Pub Date : 2021-07-01DOI: 10.1158/1538-7445.AM2021-LB015
Wei-Lei Yang, Jen-Fan Hang, Chi-Bin Li, Ching-Ming Lee, Yi-Sheng Lin, T. Tsao, M. Chang, Y. Ou, Tien-Jen Liu
BACKGROUND: The Paris System (TPS) for Reporting Urinary Cytology provides standardized diagnostic criteria for urinary tract cytology specimens, focusing on the detection of high-grade urothelial carcinoma (HGUC). Since the publication in 2016, numerous studies have reported a decrease in atypical diagnosis and a significant improvement in the detection of HGUC after adopting TPS. However, the major challenges include labor-intensive screening and interobserver variations. Artificial intelligence (AI) in medical imaging analysis is an emerging tool for ancillary diagnosis. To this end, we have developed an AI algorithm and conducted a retrospective study to evaluate the AI-assisted urine cytology reporting workflow. METHODS: A total of 131 urine cytology slides from bladder cancer patients, either first diagnosis or post-treatment follow-up, were retrieved and digitized as whole slide images (WSIs). A deep learning-based computational model was used to analyze these WSIs. Candidate urothelial cells were automatically highlighted and classified into high-risk and low-risk atypia categories in each sample based on TPS criteria. Slide-wide statistical data, including a total number of high-risk and low-risk cells, nuclear-cytoplasmic ratio (N:C ratio) and nuclear area for each cell, and the distribution and mean values of these variables, were also provided. In a blind study, a cytotechnologist and a cytopathologist parallelly reviewed the AI-annotated images and quantitative data for each WSI sample. Suspicious for HUGC and HGUC were considered to be "positive" and the other diagnostic categories were considered to be "negative" according to whether trigger cystoscopy. The results were compared with the final diagnosis reviewed by a senior cytopathologist via microscopy to evaluate the performance of the AI-assisted model. RESULTS: There were 35 positive and 96 negative urine cytology samples based on the final diagnosis. The AI algorithm annotated a total of 26,502 cells and a mean of 757.2 cells at cancer risk from all positive samples and a total of 950 cells and a mean of 9.9 cells at cancer risk from all negative samples. The mean N:C ratio was 0.68 for high-risk atypical cells and 0.56 for low-risk atypical cells. The performance of the AI-assisted reports of the cytotechnologist was 88.6% sensitivity, 97.9% specificity, 93.9% positive prediction value (PPV), and 95.9% negative prediction value (NPV) and the cytopathologist was 91.4% sensitivity, 95.8% specificity, 88.9% PPV, and 96.8% NPV. CONCLUSIONS: We demonstrated an AI algorithm that can effectively assist the reporting of urine cytology by classifying urothelial cells at cancer risk and calculating quantitative data using WSI analysis. Integrating this AI model into clinical urine cytology workflow supported TPS for reporting urinary cytopathology, reduced the interobserver variations, and may potentially reduce the human labor for screening. Citation Format: Wei-Lei Yang, Jen-Fan Han
{"title":"Abstract LB015: Clinical evaluation of The Paris System-based artificial intelligence algorithm for reporting urinary cytopathology","authors":"Wei-Lei Yang, Jen-Fan Hang, Chi-Bin Li, Ching-Ming Lee, Yi-Sheng Lin, T. Tsao, M. Chang, Y. Ou, Tien-Jen Liu","doi":"10.1158/1538-7445.AM2021-LB015","DOIUrl":"https://doi.org/10.1158/1538-7445.AM2021-LB015","url":null,"abstract":"BACKGROUND: The Paris System (TPS) for Reporting Urinary Cytology provides standardized diagnostic criteria for urinary tract cytology specimens, focusing on the detection of high-grade urothelial carcinoma (HGUC). Since the publication in 2016, numerous studies have reported a decrease in atypical diagnosis and a significant improvement in the detection of HGUC after adopting TPS. However, the major challenges include labor-intensive screening and interobserver variations. Artificial intelligence (AI) in medical imaging analysis is an emerging tool for ancillary diagnosis. To this end, we have developed an AI algorithm and conducted a retrospective study to evaluate the AI-assisted urine cytology reporting workflow. METHODS: A total of 131 urine cytology slides from bladder cancer patients, either first diagnosis or post-treatment follow-up, were retrieved and digitized as whole slide images (WSIs). A deep learning-based computational model was used to analyze these WSIs. Candidate urothelial cells were automatically highlighted and classified into high-risk and low-risk atypia categories in each sample based on TPS criteria. Slide-wide statistical data, including a total number of high-risk and low-risk cells, nuclear-cytoplasmic ratio (N:C ratio) and nuclear area for each cell, and the distribution and mean values of these variables, were also provided. In a blind study, a cytotechnologist and a cytopathologist parallelly reviewed the AI-annotated images and quantitative data for each WSI sample. Suspicious for HUGC and HGUC were considered to be \"positive\" and the other diagnostic categories were considered to be \"negative\" according to whether trigger cystoscopy. The results were compared with the final diagnosis reviewed by a senior cytopathologist via microscopy to evaluate the performance of the AI-assisted model. RESULTS: There were 35 positive and 96 negative urine cytology samples based on the final diagnosis. The AI algorithm annotated a total of 26,502 cells and a mean of 757.2 cells at cancer risk from all positive samples and a total of 950 cells and a mean of 9.9 cells at cancer risk from all negative samples. The mean N:C ratio was 0.68 for high-risk atypical cells and 0.56 for low-risk atypical cells. The performance of the AI-assisted reports of the cytotechnologist was 88.6% sensitivity, 97.9% specificity, 93.9% positive prediction value (PPV), and 95.9% negative prediction value (NPV) and the cytopathologist was 91.4% sensitivity, 95.8% specificity, 88.9% PPV, and 96.8% NPV. CONCLUSIONS: We demonstrated an AI algorithm that can effectively assist the reporting of urine cytology by classifying urothelial cells at cancer risk and calculating quantitative data using WSI analysis. Integrating this AI model into clinical urine cytology workflow supported TPS for reporting urinary cytopathology, reduced the interobserver variations, and may potentially reduce the human labor for screening. Citation Format: Wei-Lei Yang, Jen-Fan Han","PeriodicalId":73617,"journal":{"name":"Journal of bioinformatics and systems biology : Open access","volume":"39 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2021-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"85076691","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-07-01DOI: 10.1158/1538-7445.AM2021-258
Tomohiro Sera, M. Yashiro, Gen Tsujio, Yurie Yamamoto, Atsushi Sugimoto, S. Kushiyama, Sadaaki Nishimura, M. Ohira
{"title":"Abstract 258: Identification of characteristic genes of scirrhous-type gastric cancer cells by RNAseq","authors":"Tomohiro Sera, M. Yashiro, Gen Tsujio, Yurie Yamamoto, Atsushi Sugimoto, S. Kushiyama, Sadaaki Nishimura, M. Ohira","doi":"10.1158/1538-7445.AM2021-258","DOIUrl":"https://doi.org/10.1158/1538-7445.AM2021-258","url":null,"abstract":"","PeriodicalId":73617,"journal":{"name":"Journal of bioinformatics and systems biology : Open access","volume":"82 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2021-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"90576605","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-07-01DOI: 10.1158/1538-7445.AM2021-180
Matthew Lee, N. Tang, M. Ahluwalia, E. Fonkem, K. Fink, Harshil Dhurv, M. Berens, S. Peng
Glioblastoma is characterized by intra- and inter-tumoral heterogeneity. An umbrella trial tests multiple investigational treatment arms depending on corresponding biomarker signatures. A contingency of an efficient umbrella trial is a suite of preferably orthogonal molecular biomarkers to classify patients into the likely-most-beneficial arm. Assigning thresholds of molecular signatures to classify a patient as a “most-likely responder” for one specific treatment arm is a crucial task. Gene Set Variation Analysis (GSVA) of specimens from a GBM clinical trial of methoxyamine associated differential enrichment in DNA repair pathways activities with patient response. However, the large number of DNA-repair related pathways confound confident “high” enrichment of responder, as well as obscuring to what degree each feature contributes to the likelihood of a patient9s response. Here, we utilized semi-supervised machine learning, Entropy-Regularized Logistic Regression (ERLR) to predict vulnerability classification. By first training all available data using semi-supervised algorithms we transformed unclassified TCGA GBM samples with highest certainty of predicted response into a self-labeled dataset. In this case, we developed a predictive model which has a larger sample size and potential better performance. Our umbrella trial design currently includes three treatment arms for GBM patients: arsenic trioxide, methoxyamine, and pevonedistat. Each treatment arm manifests its own signature developed by the above (or similar) machine learning pipeline based on selected gene mutation status and whole transcriptome data. In order to increase the robustness and scalability (with future more treatment arms), we also developed a multi-label classification ensemble model that9s capable of predicting a probability of “fitness” of each novel therapeutic agent for each patient. By expansion to three, independent treatment arms within a single umbrella trial, a “mock” stratification of TCGA GBM patients labeled 56% of all cases into at least one “high likelihood of response” arm. Predicted vulnerability using genomic data from preclinical PDX models placed 4 out of 6 models into a “high likelihood of response” regimen. Our utilization of multiple vulnerability signatures in an umbrella trial demonstrates how a precision medicine model can support an efficient clinical trial for heterogeneous diseases such as GBM. Citation Format: Matthew Eric Lee, Nanyun Tang, Manmeet Ahluwalia, Ekokobe Fonkem, Karen Fink, Harshil Dhurv, Harshil Dhurv, Michael E. Berens, Sen Peng. Identifying signatures of vulnerability through machine learning in an umbrella trial for glioblastoma [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 180.
胶质母细胞瘤的特点是肿瘤内部和肿瘤间的异质性。伞式试验根据相应的生物标志物特征测试多个研究性治疗组。有效的保护伞试验的偶然性是一套优选的正交分子生物标志物，将患者分为可能最有益的组。分配分子特征阈值来将患者分类为某一特定治疗组的“最有可能应答者”是一项至关重要的任务。基因集变异分析(GSVA)来自GBM临床试验的标本甲氧基胺相关的DNA修复途径活性差异富集与患者反应。然而，大量的dna修复相关途径混淆了应答者的“高”富集，也模糊了每个特征在多大程度上有助于患者应答的可能性。在这里，我们利用半监督机器学习，熵-正则化逻辑回归(ERLR)来预测漏洞分类。通过首先使用半监督算法训练所有可用数据，我们将具有最高预测响应确定性的未分类TCGA GBM样本转换为自标记数据集。在这种情况下，我们开发了一个具有更大样本量和更好性能的预测模型。我们的伞式试验设计目前包括GBM患者的三个治疗组:三氧化二砷、甲氧基胺和佩伏奈远。每个治疗组都有自己的特征，这些特征是由上述(或类似的)机器学习管道基于选定的基因突变状态和整个转录组数据开发的。为了增加鲁棒性和可扩展性(未来会有更多的治疗组)，我们还开发了一个多标签分类集成模型，该模型能够预测每种新型治疗剂对每个患者的“适合度”概率。通过扩大到三个独立的治疗组，在一个单一的伞式试验中，TCGA GBM患者的“模拟”分层将56%的病例标记为至少一个“高可能反应”组。利用临床前PDX模型的基因组数据预测的脆弱性将6个模型中的4个置于“高可能反应”方案中。我们在一项总体性试验中利用多个脆弱性特征，展示了精准医学模型如何支持针对异质疾病(如GBM)的有效临床试验。引用格式:Matthew Eric Lee, Nanyun Tang, Manmeet Ahluwalia, Ekokobe Fonkem, Karen Fink, Harshil Dhurv, Harshil Dhurv, Michael E. Berens, Sen Peng在胶质母细胞瘤的总括性试验中，通过机器学习识别脆弱性特征[摘要]。见:美国癌症研究协会2021年年会论文集;2021年4月10日至15日和5月17日至21日。费城(PA): AACR;癌症杂志，2021;81(13 -增刊):摘要第180期。
{"title":"Abstract 180: Identifying signatures of vulnerability through machine learning in an umbrella trial for glioblastoma","authors":"Matthew Lee, N. Tang, M. Ahluwalia, E. Fonkem, K. Fink, Harshil Dhurv, M. Berens, S. Peng","doi":"10.1158/1538-7445.AM2021-180","DOIUrl":"https://doi.org/10.1158/1538-7445.AM2021-180","url":null,"abstract":"Glioblastoma is characterized by intra- and inter-tumoral heterogeneity. An umbrella trial tests multiple investigational treatment arms depending on corresponding biomarker signatures. A contingency of an efficient umbrella trial is a suite of preferably orthogonal molecular biomarkers to classify patients into the likely-most-beneficial arm. Assigning thresholds of molecular signatures to classify a patient as a “most-likely responder” for one specific treatment arm is a crucial task. Gene Set Variation Analysis (GSVA) of specimens from a GBM clinical trial of methoxyamine associated differential enrichment in DNA repair pathways activities with patient response. However, the large number of DNA-repair related pathways confound confident “high” enrichment of responder, as well as obscuring to what degree each feature contributes to the likelihood of a patient9s response. Here, we utilized semi-supervised machine learning, Entropy-Regularized Logistic Regression (ERLR) to predict vulnerability classification. By first training all available data using semi-supervised algorithms we transformed unclassified TCGA GBM samples with highest certainty of predicted response into a self-labeled dataset. In this case, we developed a predictive model which has a larger sample size and potential better performance. Our umbrella trial design currently includes three treatment arms for GBM patients: arsenic trioxide, methoxyamine, and pevonedistat. Each treatment arm manifests its own signature developed by the above (or similar) machine learning pipeline based on selected gene mutation status and whole transcriptome data. In order to increase the robustness and scalability (with future more treatment arms), we also developed a multi-label classification ensemble model that9s capable of predicting a probability of “fitness” of each novel therapeutic agent for each patient. By expansion to three, independent treatment arms within a single umbrella trial, a “mock” stratification of TCGA GBM patients labeled 56% of all cases into at least one “high likelihood of response” arm. Predicted vulnerability using genomic data from preclinical PDX models placed 4 out of 6 models into a “high likelihood of response” regimen. Our utilization of multiple vulnerability signatures in an umbrella trial demonstrates how a precision medicine model can support an efficient clinical trial for heterogeneous diseases such as GBM. Citation Format: Matthew Eric Lee, Nanyun Tang, Manmeet Ahluwalia, Ekokobe Fonkem, Karen Fink, Harshil Dhurv, Harshil Dhurv, Michael E. Berens, Sen Peng. Identifying signatures of vulnerability through machine learning in an umbrella trial for glioblastoma [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 180.","PeriodicalId":73617,"journal":{"name":"Journal of bioinformatics and systems biology : Open access","volume":"82 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2021-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"79612666","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-07-01DOI: 10.1158/1538-7445.AM2021-156
Felipe Batalini, D. Gulhan, V. Mao, Madeline Polak, E. Winer, E. Mayer, U. Matulonis, P. Konstantinopoulos, P. J. Park, G. Wulf
{"title":"Abstract 156: Mutational signature 3 predicts responses to olaparib plus buparlisib in triple-negative breast cancer and high-grade serous ovarian cancer","authors":"Felipe Batalini, D. Gulhan, V. Mao, Madeline Polak, E. Winer, E. Mayer, U. Matulonis, P. Konstantinopoulos, P. J. Park, G. Wulf","doi":"10.1158/1538-7445.AM2021-156","DOIUrl":"https://doi.org/10.1158/1538-7445.AM2021-156","url":null,"abstract":"","PeriodicalId":73617,"journal":{"name":"Journal of bioinformatics and systems biology : Open access","volume":"507 ","pages":""},"PeriodicalIF":0.0,"publicationDate":"2021-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"72435738","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-07-01DOI: 10.1158/1538-7445.AM2021-159
P. Chati, E. Storrs, A. Usmani, B. Krasnick, C. Wetzel, T. Hollander, Faridi Quium, I. Sloan, H. Anthony, Badiyan Shahed, G. Lang, N. Cosgrove, V. Kushnir, D. Early, W. Hawkins, L. Ding, R. Fields, K. Das, A. Chaudhuri
INTRODUCTION: Pancreatic ductal adenocarcinoma (PDAC) is an aggressive cancer type with a poor prognosis. Patients with the classical histologic subtype typically have a better prognosis than those with a squamous-like histology. Still, survival outcomes vary significantly, even in early-stage patients, making it challenging to personalize treatment via subtyping. Here, we utilize CytoTRACE to better classify PDAC based on tumor cell-intrinsic developmental states, to more accurately prognosticate patients from the time of initial surgical resection. METHODS: We performed core needle pancreatic biopsies in 13 patients and surgical PDAC resections in five patients, and analyzed the resulting single-cell RNA sequencing (scRNA-seq) data to identify tumor cell clusters. We then applied CytoTRACE for developmental state analysis. Following developmental state quantification, we classified PDAC tumor cells into 3 distinct subtypes: squamous-like, classical early developmental (ED), and classical late developmental (LD). We developed a gene signature for each subtype, which we then applied to two external bulk RNA-seq datasets - 1) The Cancer Genome Atlas (TCGA): 125 early-stage PDAC tumors, and 2) Bailey et al (Nature 2016): 86 predominantly early-stage PDAC tumors. RESULTS: scRNA-seq data was partitioned into two subtypes, classical and squamous-like, based on marker gene expression. The classical subtype was further partitioned into ED versus LD cell states using the developmental index from CytoTRACE. For the squamous-like group, we identified the top 20 differentially expressed genes (squamous-like gene signature). For the ED and LD subtypes, we identified the top 20 genes correlating with the CytoTRACE developmental index (ED gene signature). Using a multivariate cox proportional hazards regression, we showed that the squamous-like signature was associated with significantly worse overall survival in TCGA (HR = 6.8, P = .01). Strikingly, our newly derived ED cell state signature was also associated with inferior overall survival in TCGA (HR = 5.9, P = .02). Kaplan-Meier analysis using optimized cutpoints between squamous-like and classical subtype scores, and between ED and LD cell state scores, again showed that patients with predominantly squamous-like tumors had significantly worse survival (HR = 4.4, P = .04); and that predominantly classical tumors enriched for the ED cell state had significantly inferior overall survival compared to LD (median 15.0 vs. 22.0 months, HR = 4.6, P = .03). The same trends were observed in the less-powered Bailey et al cohort. CONCLUSION: We showed that three developmental cell states, learned through the analysis of PDAC scRNA-seq data, can prognosticate patients with bulk RNA-seq expression data. This could help facilitate more personalized risk-adapted approaches for PDAC in the future. Citation Format: Prathamesh Mandar Chati, Erik Storrs, Abul Usmani, Bradley Krasnick, Chris Wetzel, Thomas Hollander, Faridi Q
简介:胰腺导管腺癌(PDAC)是一种预后不良的侵袭性癌症。典型组织学亚型的患者通常比鳞状样组织学的患者预后更好。尽管如此，即使在早期患者中，生存结果也存在显着差异，这使得通过分型进行个性化治疗具有挑战性。在这里，我们利用CytoTRACE根据肿瘤细胞内在发育状态更好地对PDAC进行分类，从而更准确地预测患者从最初手术切除时开始的预后。方法:我们对13例患者进行了核心针胰腺活检，对5例患者进行了PDAC手术切除，并分析了由此产生的单细胞RNA测序(scRNA-seq)数据，以鉴定肿瘤细胞簇。然后应用CytoTRACE进行发育状态分析。根据发育状态量化，我们将PDAC肿瘤细胞分为3个不同的亚型:鳞状样，经典早期发育(ED)和经典晚期发育(LD)。我们为每个亚型开发了一个基因标记，然后将其应用于两个外部批量RNA-seq数据集- 1)癌症基因组图谱(TCGA): 125个早期PDAC肿瘤，以及2)Bailey等人(Nature 2016): 86个主要是早期PDAC肿瘤。结果:基于标记基因的表达，scRNA-seq数据被分为经典型和鳞状型两种亚型。利用CytoTRACE的发育指数进一步将经典亚型划分为ED和LD细胞状态。对于鳞状样组，我们确定了前20个差异表达基因(鳞状样基因特征)。对于ED和LD亚型，我们确定了与CytoTRACE发育指数(ED基因标记)相关的前20个基因。通过多变量cox比例风险回归，我们发现鳞状样特征与TCGA患者的总生存率显著降低相关(HR = 6.8, P = 0.01)。引人注目的是，我们新获得的ED细胞状态特征也与TCGA患者较低的总生存率相关(HR = 5.9, P = 0.02)。Kaplan-Meier分析采用优化的切点在鳞状样和经典亚型评分之间、ED和LD细胞状态评分之间进行分析，再次显示以鳞状样肿瘤为主的患者生存率明显较差(HR = 4.4, P = 0.04);与LD相比，以ED细胞状态富集为主的典型肿瘤的总生存期明显较低(中位15.0个月vs. 22.0个月，HR = 4.6, P = 0.03)。同样的趋势在较弱的Bailey等人的队列中也观察到了。结论:我们发现通过PDAC scRNA-seq数据分析了解到的三种发育细胞状态可以通过大量RNA-seq表达数据预测患者的预后。这有助于在未来为PDAC提供更个性化的风险适应方法。引文格式:Prathamesh Mandar Chati, Erik Storrs, Abul Usmani, Bradley Krasnick, Chris Wetzel, Thomas Hollander, Faridi Quium, Ian Sloan, Hephzibah Anthony, Badiyan Shahed, Gabriel D. Lang, Natalie D. Cosgrove, Vladimir M. Kushnir, Dayna S. Early, William G. Hawkins, Li Ding, Ryan C. Fields, Koushik K. Das, Aadel A. Chaudhuri。单细胞RNA测序鉴定的胰腺导管腺癌发育细胞状态特征在应用于大量RNA-seq数据时具有预后作用[摘要]。见:美国癌症研究协会2021年年会论文集;2021年4月10日至15日和5月17日至21日。费城(PA): AACR;癌症杂志，2021;81(13 -增刊):摘要第159期。
{"title":"Abstract 159: Pancreatic ductal adenocarcinoma developmental cell state signatures identified by single cell RNA sequencing are prognostic when applied to bulk RNA-seq data","authors":"P. Chati, E. Storrs, A. Usmani, B. Krasnick, C. Wetzel, T. Hollander, Faridi Quium, I. Sloan, H. Anthony, Badiyan Shahed, G. Lang, N. Cosgrove, V. Kushnir, D. Early, W. Hawkins, L. Ding, R. Fields, K. Das, A. Chaudhuri","doi":"10.1158/1538-7445.AM2021-159","DOIUrl":"https://doi.org/10.1158/1538-7445.AM2021-159","url":null,"abstract":"INTRODUCTION: Pancreatic ductal adenocarcinoma (PDAC) is an aggressive cancer type with a poor prognosis. Patients with the classical histologic subtype typically have a better prognosis than those with a squamous-like histology. Still, survival outcomes vary significantly, even in early-stage patients, making it challenging to personalize treatment via subtyping. Here, we utilize CytoTRACE to better classify PDAC based on tumor cell-intrinsic developmental states, to more accurately prognosticate patients from the time of initial surgical resection. METHODS: We performed core needle pancreatic biopsies in 13 patients and surgical PDAC resections in five patients, and analyzed the resulting single-cell RNA sequencing (scRNA-seq) data to identify tumor cell clusters. We then applied CytoTRACE for developmental state analysis. Following developmental state quantification, we classified PDAC tumor cells into 3 distinct subtypes: squamous-like, classical early developmental (ED), and classical late developmental (LD). We developed a gene signature for each subtype, which we then applied to two external bulk RNA-seq datasets - 1) The Cancer Genome Atlas (TCGA): 125 early-stage PDAC tumors, and 2) Bailey et al (Nature 2016): 86 predominantly early-stage PDAC tumors. RESULTS: scRNA-seq data was partitioned into two subtypes, classical and squamous-like, based on marker gene expression. The classical subtype was further partitioned into ED versus LD cell states using the developmental index from CytoTRACE. For the squamous-like group, we identified the top 20 differentially expressed genes (squamous-like gene signature). For the ED and LD subtypes, we identified the top 20 genes correlating with the CytoTRACE developmental index (ED gene signature). Using a multivariate cox proportional hazards regression, we showed that the squamous-like signature was associated with significantly worse overall survival in TCGA (HR = 6.8, P = .01). Strikingly, our newly derived ED cell state signature was also associated with inferior overall survival in TCGA (HR = 5.9, P = .02). Kaplan-Meier analysis using optimized cutpoints between squamous-like and classical subtype scores, and between ED and LD cell state scores, again showed that patients with predominantly squamous-like tumors had significantly worse survival (HR = 4.4, P = .04); and that predominantly classical tumors enriched for the ED cell state had significantly inferior overall survival compared to LD (median 15.0 vs. 22.0 months, HR = 4.6, P = .03). The same trends were observed in the less-powered Bailey et al cohort. CONCLUSION: We showed that three developmental cell states, learned through the analysis of PDAC scRNA-seq data, can prognosticate patients with bulk RNA-seq expression data. This could help facilitate more personalized risk-adapted approaches for PDAC in the future. Citation Format: Prathamesh Mandar Chati, Erik Storrs, Abul Usmani, Bradley Krasnick, Chris Wetzel, Thomas Hollander, Faridi Q","PeriodicalId":73617,"journal":{"name":"Journal of bioinformatics and systems biology : Open access","volume":"7 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2021-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"83191891","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Abstract 255: TransVAF: A transfer learning approach for recognize genomic mutations with various tumor purity and clonal proportions","authors":"Tian Zheng, Jiayin Wang, Xiao-e Xiao, Xiaoyan Zhu, Xuanping Zhang, Xin Lai, Yanfang Guan, X. Yi","doi":"10.1158/1538-7445.AM2021-255","DOIUrl":"https://doi.org/10.1158/1538-7445.AM2021-255","url":null,"abstract":"","PeriodicalId":73617,"journal":{"name":"Journal of bioinformatics and systems biology : Open access","volume":"21 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2021-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"73277649","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-07-01DOI: 10.1158/1538-7445.AM2021-176
D. Thompson, O. Vaske, A. Rao, Holly C. Beale
Epitopes are peptides that present on the surface of the cell and can be recognized by immune cells to initiate the immune response. Identification of neoepitopes – tumor-specific, MHC-bound epitopes recognized specifically by T-cells – is valuable for predicting response to immunotherapies, including checkpoint blockade therapies. Tumors with more neoepitopes tend to be more responsive to immune checkpoint therapies compared to tumors with fewer neoepitopes. ProTECT is a previously published computational method that uses Illumina whole genome and transcriptome sequencing data from tumor and matched normal tissues to identify neoepitopes. Tumor and normal whole genome sequencing data are used to infer a patient9s HLA haplotypes, as well as annotate variants as either somatic or germline. While whole genome sequencing is comprehensive, it is quite costly and not available for many samples. Here we adapt ProTECT to use only tumor RNA sequencing data and HLA haplotype information available to the clinician to identify neoepitopes in a tumor sample. Prior to running ProTECT, we use the computational tools Opossum and Platypus for variant calling instead of Radia (which is designed for variant calling using both RNA and DNA sequencing data as input). To determine which variants are somatic and therefore could represent tumor neoepitopes, variants found in RNA are compared to a panel of normals, for example the Genome Aggregation Database (gnomAD; containing variants from 125,748 exome sequences and 15,708 whole-genome sequences). With the resulting somatic variants and the HLA type, ProTECT proceeds as usual, with translation of variants into proteins, MHC:Peptide binding predictions and neoepitope ranking. We find that high quality neoepitopes are identifiable using an RNA-only approach, when genomic data is absent. Future work will validate the sensitivity of our method by benchmarking it against the original ProTECT predictions in the TCGA Prostate Adenocarcinoma cohort. Citation Format: Drew Thompson, Olena M. Vaske, Arjun Rao, Holly C. Beale. Detecting neoepitopes from tumor RNA sequencing datasets [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 176.
表位是存在于细胞表面的多肽，可以被免疫细胞识别并启动免疫反应。鉴定新表位-肿瘤特异性，由t细胞特异性识别的mhc结合表位-对于预测免疫疗法的反应是有价值的，包括检查点阻断疗法。与新表位较少的肿瘤相比，具有更多新表位的肿瘤往往对免疫检查点治疗更有反应。ProTECT是一种先前发表的计算方法，它使用来自肿瘤和匹配正常组织的Illumina全基因组和转录组测序数据来识别新表位。肿瘤和正常全基因组测序数据用于推断患者的HLA单倍型，以及注释体细胞或种系变异。虽然全基因组测序是全面的，但它非常昂贵，而且许多样本无法获得。在这里，我们使ProTECT仅使用肿瘤RNA测序数据和临床医生可用的HLA单倍型信息来识别肿瘤样本中的新表位。在运行ProTECT之前，我们使用计算工具possum和鸭嘴兽来调用变体，而不是Radia (Radia是设计用于使用RNA和DNA测序数据作为输入的变体调用)。为了确定哪些变异是体细胞的，因此可能代表肿瘤新表位，将RNA中发现的变异与一组正常的变异进行比较，例如基因组聚集数据库(gnomAD;包含125,748个外显子组序列和15,708个全基因组序列的变体)。根据产生的体细胞变异和HLA类型，ProTECT照常进行，将变异翻译成蛋白质，MHC:肽结合预测和新表位排序。我们发现，在基因组数据缺失的情况下，高质量的新表位可以通过仅使用rna的方法进行识别。未来的工作将通过对照TCGA前列腺腺癌队列中原始的ProTECT预测来验证我们方法的敏感性。引用格式:Drew Thompson, Olena M. Vaske, Arjun Rao, Holly C. Beale。从肿瘤RNA测序数据集中检测新表位[摘要]。见:美国癌症研究协会2021年年会论文集;2021年4月10日至15日和5月17日至21日。费城(PA): AACR;癌症杂志，2021;81(13 -增刊):摘要nr 176。
{"title":"Abstract 176: Detecting neoepitopes from tumor RNA sequencing datasets","authors":"D. Thompson, O. Vaske, A. Rao, Holly C. Beale","doi":"10.1158/1538-7445.AM2021-176","DOIUrl":"https://doi.org/10.1158/1538-7445.AM2021-176","url":null,"abstract":"Epitopes are peptides that present on the surface of the cell and can be recognized by immune cells to initiate the immune response. Identification of neoepitopes – tumor-specific, MHC-bound epitopes recognized specifically by T-cells – is valuable for predicting response to immunotherapies, including checkpoint blockade therapies. Tumors with more neoepitopes tend to be more responsive to immune checkpoint therapies compared to tumors with fewer neoepitopes. ProTECT is a previously published computational method that uses Illumina whole genome and transcriptome sequencing data from tumor and matched normal tissues to identify neoepitopes. Tumor and normal whole genome sequencing data are used to infer a patient9s HLA haplotypes, as well as annotate variants as either somatic or germline. While whole genome sequencing is comprehensive, it is quite costly and not available for many samples. Here we adapt ProTECT to use only tumor RNA sequencing data and HLA haplotype information available to the clinician to identify neoepitopes in a tumor sample. Prior to running ProTECT, we use the computational tools Opossum and Platypus for variant calling instead of Radia (which is designed for variant calling using both RNA and DNA sequencing data as input). To determine which variants are somatic and therefore could represent tumor neoepitopes, variants found in RNA are compared to a panel of normals, for example the Genome Aggregation Database (gnomAD; containing variants from 125,748 exome sequences and 15,708 whole-genome sequences). With the resulting somatic variants and the HLA type, ProTECT proceeds as usual, with translation of variants into proteins, MHC:Peptide binding predictions and neoepitope ranking. We find that high quality neoepitopes are identifiable using an RNA-only approach, when genomic data is absent. Future work will validate the sensitivity of our method by benchmarking it against the original ProTECT predictions in the TCGA Prostate Adenocarcinoma cohort. Citation Format: Drew Thompson, Olena M. Vaske, Arjun Rao, Holly C. Beale. Detecting neoepitopes from tumor RNA sequencing datasets [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 176.","PeriodicalId":73617,"journal":{"name":"Journal of bioinformatics and systems biology : Open access","volume":"23 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2021-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"88737231","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-07-01DOI: 10.1158/1538-7445.AM2021-231
Kaitlyn E. Johnson, M. Ciocanel, Josua Aponte, N. Bajeux, Fanwang Meng, D. Bottino
{"title":"Abstract 231: Evaluation of pharmacologic mechanisms to overcome IgG1 antibody (Ab) resistance via quantitative systems pharmacology (QSP) modeling of antibody-dependent cell mediated cytotoxicity (ADCC)","authors":"Kaitlyn E. Johnson, M. Ciocanel, Josua Aponte, N. Bajeux, Fanwang Meng, D. Bottino","doi":"10.1158/1538-7445.AM2021-231","DOIUrl":"https://doi.org/10.1158/1538-7445.AM2021-231","url":null,"abstract":"","PeriodicalId":73617,"journal":{"name":"Journal of bioinformatics and systems biology : Open access","volume":"11 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2021-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"75332612","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-07-01DOI: 10.1158/1538-7445.AM2021-247
L. Piccotti, L. Mirandola, M. Chiriva-Internati
The advancement of cures for cancer needs the development of novel, more efficacious, and more specific immunotherapeutic approaches through the discovery of novel target candidates displaying differential expression between healthy and malignant tissues. CancerDiff is a proprietary software module for the identification of potential new immunotherapeutic cancer targets that originate from differentially expressed, alternatively spliced transcripts. When utilized to analyze Ovarian Cancer (OV) datasets, CancerDiff identified a selectively upregulated mesothelin (MSLN) splice variant translated into a protein isoform (IsoMSLN) bearing a distinct unique peptide absent in the canonical protein sequence. To validate this prediction and to confirm the upregulation of IsoMSLN in OV, datasets from publicly available proteomic repositories were searched for its unique signature peptide. In agreement with CancerDiff prediction, IsoMSLN peptide was detected in 71% of OV samples and 61% of adjacent normal tissues. Molecular modeling tools predicted this peptide to be part of the extracellular portion of the protein in an antibody accessible region. These results indicate IsoMSLN unique peptide as a suitable target for immunotherapy for OV cancer. Citation Format: Lucia Piccotti, Leonardo Mirandola, Maurizio Chiriva-Internati. Identification of an ovarian cancer selective splice variant of mesothelin utilizing the Kiromic proprietary search engine CancerDiff [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 247.
{"title":"Abstract 247: Identification of an ovarian cancer selective splice variant of mesothelin utilizing the Kiromic proprietary search engine CancerDiff","authors":"L. Piccotti, L. Mirandola, M. Chiriva-Internati","doi":"10.1158/1538-7445.AM2021-247","DOIUrl":"https://doi.org/10.1158/1538-7445.AM2021-247","url":null,"abstract":"The advancement of cures for cancer needs the development of novel, more efficacious, and more specific immunotherapeutic approaches through the discovery of novel target candidates displaying differential expression between healthy and malignant tissues. CancerDiff is a proprietary software module for the identification of potential new immunotherapeutic cancer targets that originate from differentially expressed, alternatively spliced transcripts. When utilized to analyze Ovarian Cancer (OV) datasets, CancerDiff identified a selectively upregulated mesothelin (MSLN) splice variant translated into a protein isoform (IsoMSLN) bearing a distinct unique peptide absent in the canonical protein sequence. To validate this prediction and to confirm the upregulation of IsoMSLN in OV, datasets from publicly available proteomic repositories were searched for its unique signature peptide. In agreement with CancerDiff prediction, IsoMSLN peptide was detected in 71% of OV samples and 61% of adjacent normal tissues. Molecular modeling tools predicted this peptide to be part of the extracellular portion of the protein in an antibody accessible region. These results indicate IsoMSLN unique peptide as a suitable target for immunotherapy for OV cancer. Citation Format: Lucia Piccotti, Leonardo Mirandola, Maurizio Chiriva-Internati. Identification of an ovarian cancer selective splice variant of mesothelin utilizing the Kiromic proprietary search engine CancerDiff [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 247.","PeriodicalId":73617,"journal":{"name":"Journal of bioinformatics and systems biology : Open access","volume":"10 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2021-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"74091036","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-07-01DOI: 10.1158/1538-7445.AM2021-244
Weiwei Bian, F. Kebede, Zongli Zheng
{"title":"Abstract 244: A computation method for noise reduction based on ultra-deep targeted sequencing data","authors":"Weiwei Bian, F. Kebede, Zongli Zheng","doi":"10.1158/1538-7445.AM2021-244","DOIUrl":"https://doi.org/10.1158/1538-7445.AM2021-244","url":null,"abstract":"","PeriodicalId":73617,"journal":{"name":"Journal of bioinformatics and systems biology : Open access","volume":"256 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2021-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"74510641","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：