Lung Cancer Diagnosis on Virtual Histologically Stained Tissue Using Weakly Supervised Learning.

Lung adenocarcinoma (LUAD) is the most common primary lung cancer and accounts for 40% of all lung cancer cases. The current gold standard for lung cancer analysis is based on the pathologists' interpretation of hematoxylin and eosin (H&E)-stained tissue slices viewed under a brightfield microscope or a digital slide scanner. Computational pathology using deep learning has been proposed to detect lung cancer on histology images. However, the histological staining workflow to acquire the H&E-stained images and the subsequent cancer diagnosis procedures are labor-intensive and time-consuming with tedious sample preparation steps and repetitive manual interpretation, respectively. In this work, we propose a weakly supervised learning method for LUAD classification on label-free tissue slices with virtual histological staining. The autofluorescence images of label-free tissue with histopathological information can be converted into virtual H&E-stained images by a weakly supervised deep generative model. For the downstream LUAD classification task, we trained the attention-based multiple-instance learning model with different settings on the open-source LUAD H&E-stained whole-slide images (WSIs) dataset from the Cancer Genome Atlas (TCGA). The model was validated on the 150 H&E-stained WSIs collected from patients in Queen Mary Hospital and Prince of Wales Hospital with an average area under the curve (AUC) of 0.961. The model also achieved an average AUC of 0.973 on 58 virtual H&E-stained WSIs, comparable to the results on 58 standard H&E-stained WSIs with an average AUC of 0.977. The attention heatmaps of virtual H&E-stained WSIs and ground-truth H&E-stained WSIs can indicate tumor regions of LUAD tissue slices. In conclusion, the proposed diagnostic workflow on virtual H&E-stained WSIs of label-free tissue is a rapid, cost effective, and interpretable approach to assist clinicians in postoperative pathological examinations. The method could serve as a blueprint for other label-free imaging modalities and disease contexts.

Function and mechanism of MCM8 in the development and progression of colorectal cancer.

Colorectal cancer (CRC) has become a global health problem which has almost highest morbidity and mortality in all types of cancers. This study aimed to uncover the biological functions and underlying mechanism of MCM8 in the development and progression of CRC. The expression level of MCM8 was found to be upregulated in CRC tissues and significantly associated with tumor grade and patients' survival. Knocking down MCM8 expression in CRC cells could restrain cell growth and cell motility while promoting cell apoptosis in vitro, as well as inhibit tumor growth in xenograft mice model. Based on the RNA screening performing on CRC cells with or without MCM8 knockdown and the following IPA analysis, CHSY1 was identified as a potential target of MCM8 in CRC, whose expression was also found to be higher in tumor tissues than in normal tissues. Moreover, it was demonstrated that MCM8 may regulate the expression of CHSY1 through affecting its NEDD4-mediated ubiquitination, both of which synergistically execute tumor promotion effects on CRC. In conclusion, the outcomes of our study showed the first evidence that MCM8 act as a tumor promotor in CRC, and may be a promising therapeutic target of CRC treatment.

Noninvasive predictive models based on lifestyle analysis and risk factors for early-onset colorectal cancer.

BACKGROUND: Colorectal cancer (CRC) incidence has increased among patients aged <50 years. Exploring high-risk factors and screening high-risk populations may help lower early-onset CRC (EO-CRC) incidence. We developed noninvasive predictive models for EO-CRC and investigated its risk factors. METHODS: This retrospective multicenter study collected information on 1756 patients (811 patients with EO-CRC and 945 healthy controls) from two medical centers in China. Sociodemographic features, clinical symptoms, medical and family history, lifestyle, and dietary factors were measured. Patients from one cohort were randomly assigned (8:2) to two groups for model establishment and internal validation, and another independent cohort was used for external validation. Multivariable logistic regression, random forest, and eXtreme Gradient Boosting (XGBoost) were performed to establish noninvasive predictive models for EO-CRC. Some variables in the model influenced EO-CRC occurrence and were further analyzed. Multivariable logistic regression analysis yielded adjusted odd ratios (ORs) and 95% confidence intervals (CIs). RESULTS: All three models showed good performance, with areas under the receiver operator characteristic curves (AUCs) of 0.82, 0.84, and 0.82 in the internal and 0.78, 0.79, and 0.78 in the external validation cohorts, respectively. Consumption of sweet (OR 2.70, 95% CI 1.89-3.86, P < 0.001) and fried (OR 2.16, 95% CI 1.29-3.62, P < 0.001) foods ≥3 times per week was significantly associated with EO-CRC occurrence. CONCLUSION: We established noninvasive predictive models for EO-CRC and identified multiple nongenetic risk factors, especially sweet and fried foods. The model has good performance and can help predict the occurrence of EO-CRC in the Chinese population.

Bulk and single-cell transcriptome profiling reveal necroptosis-based molecular classification, tumor microenvironment infiltration characterization, and prognosis prediction in colorectal cancer.

BACKGROUND: Necroptosis is a new form of programmed cell death that is associated with cancer initiation, progression, immunity, and chemoresistance. However, the roles of necroptosis-related genes (NRGs) in colorectal cancer (CRC) have not been explored comprehensively. METHODS: In this study, we obtained NRGs and performed consensus molecular subtyping by "ConsensusClusterPlus" to determine necroptosis-related subtypes in CRC bulk transcriptomic data. The ssGSEA and CIBERSORT algorithms were used to evaluate the relative infiltration levels of different cell types in the tumor microenvironment (TME). Single-cell transcriptomic analysis was performed to confirm classification related to NRGs. NRG_score was developed to predict patients' survival outcomes with low-throughput validation in a patients' cohort from Fudan University Shanghai Cancer Center. RESULTS: We identified three distinct necroptosis-related classifications (NRCs) with discrepant clinical outcomes and biological functions. Characterization of TME revealed that there were two stable necroptosis-related phenotypes in CRC: a phenotype characterized by few TME cells infiltration but with EMT/TGF-pathways activation, and another phenotype recognized as immune-excluded. NRG_score for predicting survival outcomes was established and its predictive capability was verified. In addition, we found NRCs and NRG_score could be used for patient or drug selection when considering immunotherapy and chemotherapy. CONCLUSIONS: Based on comprehensive analysis, we revealed the potential roles of NRGs in the TME, and their correlations with clinicopathological parameters and patients' prognosis in CRC. These findings could enhance our understanding of the biological functions of necroptosis, which thus may aid in prognosis prediction, drug selection, and therapeutics development.

The Critical Role of Tumor Size in Predicting Prognosis for T1 Colon Cancer.

BACKGROUND: The role of horizontal growth index of tumor size in survival prediction is still underappreciated in colon cancer because of the identification of vertical infiltration index reflected by T stage. We sought to reveal the impact of T stage on the prognostic and predictive value of tumor size in colon cancer. MATERIALS AND METHODS: Data of patients with stage I-III colon cancer were extracted from Surveillance, Epidemiology, and End Results Program (SEER) and Fudan University Shanghai Cancer Center (FUSCC) databases. Harrell's concordance index (c-index) and time-dependent receiver operating characteristic curve (ROC) were used to analyze the discriminative ability of prognostic factors. RESULTS: Stratified analyses based on T stage found that the increase of T stage significantly and negatively repressed the effect of tumor size on death and recurrence risk. In addition, tumor size showed the greatest hazard ratio of cancer-specific death and relapse in T1 colon cancer. Even more importantly, the discriminatory ability of tumor size outperformed any other widely accepted prognostic clinical features in predicting cancer-specific survival (SEER: c-index 0.637, area under the ROC [AUC] 0.649; FUSCC: c-index 0.673, AUC 0.686) and disease-free survival (FUSCC: c-index 0.645, AUC 0.656) in T1 stage colon cancer. CONCLUSION: Tumor size is a critical clinical factor with considerable prognostic and predictive value for T1 colon cancer, and it should be selectively incorporated into the current staging system to facilitate prediction of death and recurrence risk. IMPLICATIONS FOR PRACTICE: To date, no consensus has been reached about the prognostic and predictive value of tumor size in colon cancer. Although tumor size is an independent prognostic factor for patients with colon cancer, the impact of tumor size on death or recurrence risk decreased notably with the increase of T stage. More importantly, the discriminative ability of tumor size outperformed any other clinical factors including N stage in patients with T1 colon cancer. Therefore, tumor size should be recommended to be incorporated into current staging systems to facilitate prognosis prediction for patients with T1 colon cancer.

A clinical-radiomics nomogram for the preoperative prediction of lymph node metastasis in colorectal cancer.

BACKGROUND: Accurate lymph node metastasis (LNM) prediction in colorectal cancer (CRC) patients is of great significance for treatment decision making and prognostic evaluation. We aimed to develop and validate a clinical-radiomics nomogram for the individual preoperative prediction of LNM in CRC patients. METHODS: We enrolled 766 patients (458 in the training set and 308 in the validation set) with clinicopathologically confirmed CRC. We included nine significant clinical risk factors (age, sex, preoperative carbohydrate antigen 19-9 (CA19-9) level, preoperative carcinoembryonic antigen (CEA) level, tumor size, tumor location, histotype, differentiation and M stage) to build the clinical model. We used analysis of variance (ANOVA), relief and recursive feature elimination (RFE) for feature selection (including clinical risk factors and the imaging features of primary lesions and peripheral lymph nodes), established classification models with logistic regression analysis and selected the respective candidate models by fivefold cross-validation. Then, we combined the clinical risk factors, primary lesion radiomics features and peripheral lymph node radiomics features of the candidate models to establish combined predictive models. Model performance was assessed by the area under the receiver operating characteristic (ROC) curve (AUC). Finally, decision curve analysis (DCA) and a nomogram were used to evaluate the clinical usefulness of the model. RESULTS: The clinical-primary lesion radiomics-peripheral lymph node radiomics model, with the highest AUC value (0.7606), was regarded as the candidate model and had good discrimination and calibration in both the training and validation sets. DCA demonstrated that the clinical-radiomics nomogram was useful for preoperative prediction in the clinical environment. CONCLUSION: The present study proposed a clinical-radiomics nomogram with a combination of clinical risk factors and radiomics features that can potentially be applied in the individualized preoperative prediction of LNM in CRC patients.

FOXE1 represses cell proliferation and Warburg effect by inhibiting HK2 in colorectal cancer.

BACKGROUND: Low expression of FOXE1, a member of Forkhead box (FOX) transcription factor family that plays vital roles in cancers, contributes to poor prognosis of colorectal cancer (CRC) patients. However, the underlying mechanism remains unclear. MATERIALS AND METHODS: The effects of FOXE1 on the growth of colon cancer cells and the expression of glycolytic enzymes were investigated in vitro and in vivo. Molecular biological experiments were used to reveal the underlying mechanisms of altered aerobic glycolysis. CRC tissue specimens were used to determine the clinical association of ectopic metabolism caused by dysregulated FOXE1. RESULTS: FOXE1 is highly expressed in normal colon tissues compared with cancer tissues and low expression of FOXE1 is significantly associated with poor prognosis of CRC patients. Silencing FOXE1 in CRC cell lines dramatically enhanced cell proliferation and colony formation and promoted glucose consumption and lactate production, while enforced expression of FOXE1 manifested the opposite effects. Mechanistically, FOXE1 bound directly to the promoter region of HK2 and negatively regulated its transcription. Furthermore, the expression of FOXE1 in CRC tissues was negatively correlated with that of HK2. CONCLUSION: FOXE1 functions as a critical tumor suppressor in regulating tumor growth and glycolysis via suppressing HK2 in CRC.

Nomogram to predict the risk and survival of synchronous bone metastasis in colorectal cancer: a population-based real-world analysis.

PURPOSE: Bone metastasis (BM) can obviously affect the quality of life of patients in colorectal cancer (CRC), and the whole management of patients with BM would be attractive in current clinical practice. METHODS: A total of 52,859 patients during 2010-2015 were collected from Surveillance, Epidemiology, and End Results (SEER) database. After propensity score matching (PSM), cancer-specific survival (CCS) and overall survival (OS) with BM were adopted to assess survival probability difference. Logistic regression was used to identify risk factors for BM; COX proportion hazard regression was applied to explore prognosticators for OS in patients with BM. Subsequently, nomograms were constructed and receiver operating curves (ROCs) were used to confirm the validation of nomogram. RESULTS: Three hundred and forty-two (0.65%) patients were diagnosed with synchronous BM. After PSM, 16 variables were balanced. Tumor site, histology, grade, T stage, N stage, CEA, radiochemotherapy, surgery, and liver/lung/brain metastases were associated with BM, and histology, grade, T stage, N stage, CEA, chemotherapy, surgery, and liver/lung metastases were prognosticators for BM survival. Nomograms were applied and the ROC curve proved the predictive effects. CONCLUSION: CRC patients with BM have worse real-world survival. Nomogram can predict incidence of BM in CRC patients and survival among patients with BM.

Prognostic accuracy of different lymph node staging system in predicting overall survival in stage IV colon cancer.

PURPOSE: With emphasis of surgical management, the lymph node (LN) status has been advocated to predict prognosis in colon cancer with distant metastatic. Therefore, we tend to compare the prognostic performance of American Joint Committee on Cancer (AJCC) N-staging relative to lymph node ratio (LNR), log odds of metastatic lymph nodes (LODDS), and N-score in stage IV colon cancer. METHODS: About 20,961 patients who underwent primary surgical resection for stage IV colon cancer were extracted from Surveillance, Epidemiology, and End Results (SEER) Program database. Harrell's C statistic (C-index) and Akaike's Information Criterion (AIC) were used to distinguish the prognostic performance of the different LN-staging schemes. RESULTS: Of the 20,961 patients, 17,043 (81.3%) had been with lymph node metastasis, and the median number of examined lymph nodes (ELNs) was 15. When assessed as continuous values, the LODDS shown as the best system with greatest discriminatory power (C-index, 0.6241; AIC, 29114.29) generally and each subgroups divided by ELNs. When modeled as categorical cutoff variables for further clinical usage, the 8th AJCC N-stage outperformed the other three schemes with either ELNs less than 12 (C-index, 0.5770; AIC, 8992.638), between 12 and 25 (C-index, 0.6084; AIC, 13905.72), or more than 25(C-index, 0.6192; AIC, 3138.018) with increasing C-index and less AIC value. CONCLUSIONS: When assessed as categorical variables, N-stage performed superiorly regardless of ELNs. When assessed as a continuous variable, LODDS exhibited good discriminative ability and goodness of fit in predicting survival for colon cancer patients regardless of ELNs.

Increasing age-related survival gap among patients with colorectal cancer: a population-based retrospective study.

BACKGROUND: Survival for patients with colorectal cancer (CRC) has improved over the past decades. However, it is unclear whether older patients have benefited to the same extent as younger patients. METHODS: The Surveillance, Epidemiology, and End Results (SEER) 9 registries database was queried for CRC patients from 1975 to 2009. We presented yearly data for survival with overlying loess-smoothing lines across all age groups. Another cohort was created using the SEER 18 registries database for patients diagnosed with CRC from 1973 to 2014. Yearly data for surgery-performed rate, stage proportion, and multivariate hazard ratio were performed with overlying smoothing lines across all age groups. RESULTS: In the analysis SEER 9, 5-year cause-specific survival (CSS) of patients aged ≤ 54, 55-64, and 65-74 years showed robust increase since 1975; however, the survival of patients aged 75-84 years remained low despite modest improvement, and patients aged 85 or older even showed no survival gains since 1990. In the analysis of SEER 18, there has been a steady increase in the survival of patients aged ≤ 54, 55-64, 65-74, and 75-84 years as time period advanced; however, of CRC patients aged ≥ 85 years, the survival curves of period 1990-1999 and 2000-2012 could not be distinguished from each other presented with negligibly a small gap from the curve of 1980-1989. CONCLUSIONS: The strong interaction between age and year of diagnosis implies that older patients have benefited less over time than younger patients, especially for patients aged ≥ 85 years.

Prognostic and predictive value of an autophagy-related signature for early relapse in stages I-III colon cancer.

We postulated that expression differences of autophagy-related genes are instrumental in stratifying the risk of early relapse after surgery and evaluating the prognosis of patients with stages I-III colon cancer. Therefore, propensity score matching analysis was performed between patients in early relapse group and long-term survival group from GSE39582 test series and internal validation series. Using Cox regression model, a nine-autophagy-related signature (CAPN2, ATG16L2, TP63, SIRT1, RPS6KB1, PEX3, ATG5, UVRAG, NAF1) was established to classify patients into those at high risk of early relapse (high-risk group), and those at low risk of early relapse (low-risk group). Relapse-free survival (RFS) was significantly different between the two groups in test [hazard ratio (HR): 2.019, 95% confidence interval (CI): 1.362-2.992, P < 0.001], internal validation (HR: 2.464, 95% CI: 1.196-5.079, P < 0.001) and another two external validation series (GSE14333-HR: 2.250, 95% CI: 1.227-4.126, P = 0.007; GSE33113-HR: 5.552, 95% CI: 2.098-14.693, P < 0.001). Then, based on RFS, we developed a nomogram, integrating the nine-autophagy-related classifier and four clinicopathological risk factors to evaluate prognosis of stages I-III colon cancer patients. Time-dependent receiver operating curve at 2 years showed that the integrated signature (area under curve = 0.758) had better prognostic accuracy than American Joint Committee on Cancer TNM stage (area under curve = 0.620). In conclusion, we identified and built a nine-autophagy-related signature, a credible approach to early relapse prediction in stages I-III colon cancer patients, which can assist physicians in devising more efficient therapeutic strategies.

Prognostic nomograms for predicting cause-specific survival and overall survival of stage I-III colon cancer patients: a large population-based study.

BACKGROUND: The purpose of this study was to build functional nomograms based on significant clinicopathological features to predict cause-specific survival (CSS) and overall survival (OS) in patients with stage I-III colon cancer. METHODS: Data on patients diagnosed with stage I-III colon cancer between 2010 and 2015 were downloaded from the Surveillance, Epidemiology, and End Results (SEER) database. Univariate and multivariate Cox analyses were used to identify independent prognostic factors, which were used to construct nomograms to predict the probabilities of CSS and OS. The performance of the nomogram was assessed by C-indexes, receiver operating characteristic (ROC) curves and calibration curves. Decision curve analysis (DCA) was used to compare clinical usage between the nomogram and the tumor-node-metastasis (TNM) staging system. RESULTS: Based on the univariate and multivariate analyses, features that correlated with survival outcomes were used to establish nomograms for CSS and OS prediction. The nomograms showed favorable sensitivity at predicting 1-, 3-, and 5-year CSS and OS, with a C-index of 0.78 (95% confidence interval (CI) 0.77-0.80) for CSS and 0.74 (95% CI 0.73-0.75) for OS. Calibration curves and ROC curves revealed excellent predictive accuracy. The clinically and statistically significant prognostic performance of the nomogram generated with the entire group of patients and risk scores was validated by a stratified analysis. DCA showed that the nomograms were more clinically useful than TNM stage. CONCLUSION: Novel nomograms based on significant clinicopathological characteristics were developed and can be used as a tool for clinicians to predict CSS and OS in stage I-III colon cancer patients. These models could help facilitate a personalized postoperative evaluation.

PRMT5 enhances tumorigenicity and glycolysis in pancreatic cancer via the FBW7/cMyc axis.

BACKGROUND: The epigenetic factor protein arginine methyltransferase 5 (PRMT5) has been reported to play vital roles in a wide range of cellular processes, such as gene transcription, genomic organization, differentiation and cell cycle control. However, its role in pancreatic cancer remains unclear. Our study aimed to investigate the roles of PRMT5 in pancreatic cancer prognosis and progression and to explore the underlying molecular mechanism. METHODS: Real-time PCR, immunohistochemistry and analysis of a dataset from The Cancer Genome Atlas (TCGA) were performed to study the expression of PRMT5 at the mRNA and protein levels in pancreatic cancer. Cell proliferation assays, including cell viability, colony formation ability and subcutaneous mouse model assays, were utilized to confirm the role of PRMT5 in cell proliferation and tumorigenesis. A Seahorse extracellular flux analyzer, a glucose uptake kit, a lactate level measurement kit and the measurement of 18F-FDG (fluorodeoxyglucose) uptake by PET/CT (positron emission tomography/computed tomography) imaging were used to verify the role of PRMT5 in aerobic glycolysis, which sustains cell proliferation. The regulatory effect of PRMT5 on cMyc, a master regulator of oncogenesis and aerobic glycolysis, was explored by quantitative PCR and protein stability measurements. RESULTS: PRMT5 expression was significantly upregulated in pancreatic cancer tissues compared with that in adjacent normal tissues. Clinically, elevated expression of PRMT5 was positively correlated with worse overall survival in pancreatic cancer patients. Silencing PRMT5 expression inhibited the proliferation of pancreatic cancer cells both in vitro and in vivo. Moreover, PRMT5 regulated aerobic glycolysis in vitro in cell lines, in vivo in pancreatic cancer patients and in a xenograft mouse model used to measure 18F-FDG uptake. We found that mechanistically, PRMT5 posttranslationally regulated cMyc stability via F-box/WD repeat-containing protein 7 (FBW7), an E3 ubiquitin ligase that controls cMyc degradation. Moreover, PRMT5 epigenetically regulated the expression of FBW7 in pancreatic cancer cells. CONCLUSIONS: The present study demonstrated that PRMT5 epigenetically silenced the expression of the tumor suppressor FBW7, leading to increased cMyc levels and the subsequent enhancement of the proliferation of and aerobic glycolysis in pancreatic cancer cells. The PRMT5/FBW7/cMyc axis could be a potential therapeutic target for the treatment of pancreatic cancer.

Overexpression of MTA1 inhibits the metastatic ability of ZR-75-30 cells in vitro by promoting MTA2 degradation.

BACKGROUND: As the first member of the metastasis-associated protein (MTA) family, MTA1 and another MTA family member, MTA2, have both been reported to promote breast cancer progression and metastasis. However, the difference and relationship between MTA1 and MTA2 have not been fully elucidated. METHODS: Transwell assays were used to assess the roles of MTA1 and MTA2 in the metastasis of ZR-75-30 luminal B breast cancer cells in vitro. Immunoblotting and qRT-PCR were used to evaluate the effect of MTA1 overexpression on MTA2. Proteases that cleave MTA2 were predicted using an online web server. The role of neutrophil elastase (NE) in MTA1 overexpression-induced MTA2 downregulation was confirmed by specific inhibitor treatment, knockdown, overexpression and immunocytochemistry, and NE cleavage sites in MTA2 were confirmed by MTA2 truncation and mutation. The effect of MTA1 overexpression on the intrinsic inhibitor of NE, elafin, was detected by qRT-PCR, immunoblotting and treatment with inhibitors. RESULTS: MTA1 overexpression inhibited, while MTA2 promoted the metastasis of ZR-75-30 cells in vitro. MTA1 overexpression downregulated MTA2 expression at the protein level rather than the mRNA level. NE was predicted to cleave MTA2 and was responsible for MTA1 overexpression-induced MTA2 degradation. NE was found to cleave MTA2 in the C-terminus at the 486, 497, 542, 583 and 621 sites. MTA1 overexpression activated NE by downregulating elafin in a histone deacetylase- and DNA methyltransferase-dependent manner. CONCLUSIONS: MTA1 and MTA2 play opposing roles in the metastasis of ZR-75-30 luminal B breast cancer cells in vitro. MTA1 downregulates MTA2 at the protein level by epigenetically repressing the expression of elafin and releasing the inhibition of neutrophil elastase, which cleaves MTA2 in the C-terminus at multiple specific sites.

Epidemiology of and prognostic factors for appendiceal carcinomas: a retrospective, population-based study.

BACKGROUND: The aim of this study is to evaluate the epidemiology of and prognostic factors for appendiceal carcinomas (ACs). METHODS: All cases of ACs registered in the Surveillance, Epidemiology, and End Results (SEER) database from 1973 to 2014 were retrospectively identified in this study. Age-adjusted incidence and survival rates were calculated. RESULTS: We analyzed 7170 patients with ACs. We observed a significant increase in the reported annual age-adjusted incidence of ACs from 1973 (0.18/100,000) to 2014 (1.11/100,000). The elevation of the incidence was noted in all the histological types, stages, and grades. The most common histological type varied by race, with the appendiceal mucinous adenocarcinoma (AMA) being the most common in white, Asian/Pacific Islander, and American Indian/Alaskan Native patients, and the appendiceal adenocarcinoma (AA) being the most common in African American patients. In multivariate analysis of patients with all ACs, gender (P < 0.001), year of diagnosis (P < 0.001), age (P < 0.001), race (P < 0.001), tumor grade (P < 0.001), disease stage (P < 0.001), retrieved regional lymph nodes (P < 0.001), type of surgery performed (P = 0.002), and histologic subtype (P < 0.001) were predictors of outcome. Survival time for all ACs increased from the 1973-1993 period to the 1994-2014 period (HR 0.76; 95% CI, 0.69 to 0.85). Additionally, the 5-year survival rates were 88% for malignant carcinoid, 70% for goblet cell carcinoid, 51% for colonic type adenocarcinoma, 59% for mucinous adenocarcinoma, and 25% for signet ring cell type. CONCLUSIONS: We observed increased reported incidence of ACs and increased survival durations over time, suggesting that clinicians pay more attention to ACs and mastering the characteristic of these tumors.

Laparoscopic surgery may decrease the risk of clinical anastomotic leakage and a nomogram to predict anastomotic leakage after anterior resection for rectal cancer.

INTRODUCTION: Anastomotic leakage is still one of the most dreaded complications after anterior resection for rectal cancer. This study aimed to identify risk factors for anastomotic leakage and to create a nomogram for precise prediction of anastomotic leakage after anterior resection for rectal cancer. METHODS: Two thousand six hundred eighteen consecutive patients who underwent anterior resection for rectal cancer with primary anastomosis, with or without diverting stoma, were retrospectively analyzed as a training dataset. Univariate and multivariable Cox regression analyses were used to determine independent risk factors associated with anastomotic leakage. A nomogram was constructed to predict anastomotic leakage. Data containing 611 patients were prospectively collected as a test dataset. The performance of the nomogram was evaluated by using a bootstrapped-concordance index and calibration plots. RESULTS: The rate of clinical anastomotic leakage was 9.3% in the training dataset. Multivariate analysis identifies the following variables as independent risk factors for anastomotic leakage: gender (male) (odds ratio (OR) = 2.286), distance of tumor to anal verge (OR = 0.791), tumor size (OR = 1.175), operating time (OR = 1.009), diabetes mellitus (OR = 1.704), laparoscopic surgery (OR = 0.445), anastomotic bleeding (OR = 13.46), and diverting stoma (OR = 0.386). We created a nomogram with high discriminative ability (concordance index, 0.722). The area under the curve value, which evaluated the predictive performance of external validation, was 0.723. CONCLUSIONS: A protective diverting stoma and laparoscopic surgery significantly decrease the risk of anastomotic leakage. Our nomogram was a useful tool for precise prediction of anastomotic leakage after anterior resection for rectal cancer.

A Ligand-Based Virtual Screening Method Using Direct Quantification of Generalization Ability.

Machine learning plays an important role in ligand-based virtual screening. However, conventional machine learning approaches tend to be inefficient when dealing with such problems where the data are imbalanced and features describing the chemical characteristic of ligands are high-dimensional. We here describe a machine learning algorithm LBS (local beta screening) for ligand-based virtual screening. The unique characteristic of LBS is that it quantifies the generalization ability of screening directly by a refined loss function, and thus can assess the risk of over-fitting accurately and efficiently for imbalanced and high-dimensional data in ligand-based virtual screening without the help of resampling methods such as cross validation. The robustness of LBS was demonstrated by a simulation study and tests on real datasets, in which LBS outperformed conventional algorithms in terms of screening accuracy and model interpretation. LBS was then used for screening potential activators of HIV-1 integrase multimerization in an independent compound library, and the virtual screening result was experimentally validated. Of the 25 compounds tested, six were proved to be active. The most potent compound in experimental validation showed an EC50 value of 0.71 µM.

Beta Distribution-Based Cross-Entropy for Feature Selection.

Analysis of high-dimensional data is a challenge in machine learning and data mining. Feature selection plays an important role in dealing with high-dimensional data for improvement of predictive accuracy, as well as better interpretation of the data. Frequently used evaluation functions for feature selection include resampling methods such as cross-validation, which show an advantage in predictive accuracy. However, these conventional methods are not only computationally expensive, but also tend to be over-optimistic. We propose a novel cross-entropy which is based on beta distribution for feature selection. In beta distribution-based cross-entropy (BetaDCE) for feature selection, the probability density is estimated by beta distribution and the cross-entropy is computed by the expected value of beta distribution, so that the generalization ability can be estimated more precisely than conventional methods where the probability density is learnt from data. Analysis of the generalization ability of BetaDCE revealed that it was a trade-off between bias and variance. The robustness of BetaDCE was demonstrated by experiments on three types of data. In the exclusive or-like (XOR-like) dataset, the false discovery rate of BetaDCE was significantly smaller than that of other methods. For the leukemia dataset, the area under the curve (AUC) of BetaDCE on the test set was 0.93 with only four selected features, which indicated that BetaDCE not only detected the irrelevant and redundant features precisely, but also more accurately predicted the class labels with a smaller number of features than the original method, whose AUC was 0.83 with 50 features. In the metabonomic dataset, the overall AUC of prediction with features selected by BetaDCE was significantly larger than that by the original reported method. Therefore, BetaDCE can be used as a general and efficient framework for feature selection.

Transcriptome profiling reveals an integrated mRNA-lncRNA signature with predictive value of early relapse in colon cancer.

The purpose of our study was to develop a multigene signature based on transcriptome profiles of both mRNAs and lncRNAs to identify a group of patients who are at high risk of early relapse in stages II-III colon cancer. Firstly, propensity score matching was conducted between patients in early relapse group and long-term survival group from GSE39582 training series (N = 359) and patients were matched 1:1. Global transcriptome analysis was then performed between the paired groups to identify tumor specific mRNAs and lncRNAs. Finally, using LASSO Cox regression model, we built a multigene early relapse classifier incorporating 15 mRNAs and three lncRNAs. The prognostic and predictive accuracy of the signature was internally validated in 102 colon cancer patients and externally validated in other 241 patients. In the training set, patients with high risk score were more likely to suffer from relapse than those with low risk score (HR: 2.67, 95% CI: 2.07-3.46, P < 0.001). The results were validated in the internal validation set (HR: 2.23, 95% CI: 1.23-3.78, P = 0.003) and external validation (HR 1.88, 95% CI 1.42-2.48; P < 0.001) set. Time-dependent receiver operating curve at 1 year showed that the integrated mRNA-lncRNA signature [area under curve (AUC) = 0.742] had better prognostic accuracy than AJCC TNM stage (AUC = 0.615) in the entire 702 patients. In addition, survival decision curve analyses at 12 months revealed a good clinical usefulness of the integrated mRNA-lncRNA signature. In conclusion, we successfully developed an integrated mRNA-lncRNA signature that can accurately predict early relapse.

Clinicopathologic Features and Prognostic Factors in Alpha-Fetoprotein-Producing Colorectal Cancer: Analysis of 78 Cases.

BACKGROUND/AIMS: Alpha-fetoprotein-producing colorectal cancer (AFPP-CRC) is quite rarely seen. This study aimed to elucidate the clinicopathologic characteristics and prognostic factors of AFPP-CRC. METHODS: Among 5,051 colorectal cancer patients receiving surgery in the Fudan University Shanghai Cancer Center from 2006 to 2016, we identified 78 patients with elevated serum level of AFP (> 10 µg/L) preoperatively. A propensity score matching (PSM) analysis was performed which matched 75 AFPP-CRC patients to the same number of AFP-negative colorectal cancer (AFPN-CRC) patients. Kaplan-Meier curves were compared using the log-rank test and multivariable analysis was performed to evaluate the effect of AFP-positivity while adjusting confounding factors. 27 patients were available for immunohistochemical analysis. We conducted functional experiments to characterize the tumorigenicity of AFP. RESULTS: Patients with AFPP-CRC had a significantly higher incidence of advanced TNM stage and liver metastasis. Overall survival was significantly different between two groups before and after PSM, and AFP-positivity was one of the strongest predictors of overall survival in the multivariable model (HR 4.11, CI 95%: 1.43-11.76, p = 0.009) after PSM. We further investigated prognostic factors affecting prognosis in AFPP-CRC and found that the presence of liver metastasis was the only independent prognostic factor (HR 4.95, CI 95%: 1.48-16.48, p = 0.009). AFP expression was significantly positively correlated with HGF and c-Met expression. Transwell invasion assay revealed significantly increased cell motility with AFP overexpression. CONCLUSION: AFP-positivity is a significant negative predictor of overall survival in patients with colorectal cancer, which may be mediated by HGF/c-Met signaling pathway.