risk prediction model of these four miRNAs was constructed by using Cox regression analysis with the least absolute shrinkage and selection operator (LASSO) regression analysis. This model stratified TCGA PAAD cohort into the low-risk and high-risk groups based on the panel-based risk score, which was significantly associated with 1-, 2-, 3-year OS (AUC=0.836, AUC=0.844, AUC=0.952, respectively). The nomogram was then established with a robust performance signature for predicting prognosis compared to clinical characteristics of pancreatic cancer (PC) patients, including age, gender and clinical stage. Moreover, two GSE data were validated the expressions of 4 miRNAs with prognosis/survival outcome in PC. In the external clinical sample validation, the high-risk group with the upregulated expressions of miR-934/miR-4444-2 and downregulated expressions of miR-1301/miR-3655 were indicated a poor prognosis. Furthermore, the cell counting kit-8 (CCK-8) assay, clone formation, transwell and wound healing assay also confirmed the promoting effect of miR-934/miR-4444-2 and the inhibiting effect of miR-1301/miR-3655 in PC cell proliferation and migration. Taken together, we identified a new 4-miRNA risk stratification model could be used in predicting prognosis in PAAD.
| S-EPMC8851918 | biostudies-literature
Project description:BackgroundThe prognosis of patients with pancreatic ductal adenocarcinoma (PDAC) of pancreatic head remains poor, even after potentially curative R0 resection. The aim of this study was to develop an accurate model to predict patients' prognosis for PDAC of pancreatic head following pancreaticoduodenectomy.MethodsWe retrospectively reviewed 112 patients with PDAC of pancreatic head after pancreaticoduodenectomy in Guangdong Provincial People's Hospital between 2014 and 2018.ResultsFive prognostic factors were identified using univariate Cox regression analysis, including age, histologic grade, American Joint Committee on Cancer (AJCC) Stage 8th, total bilirubin (TBIL), CA19-9. Using all subset analysis and multivariate Cox regression analysis, we developed a nomogram consisted of age, AJCC Stage 8th, perineural invasion, TBIL, and CA19-9, which had higher C-indexes for OS (0.73) and RFS (0.69) compared with AJCC Stage 8th alone (OS: 0.66; RFS: 0.67). The area under the curve (AUC) values of the receiver operating characteristic (ROC) curve for the nomogram for OS and RFS were significantly higher than other single parameter, which are AJCC Stage 8th, age, perineural invasion, TBIL, and CA19-9. Importantly, our nomogram displayed higher C-index for OS than previous reported models, indicating a better predictive value of our model.ConclusionsA simple and practical nomogram for patient prognosis in PDAC of pancreatic head following pancreaticoduodenectomy was established, which shows satisfactory predictive efficacy and deserves further evaluation in the future.
| S-EPMC8216341 | biostudies-literature
Project description:PurposeClinical prognostic criteria using preoperative factors were not developed for post-neoadjuvant therapy (NAT) surgery of pancreatic ductal adenocarcinoma (PDAC). We aimed to identify preoperative factors associated with overall survival (OS) in PDAC patients who underwent post-NAT curative-intent surgery and develop risk stratification criteria.Materials and methodsConsecutive PDAC patients who underwent post-NAT curative-intent surgeries between 2007 and 2020 were retrospectively analyzed. Demographic, laboratory, surgical, and histopathologic variables were collected. Baseline, preoperative, and interval changes of computed tomography (CT) findings proposed by the Society of Abdominal Radiology and the American Pancreatic Association were analyzed. Cox proportional hazard analysis was used to select preoperative variables associated with OS. We developed risk stratification criteria composed of the significant preoperative variables, i.e., post-NAT response criteria. We compared the discrimination performance of post-NAT response criteria with that of post-NAT pathological (yp) American Joint Cancer Committee TNM staging system.ResultsOne hundred forty-five PDAC patients were included. Stable or increased tumor size on CT (hazard ratio [HR], 2.58; 95% confidence interval [CI], 1.58 to 4.21; p < 0.001) and elevated preoperative carbohydrate antigen 19-9 (CA19-9) level (HR, 1.98; 95% CI, 1.11 to 3.55; p=0.021) were independent factors of OS. The OS of the patient groups stratified by post-NAT response criteria which combined changes in tumor size and CA19-9 showed significant difference (p < 0.001). Such stratification was comparable to ypTNM staging in discrimination performance (difference of C-index, 0.068; 95% CI, -0.012 to 0.142).Conclusion"Any degree of decrease in tumor size on CT" and CA19-9 normalization or staying normal were independent favorable factors of OS. The combination of the two factors discriminated OS comparably to ypTNM staging.
| S-EPMC10789942 | biostudies-literature
Project description:BackgroundReliable methods are needed to identify patients with early-stage cancer or high-grade precancerous lesions in the pancreas. Analysis of pancreatic juice to detect somatic mutations could represent one such approach. Here we investigated the concordance between mutations found in the primary tumor and pancreatic juice from the same patient.MethodsAmplicon-based targeted deep sequencing was performed on samples from 21 patients with pancreatic ductal adenocarcinoma (PDAC) who had undergone Whipple's operation. Mutation profiles were determined in formalin-fixed sections of the primary tumor and in pancreatic juice sampled from the main pancreatic duct during surgery.ResultsUsing a cut-off of 3% for variant allele frequency, KRAS mutations were detected in 20/21 primary tumors (95%) and in 15/21 (71%) juice samples. When also considering low-frequency variants, KRAS mutations were found in 20/21 juice samples. Most juice samples exhibited multiple KRAS variants not seen in the primary tumor, and only in 11 cases (52%) did the most abundant variant of the juice correspond to the KRAS mutation detected in the tumor. TP53 mutations were found in 16 tumors (76%) and six juice samples (29%). Among the positive juice samples, only one exhibited more than a single TP53 mutation. Detection of both KRAS and TP53 mutations was fully concordant in the primary tumor and juice sample in 7/21 cases (33%).ConclusionsPancreatic juice from PDAC patients is rich in KRAS mutations often not seen in the primary tumor and possibly reflecting precancerous lesions in other regions of the pancreas. The inclusion of TP53 mutation detection and additional markers must therefore be considered for fully exploiting the clinical potential of pancreatic juice samples in early cancer detection.
| S-EPMC6321709 | biostudies-literature
Project description:BackgroundPancreatic ductal carcinoma patients have a really poor prognosis given its difficult early detection and the lack of early symptoms. Digital pathology is routinely used by pathologists to diagnose the disease. However, visually inspecting the tissue is a time-consuming task, which slows down the diagnostic procedure. With the advances occurred in the area of artificial intelligence, specifically with deep learning models, and the growing availability of public histology data, clinical decision support systems are being created. However, the generalization capabilities of these systems are not always tested, nor the integration of publicly available datasets for pancreatic ductal carcinoma detection (PDAC).MethodsIn this work, we explored the performace of two weakly-supervised deep learning models using the two more widely available datasets with pancreatic ductal carcinoma histology images, The Cancer Genome Atlas Project (TCGA) and the Clinical Proteomic Tumor Analysis Consortium (CPTAC). In order to have sufficient training data, the TCGA dataset was integrated with the Genotype-Tissue Expression (GTEx) project dataset, which contains healthy pancreatic samples.ResultsWe showed how the model trained on CPTAC generalizes better than the one trained on the integrated dataset, obtaining an inter-dataset accuracy of 90.62% ± 2.32 and an outer-dataset accuracy of 92.17% when evaluated on TCGA + GTEx. Furthermore, we tested the performance on another dataset formed by tissue micro-arrays, obtaining an accuracy of 98.59%. We showed how the features learned in an integrated dataset do not differentiate between the classes, but between the datasets, noticing that a stronger normalization might be needed when creating clinical decision support systems with datasets obtained from different sources. To mitigate this effect, we proposed to train on the three available datasets, improving the detection performance and generalization capabilities of a model trained only on TCGA + GTEx and achieving a similar performance to the model trained only on CPTAC.ConclusionsThe integration of datasets where both classes are present can mitigate the batch effect present when integrating datasets, improving the classification performance, and accurately detecting PDAC across different datasets.
| S-EPMC10294485 | biostudies-literature
Project description:Purpose: Pancreatic ductal adenocarcinoma (PDAC) is associated with an immunosuppressive milieu that supports immune system evasion and disease progression. Here, we interrogated genetic, stromal, and immunologic features of PDAC to delineate impact on prognosis and means to more effectively employ immunotherapy.Experimental Design: A cohort of 109 PDAC cases annotated for overall survival was utilized as a primary discovery cohort. Gene expression analysis defined immunologic subtypes of PDAC that were confirmed in the Cancer Genome Atlas dataset. Stromal and metabolic characteristics of PDAC cases were evaluated by histologic analysis and immunostaining. Enumeration of lymphocytes, as well as staining for CD8, FOXP3, CD68, CD163, PDL1, and CTLA4 characterized immune infiltrate. Neoantigens were determined by analysis of whole-exome sequencing data. Random-forest clustering was employed to define multimarker subtypes, with univariate and multivariate analyses interrogating prognostic significance.Results: PDAC cases exhibited distinct stromal phenotypes that were associated with prognosis, glycolytic and hypoxic biomarkers, and immune infiltrate composition. Immune infiltrate was diverse among PDAC cases and enrichment for M2 macrophages and select immune checkpoints regulators were specifically associated with survival. Composite analysis with neoantigen burden, immunologic, and stromal features defined novel subtypes of PDAC that could have bearing on sensitivity to immunologic therapy approaches. In addition, a subtype with low levels of neoantigens and minimal lymphocyte infiltrate was associated with improved overall survival.Conclusions: The mutational burden of PDAC is associated with distinct immunosuppressive mechanisms that are conditioned by the tumor stromal environment. The defined subtypes have significance for utilizing immunotherapy in the treatment of PDAC. Clin Cancer Res; 23(15); 4429-40. ©2017 AACR.
| S-EPMC5951386 | biostudies-literature