Artificial neural network for prediction of distant metastasis in colorectal cancer

Background and Objectives: Artificial neural networks (ANNs) are flexible and nonlinear models which can be used by clinical oncologists in medical research as decision making tools. This study aimed to predict distant metastasis (DM) of colorectal cancer (CRC) patients using an ANN model. Methods: The data of this study were gathered from 1219 registered CRC patients at the Research Center for Gastroenterology and Liver Disease of Shahid Beheshti University of Medical Sciences, Tehran, Iran (January 2002 and October 2007). For prediction of DM in CRC patients, neural network (NN) and logistic regression (LR) models were used. Then, the concordance index (C index) and the area under receiver operating characteristic curve (AUROC) were used for comparison of neural network and logistic regression models. Data analysis was performed with R 2.14.1 software. Results: The C indices of ANN and LR models for colon cancer data were calculated to be 0.812 and 0.779, respectively. Based on testing dataset, the AUROC for ANN and LR models were 0.82 and 0.77, respectively. This means that the accuracy of ANN prediction was better than for LR prediction. Conclusion: The ANN model is a suitable method for predicting DM and in that case is suggested as a good classifier that usefulness to treatment goals.     

Supervised learning applied to classifying fallers versus non-fallers among older adults with cancer

IntroductionSupervised machine learning approaches are increasingly used to analyze clinical data, including in geriatric oncology. This study presents a machine learning approach to understand falls in a cohort of older adults with advanced cancer starting chemotherapy, including fall prediction and identification of contributing factors.Materials and MethodsThis secondary analysis of prospectively collected data from the GAP 70+ Trial (NCT02054741; PI: Mohile) enrolled patients aged ≥70 with advanced cancer and ≥ 1 geriatric assessment domain impairment who planned to start a new cancer treatment regimen. Of ≥2000 baseline variables (“features”) collected, 73 were selected based on clinical judgment. Machine learning models to predict falls at three months were developed, optimized, and tested using data from 522 patients. A custom data preprocessing pipeline was implemented to prepare data for analysis. Both undersampling and oversampling techniques were applied to balance the outcome measure. Ensemble feature selection was applied to identify and select the most relevant features. Four models (logistic regression [LR], k-nearest neighbor [kNN], random forest [RF], and MultiLayer Perceptron [MLP]) were trained and subsequently tested on a holdout set. Receiver operating characteristic (ROC) curves were generated and area under the curve (AUC) was calculated for each model. SHapley Additive exPlanations (SHAP) values were utilized to further understand individual feature contributions to observed predictions.ResultsBased on the ensemble feature selection algorithm, the top eight features were selected for inclusion in the final models. Selected features aligned with clinical intuition and prior literature. The LR, kNN, and RF models performed equivalently well in predicting falls in the test set, with AUC values 0.66–0.67, and the MLP model showed AUC 0.75. Ensemble feature selection resulted in improved AUC values compared to using LASSO alone. SHAP values, a model-agnostic technique, revealed logical associations between selected features and model predictions.DiscussionMachine learning techniques can augment hypothesis-driven research, including in older adults for whom randomized trial data are limited. Interpretable machine learning is particularly important, as understanding which features impact predictions is a critical aspect of decision-making and intervention. Clinicians should understand the philosophy, strengths, and limitations of a machine learning approach applied to patient data.     

Development of a machine learning algorithm to predict complications of total laparoscopic anterior resection and natural orifice specimen extraction surgery in rectal cancer

BackgroundTotal laparoscopic anterior resection (tLAR) and natural orifice specimen extraction surgery (NOSES) has been widely adopted in the treatment of rectal cancer (RC). However, no study has been performed to predict the short-term outcomes of tLAR using machine learning algorithms to analyze a national cohort.MethodsData from consecutive RC patients who underwent tLAR were collected from the China NOSES Database (CNDB). The random forest (RF), extreme gradient boosting (XGBoost), support vector machine (SVM), deep neural network (DNN), logistic regression (LR) and K-nearest neighbor (KNN) algorithms were used to develop risk models to predict short-term complications of tLAR. The area under the receiver operating characteristic curve (AUROC), Gini coefficient, specificity and sensitivity were calculated to assess the performance of each risk model. The selected factors from the models were evaluated by relative importance.ResultsA total of 4313 RC patients were identified, and 667 patients (15.5%) developed postoperative complications. The machine learning model of XGBoost showed more promising results in the prediction of complication than other models (AUROC 0.90, P < 0.001). The performance was similar when internal and external validation was used. In the XGBoost model, the top four influential factors were the distance from the lower edge of the tumor to the anus, age at diagnosis, surgical time and comorbidities. In risk stratification analysis, the rate of postoperative complications in the high-risk group was significantly higher than in the medium- and low-risk groups (P < 0.001).ConclusionThe machine learning model shows potential benefits in predicting the risk of complications in RC patients after tLAR. This novel approach can provide reliable individual information for surgical treatment recommendations.     

结肠癌淋巴结转移的风险基因及列线图预测模型的构建

目的 寻找结肠癌淋巴结转移的相关风险基因，并构建由基因组成的列线图（nomogram）预测模型。 方法 从TCGA和GEO数据库下载基因测序数据，利用差异分析和LASSO回归方法筛选基因。利用赤池信息准则确定最优的nomogram模型，ROC曲线、校准曲线及拟合优度检验评估模型预测的准确性，决策曲线分析评估临床应用价值。 结果 通过筛选得到11个有效预测结肠癌淋巴结转移的基因。由年龄、病理T分期、TH、CDH4、PNMA6A、TNNC1、KIR2DL4、STUM、SFTA2构成的nomogram模型具有最小的AIC值（440.4）。内部评估模型AUC值为0.800，外部验证AUC值为0.664，校准度及拟合优度均较佳。临床决策曲线分析法评估基于nomogram模型的风险判断可以带来临床获益。结论 共筛选出11个结肠癌淋巴结转移的风险基因。构建的nomogram预测模型的一致性和区分度良好，可帮助评估患者淋巴结转移状态。     

Comparison of Classification Success Rates of Different Machine Learning Algorithms in the Diagnosis of Breast Cancer

Objective: To identify which Machine Learning (ML) algorithms are the most successful in predicting and diagnosing breast cancer according to accuracy rates. Methods: The “College of Wisconsin Breast Cancer Dataset”, which consists of 569 data and 30 features, was classified using Support Vector Machine (SVM), Naive Bayes (NB), Random Forest (RF), Decision Tree (DT), K-Nearest Neighbor (KNN), Logistic Regression (LR), Multilayer Perceptron (MLP), Linear Discriminant Analysis (LDA), XgBoost (XGB), Ada-Boost (ABC) and Gradient Boosting (GBC) ML algorithms. Before the classification process, the dataset was preprocessed. Sensitivity, accuracy, and definiteness metrics were used to measure the success of the methods. Result: Compared to other ML algorithms used in the study, the GBC ML algorithm was found to be the most successful method in the classification of tumors with an accuracy of 99.12%. The XGB ML algorithm was found to be the lowest method with an accuracy rate of 88.10%. In addition, it was determined that the general accuracy rates of the 11 ML algorithms used in the study varied between 88-95%.Conclusion: When the results obtained from the ML classifiers used in the study are evaluated, the efficiency of the GBC algorithm in the classification of tumors is obvious. It can be said that the success rates obtained from 11 different ML algorithms used in the study are valuable in terms of being used to predict different cancer types.     

基于影像组学和人工智能预测非小细胞肺癌放化疗疗效的可行性

目的:构建基于影像组学特征的预测模型，以预测非小细胞肺癌患者接受序贯放化疗（sequential chemoradiotherapy,SCRT)或同步放化疗（concurrent chemoradiotherapy,CCRT）后的病情部分缓解（partial response,PR）可能性。方法:回顾性收集2016年01月至2020年06月确诊为非小细胞肺癌并接受SCRT或CCRT患者资料。符合条件的患者纳入本研究中，并随机分为训练集和验证集。采用单因素方差分析及最小绝对收缩和选择算子（least absolute shrinkage and selection operator，LASSO）算法，在训练集中筛选出最佳影像组学特征。在训练集中进行机器学习（Logistic regression,LR;Decision tree,DT;AdaBoost）模型构建。受试者工作特征曲线下面积(area under curve,AUC)、敏感性和特异性用于评估模型性能，使用列线图对模型进行可视化，决策曲线分析法检验模型应用效能。结果:共纳入75例患者，随机分为两组，训练集52例，验证集23例。在进行单因素方差分析和LASSO回归分析后，筛选出了6个放射学特征，使用机器学习方法构建预测模型。在训练集中，LR、DT、AdaBoost的模型的AUC为0.919、0.773及0.832，在验证集中为0.795、0.723及0.638。使用LR模型构建决策曲线表明，当风险阈值为0.1~0.92时，可增加患者的净效益。结论:本研究开发并验证了一个影像组学预测模型，可以预测接受SCRT/CCRT后肺癌患者的缓解概率。     

Lung Cancer and Granuloma Identification Using a Deep Learning Model to Extract 3-Dimensional Radiomics Features in CT Imaging

BackgroundWe aimed to evaluate a deep learning (DL) model combining perinodular and intranodular radiomics features and clinical features for preoperative differentiation of solitary granuloma nodules (GNs) from solid lung cancer nodules in patients with spiculation, lobulation, or pleural indentation on CT.Patients and MethodsWe retrospectively recruited 915 patients with solitary solid pulmonary nodules and suspicious signs of malignancy. Data including clinical characteristics and subjective CT findings were obtained. A 3-dimensional U-Net-based DL model was used for tumor segmentation and extraction of 3-dimensional radiomics features. We used the Maximum Relevance and Minimum Redundancy (mRMR) algorithm and the eXtreme Gradient Boosting (XGBoost) algorithm to select the intranodular, perinodular, and gross nodular radiomics features. We propose a medical image DL (IDL) model, a clinical image DL (CIDL) model, a radiomics DL (RDL) model, and a clinical image radiomics DL (CIRDL) model to preoperatively differentiate GNs from solid lung cancer. Five-fold cross-validation was used to select and evaluate the models. The prediction performance of the models was evaluated using receiver operating characteristic and calibration curves.ResultsThe CIRDL model achieved the best performance in differentiating between GNs and solid lung cancer (area under the curve [AUC] = 0.9069), which was significantly higher compared with the IDL (AUC = 0.8322), CIDL (AUC = 0.8652), intra-RDL (AUC = 0.8583), peri-RDL (AUC = 0.8259), and gross-RDL (AUC = 0.8705) models.ConclusionThe proposed CIRDL model is a noninvasive diagnostic tool to differentiate between granuloma nodules and solid lung cancer nodules and reduce the need for invasive diagnostic and surgical procedures.     

Machine learning models to prognose 30-Day Mortality in Postoperative Disseminated Cancer Patients

Patients with disseminated cancer at higher risk for postoperative mortality see improved outcomes with altered clinical management. Being able to risk stratify patients immediately after their index surgery to flag high risk patients for healthcare providers is vital. The combination of physician uncertainty and a demonstrated optimism bias often lead to an overestimation of patient life expectancy which can precent proper end of life counseling and lead to inadequate postoperative follow up. In this cohort study of 167,474 postoperative patients with multiple types of disseminated cancer, patients at high risk of 30-day postoperative mortality were accurately identified using our machine learning models based solely on clinical features and preoperative lab values. Extreme Gradient Boosting, Random Forest, and Logistic Regression machine learning models were developed on the cohort. Among 167,474 disseminated cancer patients, 50,669 (30.3%) died within 30 days of their index surgery; After preprocessing, 28 features were included in the model development. The cohort was randomly divided into 133,979 patients (80%) for training the models and 33,495 patients (20%) for testing. The extreme gradient boosting model had an AUC of 0.93 (95% CI: 0.926–0.931), the random forest model had an AUC of 0.93 (95% CI: 0.930–0.934), and the logistic regression model had an AUC of 0.90 (95% CI: 0.900–0.906 the index operation. Ultimately, Machine learning models were able to accurately predict short-term postoperative mortality among a heterogenous population of disseminated cancer patients using commonly accessible medical features. These models can be included in electronic health systems to guide clinical judgements that affect direct patient care, particularly in low-resource settings.     

Subjective assessment versus ultrasound models to diagnose ovarian cancer: A systematic review and meta-analysis

IntroductionMany national guidelines concerning the management of ovarian cancer currently advocate the risk of malignancy index (RMI) to characterise ovarian pathology. However, other methods, such as subjective assessment, International Ovarian Tumour Analysis (IOTA) simple ultrasound-based rules (simple rules) and IOTA logistic regression model 2 (LR2) seem to be superior to the RMI.Our objective was to compare the diagnostic accuracy of subjective assessment, simple rules, LR2 and RMI for differentiating benign from malignant adnexal masses prior to surgery.Materials and methodsMEDLINE, EMBASE and CENTRAL were searched (January 1990–August 2015). Eligibility criteria were prospective diagnostic studies designed to preoperatively predict ovarian cancer in women with an adnexal mass.ResultsWe analysed 47 articles, enrolling 19,674 adnexal tumours; 13,953 (70.9%) benign and 5721 (29.1%) malignant. Subjective assessment by experts performed best with a pooled sensitivity of 0.93 (95% confidence interval [CI] 0.92–0.95) and specificity of 0.89 (95% CI 0.86–0.92). Simple rules (classifying inconclusives as malignant) (sensitivity 0.93 [95% CI 0.91–0.95] and specificity 0.80 [95% CI 0.77–0.82]) and LR2 (sensitivity 0.93 [95% CI 0.89–0.95] and specificity 0.84 [95% CI 0.78–0.89]) outperformed RMI (sensitivity 0.75 [95% CI 0.72–0.79], specificity 0.92 [95% CI 0.88–0.94]). A two-step strategy using simple rules, when inconclusive added by subjective assessment, matched test performance of subjective assessment by expert examiners (sensitivity 0.91 [95% CI 0.89–0.93] and specificity 0.91 [95% CI 0.87–0.94]).ConclusionsA two-step strategy of simple rules with subjective assessment for inconclusive tumours yielded best results and matched test performance of expert ultrasound examiners. The LR2 model can be used as an alternative if an expert is not available.     

A proposed Clino-radio-pathological Risk Scoring System (CRiSS) for prediction and management of inguinal lymph-nodes metastasis in squamous cell carcinoma of the penis

ObjectiveTo develop a risk scoring system for prediction of inguinal lymph-node involvement and to suggest a management strategy according to the risk groups based on clinical, radiological and pathological parameters in squamous cell carcinoma (SCC) of penis.Materials and MethodsA retrospective analysis of all patients of SCC penis from 2014 to 2020 at our institute was done. The patients were divided into derivation cohort (2014 to 2019) and validation cohort (2019 to 2020). A total of 10 predictors were analysed in univariate analysis and those found significant were further subjected to multivariate analysis to derive regression coefficient for each. CRiSS scores were assigned based on the coefficients and three groups were created which were correlated with nodal metastasis. The predictive accuracy of the model was assessed by ROC analysis of the derivation cohort and validation cohort.ResultsA total of 102 patients were identified in derivation cohort and 23 patients in validation cohort. Size of the primary >3cm, ulceroinfiltrative growth, involving shaft, ultrasound size of lymph-nodes >1cm, loss of fatty hila, moderate and poor differentiation, and lypmphovascular/perineural invasion were independent predictors of inguinal lymphnode metastasis in multivariate analysis. CRiSS could achieve AUROC of .910 and .887 in derivation and validation cohort respectively. The rate of metastatic lymphadenopathy was 0%, 41.4%, and 89.5% in low, intermediate, and high-risk groups respectively.ConclusionsCRiSS can effectively predict inguinal lymph-node metastasis in SCC penis. We suggest a management strategy based on risk groups that will avoid morbidity of groin dissection in many patients.     

Treatment strategies and outcomes of primary Myxofibrosarcomas in a large patients cohort

Backgroundthis study analysed primary myxofibrosarcoma (MFS) to investigate patient outcomes focusing on histopathologic margins and perioperative treatments.Patients and methodsdata from consecutive patients affected by primary and localized MFS of the extremities or trunk wall who underwent surgery (2002–2017) were analysed. Local recurrence (LR), amputation rate, incidence of distant metastasis (DM), and overall survival (OS) were studied.ResultsOf 293 included patients, 52 (17%) patients received perioperative treatments and 54 (18%) had positive microscopic histopathologic margins (R1). Median follow-up was 80 months (IQR, 49–109). 5-yr CCI of LR was 0.12 (SE: 0.02). Status of histopathologic margins (P < 0.001), tumour malignancy grade (P = 0.018) and size (P = 0023) were independent prognostic factor for LR. Nine amputations (amputation rate: 3%) were performed (N = 1 for primary tumour; N = 8 for LR). Larger tumour size (P = 0.015) and higher grade (P = 0.025) were independent prognostic factor for DM. 5-year OS was 0.84 (95%CI 0.79–0.88). Patient age (P = 0.008), tumour size (P = 0.013) and malignancy grade (P = 0.018) were independently associated to OS. In the subgroup of patients who had a re-excision for a primary MFS (N = 116, 40%), the presence of residual disease was not associated with LR, DM, or OS.Conclusionin this study 5-year LR, DM and OS were 12%, 17%, and 84%, respectively. One in six patients had a positive surgical margin, which was a prognostic factor for LR, while DM and OS were predicted by tumour grade and size. Findings from this large patient cohort may set benchmarks for investigating new treatment options for MFS.     

Clinical value of lymph node dissection for stage T1b mid-thoracic esophageal squamous cell carcinoma

BackgroundSurgery remains the mainstay of treatment for esophageal squamous cell carcinoma (ESCC), during which lymph node (LN) dissection, especially recurrent laryngeal nerve (RLN) LN dissection, is particularly important and challenging. This study aimed to investigate the LN metastasis of stage T1b mid-thoracic ESCC and explore the clinical value of RLN LN dissection.MethodsThe clinicopathological data of 254 patients with stage T1b mid-thoracic ESCC who underwent the McKeown procedure (“tri-incisional esophagectomy”) and three-field LN dissection (3FD) at Fujian Cancer Hospital from January 2010 to December 2015 were retrospectively analyzed. The value of LN dissection (especially RLN LNs) was evaluated by calculating the metastasis rate of each LN station. The efficacy index (EI) of the dissection was calculated by multiplying the frequency (%) of metastases to a station and the 5-year survival rate (%) of patients with metastases to that station, and then dividing by 100.ResultsThe stage T1b mid-thoracic ESCC had the highest rate of metastasis in the paracardiac LNs (4.3%), followed by RLN LNs (2.8%) and the left gastric artery LNs (2.8%). The 5-year survival rate was highest in patients who received lesser gastric curvature LN dissection (100%), followed by patients who underwent right RLN LN dissection (80%), and was 50% in patients who had undergone dissection of the left RLN LNs, upper paraesophageal LNs, subcarinal LNs, and left gastric artery LNs, respectively. In addition, dissection of the right RLN LNs had the highest EI value (2.2), followed by the dissection of LNs along the lesser curvature of the stomach (1.6) and left gastric artery LNs (1.4).ConclusionsRight RLN LNs have a metastasis rate only lower than that of the paracardiac LNs, but could be the most valuable location for performing dissection.     

Head and neck soft tissue sarcomas: prognostic factors and outcome in a series of patients treated at a single institution

BackgroundHead and neck soft tissue sarcomas (STS) represent a rare disease.Patients and methodsOne hundred and sixty-seven patients underwent surgery at our institution with an eradicating intent between 1990 and 2010. Local recurrence (LR), distant metastasis (DM) and disease-specific mortality (DSM) incidence were studied along with clinicopathological prognostic factors.ResultsTen-year crude cumulative incidence (CCI) of LR, DM and DSM were 19%, 11% and 26%, respectively (median follow-up 66 months). Independent prognostic factors for DSM were tumor size (P < 0.001) and grade (P = 0.032), while surgical margins obtained a border-line significance (0.070); LR was affected by the tumor size (P = 0.001), while DM only by grade (P = 0.047). The median survival after LR and DM were 14 months and 7 months, respectively. Tumors sited in the paranasal sinus and supraclavicular region had the worst survival.ConclusionsHead and neck represent a very critical anatomical site for STS. Achievement of local disease control appears to be crucial, since even LR could be a life-threatening event.     

Selective Mediastinal Lymph Node Dissection Strategy for Clinical T1N0 Invasive Lung Cancer: A Prospective,Multicenter, Clinical Trial

IntroductionWe aimed to prospectively evaluate our previously proposed selective mediastinal lymph node (LN) dissection strategy for peripheral clinical T1N0 invasive NSCLC.MethodsThis is a multicenter, prospective clinical trial in China. We set six criteria for predicting negative LN stations and finally guiding selective LN dissection. Consolidation tumor ratio less than or equal to 0.5, segment location, lepidic-predominant adenocarcinoma (LPA), negative hilar nodes (stations 10–12), and negative visceral pleural invasion (VPI) were used separately or in combination as predictors of negative LN status in the whole, superior, or inferior mediastinal zone. LPA, hilar node involvement, and VPI were diagnosed intraoperatively. All patients actually underwent systematic mediastinal LN dissection. The primary end point was the accuracy of the strategy in predicting LN involvement. If LN metastasis occurred in certain mediastinal zone that was predicted to be negative, it was considered as an “inaccurate” case.ResultsA total of 720 patients were enrolled. The median number of LN dissected was 15 (interquartile range: 11–20). All negative node status in certain mediastinal zone was correctly predicted by the strategy. Compared with final pathologic findings, the accuracy of frozen section to diagnose LPA, VPI, and hilar node metastasis was 94.0%, 98.9%, and 99.6%, respectively. Inaccurate intraoperative diagnosis of LPA, VPI, or hilar node metastasis did not lead to inaccurate prediction of node-negative status.ConclusionsThis is the first prospective trial validating the specific mediastinal LN metastasis pattern in cT1N0 invasive NSCLC, which provides important evidence for clinical applications of selective LN dissection strategy.     

结肠癌患者全结肠系膜切除术后的生存状况及预后影响因素分析

目的探讨结肠癌患者全结肠系膜切除术后的生存状况及预后影响因素。方法选取行全结肠系膜切除术治疗结肠癌患者86例,收集患者性别、年龄、Dukes分期、病理类型等信息,分析全结肠系膜切除术治疗结肠癌患者的负性情绪[焦虑自评量表(SAS)、抑郁自评量表(SDS)]及生存质量评价量表(FACT-G)、术后5年生存状况和影响因素。结果术后3个月、术后6个月的SAS、SDS评分较术前低,FACT-G评分较术前高(P<0.05)。结肠系膜切除术治疗结肠癌患者术后3年OS、PFS分别为94.19%(81/86)、88.37%(76/86),5年内OS、PFS分别为76.74%(66/86)、73.26%(63/86)。年龄、Dukes分期、病理类型、肿瘤病灶位置、切除平面分级、淋巴结转移、淋巴结清扫数目、手术入路、术中出血量是影响结肠癌患者全结肠系膜切除术后生存状况及预后的单因素(P<0.05);logistic回归分析显示,年龄>60岁、Dukes分期(C期)、病理类型(未分化癌、腺鳞癌、黏液腺癌)、肿瘤病灶位置(右半结肠)、淋巴结转移、术中出血量≥200 ml是结肠癌患者全结肠系膜切除术后生存状况及预后的危险因素,手术入路(中间)、淋巴结清扫数目≥12、切除平面分级(优)是结肠癌患者全结肠系膜切除术后生存状况及预后的保护因素(P<0.05)。结论影响结肠癌患者全结肠系膜切除术后远期生存状况的因素较多,临床可依据其影响因素进行针对性防控,以改善预后。     

Metachronous hepatic resection for liver only pancreatic metastases

BackgroundThe value of liver resection (LR) for metachronous pancreatic ductal adenocarcinoma (PDAC) metastases remains controversial. However, in light of increasing safety of liver resections, surgery might be a valuable option for metastasized PDAC in selected patients.MethodsWe performed a retrospective, multicenter study including patients undergoing hepatectomy for metachronous PDAC liver metastases between 2004 and 2015 to analyze postoperative outcome and overall survival. All patients were operated with curative intent. Patients with oligometastatic metachronous liver metastasis with definitive chemotherapy (n = 8) served as controls.ResultsOverall 25 patients in seven centers were included in this study. The median age at the time of LR was 63.8 years (56.9–69.9) and the median number of metastases in the liver was 1 (IQR 1–2). There were eight non-anatomical resections (32%), 15 anatomical minor (60%) and 2 major LR (8%). Postoperative complications occurred in eleven patients (eight Clavien-Dindo grade I complications (32%) and three grade IIIa complications (12%), respectively). The 30-day mortality was 0%. The median length of stay was 8.6 days (IQR 5–11). Median overall survival following LR was 36.8 months compared to 9.2 months in patients with metachronous liver metastasis with chemotherapy (p = 0007).DiscussionLiver resection for metachronous PDAC metastasis is safe and feasible in selected patients. To address general applicability and to find factors for patient selection, larger trials are urgently warranted.     

Prediction and Detection of Cervical Malignancy Using Machine Learning Models

Objective: Human papillomavirus and other predicting factors are responsible causing cervical cancer, and early prediction and diagnosis is the solution for preventing this condition.  The objective is to find out and analyze the predictors of cervical cancer and to study the issues of unbalanced datasets using various Machine Learning (ML) algorithm-based models. Methods: A multi-stage sampling strategy was used to recruit 501 samples for the study. The educational intervention was the video-assisted counseling which is consisted of two educational methods: a documentary film and face-to- face interaction with women followed by reminders. Following the collection of baseline data from these subjects, they were encouraged to undergo Pap smear screening. Women having abnormal Pap tests were sent for biopsy. Machine learning classification methods such as Decision Tree (DT), Random Forest (RF), Logistic Regression (LR), Multi-layer Perceptron (MLP) and Naive Bayes(NB) were used to evaluate the unbalanced input and target datasets. Result: Merely 398 women out of 501 showed an interest to participate in the study, but only 298 stated a willingness for cervical screening. Atypical malignant cells were discovered on the cervix of 26 women who had abnormal pap tests. These women had guided for further tests, such as a cervical biopsy, and seven women had been diagnosed with cervical cancer. LR in models 1, 2, and 4 showed 88% to 94% sensitivity with 84% to 89% accuracy, respectively for cervical cancer prediction, whereas DT in models 3, 5, and 6 algorithms exhibited 83% to 84% sensitivity with 84% to 88% accuracy, respectively. The NB and LR algorithms produced the highest area under the ROC curve for testing dataset, but all models performed similarly for training data. Conclusion: In current study , Logistic Regression and Decision Tree algorithms were identified as the best-performed ML algorithm classifiers to detect the significant predictors.     

Can we predict trial failure among older adult-specific clinical trials using trial-level factors?

IntroductionConducting older adult-specific clinical trials can help overcome the lack of clinical evidence for older adults due to their underrepresentation in clinical trials. Understanding factors contributing to the successful completion of such trials can help trial sponsors and researchers prioritize studies and optimize study design. We aimed to develop a model that predicts trial failure among older adult-specific cancer clinical trials using trial-level factors.Materials and methodsWe identified phase 2–4 interventional cancer clinical trials that ended between 2008 and 2019 and had the minimum age limit of 60 years old or older using Aggregate Analysis of ClinicalTrials.gov data. We defined trial failure as closed early for reasons other than interim results or toxicity or completed with a sample of <85% of the targeted size. Candidate trial-level predictors were identified from a literature review. We evaluated eight types of machine learning algorithms to find the best model. Model fitting and testing were performed using 5-fold nested cross-validation. We evaluated the model performance using the area under receiver operating characteristic curve (AUROC).ResultsOf 209 older adult-specific clinical trials, 87 were failed trials per the definition of trial failure. The model with the highest AUROC in the validation set was the least absolute shrinkage and selection operator (AUROC in the test set = 0.70; 95% confidence interval [CI]: 0.53, 0.86). Trial-level factors included in the best model were the study sponsor, the number of participating centers, the number of modalities, the level of restriction on performance score, study location, the number of arms, life expectancy restriction, and the number of target size. Among these factors, the number of centers (odds ratio [OR] = 0.83, 95% CI: 0.71, 0.94), study being in non-US only vs. US only (OR = 0.32, 95% CI: 0.12, 0.82), and life expectancy restriction (OR = 2.17, 95% CI: 1.04, 4.73) were significantly associated with the trial failure.DiscussionWe identified trial-level factors predictive of trial failure among older adult-specific clinical trials and developed a prediction model that can help estimate the risk of failure before a study is conducted. The study findings could aid in the design and prioritization of future older adult-specific clinical trials.     

Brain Metastasis Growth Kinetics: A Novel Prognosticator for Stereotactic Radiotherapy

AimsThe rate of size change in brain metastasis may have clinical implications on tumour biology and prognosis for patients who receive stereotactic radiotherapy (SRT). We analysed the prognostic value of brain metastasis size kinetics and propose a model for patients with brain metastases treated with linac-based SRT in predicting overall survival.Materials and methodsWe analysed the patients receiving linac-based SRT between 2010 and 2020. Patient and oncological factors, including the changes in sizes of brain metastasis between the diagnostic and stereotactic magnetic resonance imaging, were collected. The associations between prognostic factors and overall survival were assessed using Cox regression with least absolute selection and shrinkage operator (LASSO) checked by 500 bootstrap replications. Our prognostic score was calculated by evaluating the most statistically significant factors. Patients were grouped and compared according to our proposed score, Score Index for Radiosurgery in Brain Metastases (SIR) and Basic Score for Brain Metastases (BS-BM).ResultsIn total, 85 patients were included. We developed the prognostic model based on the most important predictors of overall survival: growth kinetics, i.e. percentage change in brain metastasis size per day between the diagnostic and stereotactic magnetic resonance imaging (hazard ratio per 1% increase, 1.32; 95% confidence interval 1.06–1.65), extracranial oligometastatic diseases (≤5 involvements) (hazard ratio 0.28; 95% confidence interval 0.16–0.52) and the presence of neurological symptoms (hazard ratio 2.99; 95% confidence interval 1.54–5.81). Patients with scores 0, 1, 2 and 3 had a median overall survival of 44.4 (95% confidence interval 9.6–not reached), 20.4 (95% confidence interval 15.6–40.8), 12.0 (95% confidence interval 7.2–22.8) and 2.4 (95% confidence interval 1.2–not reached) years, respectively. The optimism-corrected c-indices for our proposed model, SIR and BS-BM were 0.65, 0.58 and 0.54, respectively.ConclusionsBrain metastasis growth kinetics is a valuable metric for survival outcomes of SRT. Our model is useful in identifying patients with brain metastasis treated with SRT with different overall survival.     

Results of European pooled analysis of IORT-containing multimodality treatment for locally advanced rectal cancer: adjuvant chemotherapy prevents local recurrence rather than distant metastases

BackgroundThe purpose of this study is to analyze the pooled results of multimodality treatment of locally advanced rectal cancer (LARC) in four major treatment centers with particular expertise in intraoperative radiotherapy (IORT).Patients and methodsA total of 605 patients with LARC who underwent multimodality treatment up to 2005 were studied. The basic treatment principle was preoperative (chemo)radiotherapy, intended radical surgery, IORT and elective adjuvant chemotherapy (aCT). In uni- and multivariate analyses, risk factors for local recurrence (LR), distant metastases (DM) and overall survival (OS) were studied.ResultsChemoradiotherapy lead to more downstaging and complete remissions than radiotherapy alone (P < 0.001). In all, 42% of the patients received aCT, independent of tumor–node–metastasis stage or radicality of the resection. LR rate, DM rate and OS were 12.0%, 29.2% and 67.1%, respectively. Risk factors associated with LR were no downstaging, lymph node (LN) positivity, margin involvement and no postoperative chemotherapy. Male gender, preoperatively staged T4 disease, no downstaging, LN positivity and margin involvement were associated with a higher risk for DM. A risk model was created to determine a prognostic index for individual patients with LARC.ConclusionsOverall oncological results after multimodality treatment of LARC are promising. Adding aCT to the treatment can possibly improve LR rates.