Prediction models for patients with esophageal or gastric cancer: A systematic review and meta-analysis

International External Validation of Risk Prediction Model of 90-Day Mortality after Gastrectomy for Cancer Using Machine Learning

BACKGROUND

Radical gastrectomy remains the main treatment for gastric cancer, despite its high mortality. A clinical predictive model of 90-day mortality (90DM) risk after gastric cancer surgery based on the Spanish EURECCA registry database was developed using a matching learning algorithm. We performed an external validation of this model based on data from an international multicenter cohort of patients.

METHODS

A cohort of patients from the European GASTRODATA database was selected. Demographic, clinical, and treatment variables in the original and validation cohorts were compared. The performance of the model was evaluated using the area under the curve (AUC) for a random forest model.

RESULTS

The validation cohort included 2546 patients from 24 European hospitals. The advanced clinical T- and N-category, neoadjuvant therapy, open procedures, total gastrectomy rates, and mean volume of the centers were significantly higher in the validation cohort. The 90DM rate was also higher in the validation cohort (5.6%) vs. the original cohort (3.7%). The AUC in the validation model was 0.716.

CONCLUSION

The externally validated model for predicting the 90DM risk in gastric cancer patients undergoing gastrectomy with curative intent continues to be as useful as the original model in clinical practice.

Risk prediction model for gastric cancer within 5 years in healthy Korean adults

BACKGROUND

Although endoscopy is commonly used for gastric cancer screening in South Korea, predictive models that integrate endoscopy results are scarce. We aimed to develop a 5-year gastric cancer risk prediction model using endoscopy results as a predictor.

METHODS

We developed a predictive model using the cohort data of the Kangbuk Samsung Health Study from 2011 to 2019. Among the 260,407 participants aged ≥20 years who did not have any previous history of cancer, 435 cases of gastric cancer were observed. A Cox proportional hazard regression model was used to evaluate the predictors and calculate the 5-year risk of gastric cancer. Harrell's C-statistics and Nam-D'Agostino χ2 test were used to measure the quality of discrimination and calibration ability, respectively.

RESULTS

We included age, sex, smoking status, alcohol consumption, family history of cancer, and previous results for endoscopy in the risk prediction model. This model showed sufficient discrimination ability [development cohort: C-Statistics: 0.800, 95% confidence interval (CI) 0.770-0.829; validation cohort: C-Statistics: 0.799, 95% CI 0.743-0.856]. It also performed well with effective calibration (development cohort: χ2 = 13.65, P = 0.135; validation cohort: χ2 = 15.57, P = 0.056).

CONCLUSION

Our prediction model, including young adults, showed good discrimination and calibration. Furthermore, this model considered a fixed time interval of 5 years to predict the risk of developing gastric cancer, considering endoscopic results. Thus, it could be clinically useful, especially for adults with endoscopic results.

Stroke risk prediction models: A systematic review and meta-analysis

BACKGROUND

Prediction algorithms/models are viable methods for identifying individuals at high risk of stroke across diverse populations for timely intervention. However, evidence summarizing the performance of these models is limited. This study examined the performance and weaknesses of existing stroke risk-score-prediction models (SRSMs) and whether performance varied by population and region.

METHODS

PubMed, EMBASE, and Web of Science were searched for articles on SRSMs from the earliest records until February 2022. The Prediction Model Risk of Bias Assessment Tool was used to assess the quality of eligible articles. The performance of the SRSMs was assessed by meta-analyzing C-statistics (0 and 1) estimates from identified studies to determine the overall pooled C-statistics by fitting a linear restricted maximum likelihood in a random effect model.

RESULTS

Overall, 17 articles (cohort study = 15, nested case-control study = 2) comprising 739,134 stroke cases from 6,396,594 participants from diverse populations/regions (Asia; n = 8, United States; n = 3, and Europe and the United Kingdom; n = 6) were eligible for inclusion. The overall pooled c-statistics of SRSMs was 0.78 (95%CI: 0.75, 0.80; I2 = 99.9%), with most SRSMs developed using cohort studies; 0.78 (95%CI: 0.75, 0.80; I2 = 99.9%). The subgroup analyses by geographical region: Asia [0.81 (95%CI: 0.79, 0.83; I2 = 99.8%)], Europe and the United Kingdom [0.76 (95%CI: 0.69, 0.83; I2 = 99.9%)] and the United States only [0.75 (95%CI: 0.72, 0.78; I2 = 73.5%)] revealed relatively indifferent performances of SRSMs.

CONCLUSION

SRSM performance varied widely, and the pooled c-statistics of SRSMs suggested a fair predictive performance, with very few SRSMs validated in independent population group(s) from diverse world regions.

Prediction of Morbidity and Mortality After Esophagectomy: A Systematic Review

BACKGROUND

Esophagectomy for esophageal cancer has a complication rate of up to 60%. Prediction models could be helpful to preoperatively estimate which patients are at increased risk of morbidity and mortality. The objective of this study was to determine the best prediction models for morbidity and mortality after esophagectomy and to identify commonalities among the models.

PATIENTS AND METHODS

A systematic review was performed in accordance to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement and was prospectively registered in PROSPERO ( https://www.crd.york.ac.uk/prospero/ , study ID CRD42022350846). Pubmed, Embase, and Clarivate Analytics/Web of Science Core Collection were searched for studies published between 2010 and August 2022. The Prediction model Risk of Bias Assessment Tool was used to assess the risk of bias. Extracted data were tabulated and a narrative synthesis was performed.

RESULTS

Of the 15,011 articles identified, 22 studies were included using data from tens of thousands of patients. This systematic review included 33 different models, of which 18 models were newly developed. Many studies showed a high risk of bias. The prognostic accuracy of models differed between 0.51 and 0.85. For most models, variables are readily available. Two models for mortality and one model for pulmonary complications have the potential to be developed further.

CONCLUSIONS

The availability of rigorous prediction models is limited. Several models are promising but need to be further developed. Some models provide information about risk factors for the development of complications. Performance status is a potential modifiable risk factor. None are ready for clinical implementation.

Accounting for time dependency in meta-analyses of concordance probability estimates

Recent years have seen the development of many novel scoring tools for disease prognosis and prediction. To become accepted for use in clinical applications, these tools have to be validated on external data. In practice, validation is often hampered by logistical issues, resulting in multiple small-sized validation studies. It is therefore necessary to synthesize the results of these studies using techniques for meta-analysis. Here we consider strategies for meta analyzing the concordance probability for time-to-event data ("C-index"), which has become a popular tool to evaluate the discriminatory power of prediction models with a right-censored outcome. We show that standard meta-analysis of the C-index may lead to biased results, as the magnitude of the concordance probability depends on the length of the time interval used for evaluation (defined e.g., by the follow-up time, which might differ considerably between studies). To address this issue, we propose a set of methods for random-effects meta-regression that incorporate time directly as covariate in the model equation. In addition to analyzing nonlinear time trends via fractional polynomial, spline, and exponential decay models, we provide recommendations on suitable transformations of the C-index before meta-regression. Our results suggest that the C-index is best meta-analyzed using fractional polynomial meta-regression with logit-transformed C-index values. Classical random-effects meta-analysis (not considering time as covariate) is demonstrated to be a suitable alternative when follow-up times are small. Our findings have implications for the reporting of C-index values in future studies, which should include information on the length of the time interval underlying the calculations.

Prognostic value of patient reported outcomes in advanced gastro-oesophageal cancer: a systematic review

To summarise the prognostic value of patient-reported outcomes (PROs) in advanced gastro-oesophageal (GO) cancer. We systematically searched multiple databases using search terms related to advanced GO cancer, PRO and prognosis. Studies examining the relationship between baseline PROs and prognosis were included. Two reviewers independently screened articles and extracted data on study design, survival and associations between PROs and survival, in both univariable and multivariable analyses. QUIPS was used for quality assessment. From 3004 studies screened, seven studies were eligible, comprising PRO data from 2761 of 3408 (81%) participants. Median survival times ranged from 4.5 to 9.5 months. Among participants with oesophageal squamous cell carcinoma (SCC), physical functioning, social functioning and fatigue (QLQ-C30) were associated with overall survival (OS) in one univariable analysis. Among three studies of participants with adenocarcinoma, univariable analyses revealed associations between OS and global quality of life (QOL), physical functioning, role functioning and social functioning; two studies showed association with pain. There was an association between emotional functioning, fatigue, lack of mobility, lack of self-care, appetite loss/anorexia and OS in one study. One multivariable analysis among participants with oesophageal SCC showed physical and social functioning was associated with OS. Among participants with adenocarcinoma, multivariable analyses showed associations between OS and physical functioning/lack of mobility, appetite loss/anorexia (three studies), global QOL, role functioning/lack of self-care, pain (two studies) and social functioning (one study). Physical functioning, role functioning, social functioning, pain, anorexia and global QOL were associated with OS in advanced GO cancer.

Effect of a prediction tool and communication skills training on communication of treatment outcomes: a multicenter stepped wedge clinical trial (the SOURCE trial)

Background

For cancer patients to effectively engage in decision making, they require comprehensive and understandable information regarding treatment options and their associated outcomes. We developed an online prediction tool and supporting communication skills training to assist healthcare providers (HCPs) in this complex task. This study aims to assess the impact of this combined intervention (prediction tool and training) on the communication practices of HCPs when discussing treatment options.

Methods

We conducted a multicenter intervention trial using a pragmatic stepped wedge design (NCT04232735). Standardized Patient Assessments (simulated consultations) using cases of esophageal and gastric cancer patients, were performed before and after the combined intervention (March 2020 to July 2022). Audio recordings were analyzed using an observational coding scale, rating all utterances of treatment outcome information on the primary outcome-precision of provided outcome information-and on secondary outcomes-such as: personalization, tailoring and use of visualizations. Pre vs. post measurements were compared in order to assess the effect of the intervention.

Findings

31 HCPs of 11 different centers in the Netherlands participated. The tool and training significantly affected the precision of the overall communicated treatment outcome information (p = 0.001, median difference 6.93, IQR (-0.32 to 12.44)). In the curative setting, survival information was significantly more precise after the intervention (p = 0.029). In the palliative setting, information about side effects was more precise (p < 0.001).

Interpretation

A prediction tool and communication skills training for HCPs improves the precision of treatment information on outcomes in simulated consultations. The next step is to examine the effect of such interventions on communication in clinical practice and on patient-reported outcomes.

Funding

Financial support for this study was provided entirely by a grant from the Dutch Cancer Society (UVA 2014-7000).

Combination of clinical, radiomic, and "delta" radiomic features in survival prediction of metastatic gastroesophageal adenocarcinoma

Objectives

To identify combined clinical, radiomic, and delta-radiomic features in metastatic gastroesophageal adenocarcinomas (GEAs) that may predict survival outcomes.

Methods

A total of 166 patients with metastatic GEAs on palliative chemotherapy with baseline and treatment/follow-up (8-12 weeks) contrast-enhanced CT were retrospectively identified. Demographic and clinical data were collected. Three-dimensional whole-lesional radiomic analysis was performed on the treatment/follow-up scans. "Delta" radiomic features were calculated based on the change in radiomic parameters compared to the baseline. The univariable analysis (UVA) Cox proportional hazards model was used to select clinical variables predictive of overall survival (OS) and progression-free survival (PFS) (p-value <0.05). The radiomic and "delta" features were then assessed in a multivariable analysis (MVA) Cox model in combination with clinical features identified on UVA. Features with a p-value <0.01 in the MVA models were selected to assess their pairwise correlation. Only non-highly correlated features (Pearson's correlation coefficient <0.7) were included in the final model. Leave-one-out cross-validation method was used, and the 1-year area under the receiver operating characteristic curve (AUC) was calculated for PFS and OS.

Results

Of the 166 patients (median age of 59.8 years), 114 (69%) were male, 139 (84%) were non-Asian, and 147 (89%) had an Eastern Cooperative Oncology Group (ECOG) performance status of 0-1. The median PFS and OS on treatment were 3.6 months (95% CI 2.86, 4.63) and 9 months (95% CI 7.49, 11.04), respectively. On UVA, the number of chemotherapy cycles and number of lesions at the end of treatment were associated with both PFS and OS (p < 0.001). ECOG status was associated with OS (p = 0.0063), but not PFS (p = 0.054). Of the delta-radiomic features, delta conventional HUmin, delta gray-level zone length matrix (GLZLM) GLNU, and delta GLZLM LGZE were incorporated into the model for PFS, and delta shape compacity was incorporated in the model for OS. Of the treatment/follow-up radiomic features, shape compacity and neighborhood gray-level dependence matrix (NGLDM) contrast were used in both models. The combined 1-year AUC (Kaplan-Meier estimator) was 0.82 and 0.81 for PFS and OS, respectively.

Conclusions

A combination of clinical, radiomics, and delta-radiomic features may predict PFS and OS in GEAs with reasonable accuracy.

SOURCE beyond first-line: A survival prediction model for patients with metastatic esophagogastric adenocarcinoma after failure of first-line palliative systemic therapy

Prior models have been developed to predict survival for patients with esophagogastric cancer undergoing curative treatment or first-line chemotherapy (SOURCE models). Comprehensive clinical prediction models for patients with esophagogastric cancer who will receive second-line chemotherapy or best supportive care are currently lacking. The aim of our study was to develop and internally validate a new clinical prediction model, called SOURCE beyond first-line, for survival of patients with metastatic esophagogastric adenocarcinoma after failure of first-line palliative systemic therapy. Patients with unresectable or metastatic esophageal or gastric adenocarcinoma (2015-2017) who received first-line systemic therapy (N = 1067) were selected from the Netherlands Cancer Registry. Patient, tumor and treatment characteristics at primary diagnosis and at progression of disease were used to develop the model. A Cox proportional hazards regression model was developed through forward and backward selection using Akaike's Information Criterion. The model was internally validated through 10-fold cross-validations to assess performance. Model discrimination (C-index) and calibration (slope and intercept) were used to evaluate performance of the complete and cross-validated models. The final model consisted of 11 patient tumor and treatment characteristics. The C-index was 0.75 (0.73-0.78), calibration slope 1.01 (1.00-1.01) and calibration intercept 0.01 (0.01-0.02). Internal cross-validation of the model showed that the model performed adequately on unseen data: C-index was 0.79 (0.77-0.82), calibration slope 0.93 (0.85-1.01) and calibration intercept 0.02 (-0.01 to 0.06). The SOURCE beyond first-line model predicted survival with fair discriminatory ability and good calibration.

The AUGIS Survival Predictor: Prediction of Long-Term and Conditional Survival After Esophagectomy Using Random Survival Forests

OBJECTIVE

The aim of this study was to develop a predictive model for overall survival after esophagectomy using pre/postoperative clinical data and machine learning.

SUMMARY BACKGROUND DATA

For patients with esophageal cancer, accurately predicting long-term survival after esophagectomy is challenging. This study investigated survival prediction after esophagectomy using a RandomSurvival Forest (RSF) model derived from routine data from a large, well-curated, national dataset.

METHODS

Patients diagnosed with esophageal adenocarcinoma or squamous cell carcinoma between 2012 and 2018 in England and Wales who underwent an esophagectomy were included. Prediction models for overall survival were developed using the RSF method and Cox regression from 41 patient and disease characteristics. Calibration and discrimination (time-dependent area under the curve) were validated internally using bootstrap resampling.

RESULTS

The study analyzed 6399 patients, with 2625 deaths during follow-up. Median follow-up was 41 months. Overall survival was 47.1% at 5 years. The final RSF model included 14 variables and had excellent discrimination with a 5-year time-dependent area under the receiver operator curve of 83.9% [95% confidence interval (CI) 82.6%-84.9%], compared to 82.3% (95% CI 81.1%-83.3%) for the Cox model. The most important variables were lymph node involvement, pT stage, circumferential resection margin involvement (tumor at < 1 mm from cut edge) and age. There was a wide range of survival estimates even within TNM staging groups, with quintiles of prediction within Stage 3b ranging from 12.2% to 44.7% survival at 5 years.

CONCLUSIONS

An RSF model for long-term survival after esophagectomy exhibited excellent discrimination and well-calibrated predictions. At a patient level, it provides more accuracy than TNM staging alone and could help in the delivery of tailored treatment and follow-up.

Methodological guidance for the evaluation and updating of clinical prediction models: a systematic review

BACKGROUND

Clinical prediction models are often not evaluated properly in specific settings or updated, for instance, with information from new markers. These key steps are needed such that models are fit for purpose and remain relevant in the long-term. We aimed to present an overview of methodological guidance for the evaluation (i.e., validation and impact assessment) and updating of clinical prediction models.

METHODS

We systematically searched nine databases from January 2000 to January 2022 for articles in English with methodological recommendations for the post-derivation stages of interest. Qualitative analysis was used to summarize the 70 selected guidance papers.

RESULTS

Key aspects for validation are the assessment of statistical performance using measures for discrimination (e.g., C-statistic) and calibration (e.g., calibration-in-the-large and calibration slope). For assessing impact or usefulness in clinical decision-making, recent papers advise using decision-analytic measures (e.g., the Net Benefit) over simplistic classification measures that ignore clinical consequences (e.g., accuracy, overall Net Reclassification Index). Commonly recommended methods for model updating are recalibration (i.e., adjustment of intercept or baseline hazard and/or slope), revision (i.e., re-estimation of individual predictor effects), and extension (i.e., addition of new markers). Additional methodological guidance is needed for newer types of updating (e.g., meta-model and dynamic updating) and machine learning-based models.

CONCLUSION

Substantial guidance was found for model evaluation and more conventional updating of regression-based models. An important development in model evaluation is the introduction of a decision-analytic framework for assessing clinical usefulness. Consensus is emerging on methods for model updating.

Predicting mortality in patients with anastomotic leak after esophagectomy: development of a prediction model using data from the TENTACLE-Esophagus study

Anastomotic leak (AL) is a common but severe complication after esophagectomy, and over 10% of patients with AL suffer mortality. Different prognostic factors in patients with AL are known, but a tool to predict mortality after AL is lacking. This study aimed to develop a prediction model for postoperative mortality in patients with AL after esophagectomy. TENTACLE-Esophagus is an international retrospective cohort study, which included 1509 patients with AL after esophagectomy. The primary outcome was 90-day postoperative mortality. Previously identified prognostic factors for mortality were selected as predictors: patient-related (e.g. comorbidity, performance status) and leak-related predictors (e.g. leucocyte count, overall gastric conduit condition). The prediction model was developed using multivariable logistic regression and validated internally using bootstrapping. Among the 1509 patients with AL, 90-day mortality was 11.7%. Sixteen predictors were included in the prediction model. The model showed good performance after internal validation: the c-index was 0.79 (95% confidence interval 0.75-0.83). Predictions for mortality by the internally validated model aligned well with observed 90-day mortality rates. The prediction model was incorporated in an online tool for individual use and can be found at: https://www.tentaclestudy.com/prediction-model. The developed prediction model combines patient-related and leak-related factors to accurately predict postoperative mortality in patients with AL after esophagectomy. The model is useful for clinicians during counselling of patients and their families and may aid identification of high-risk patients at diagnosis of AL. In the future, the tool may guide clinical decision-making; however, external validation of the tool is warranted.

Identification of a chemotherapy-associated gene signature for a risk model of prognosis in gastric adenocarcinoma through bioinformatics analysis

BACKGROUND

Over the past few years, the overall survival rate of patients with gastric adenocarcinoma who have received different chemotherapy regimens has increased. However, not all gastric cancer patients who receive chemotherapy have a longer survival. We need better predictive biomarkers. This study is to construct a new risk model of chemotherapy-associated genes in gastric adenocarcinoma (GA) for prognostication.

METHODS

RNA-seq data and clinical information of GSE26901 (containing 44 chemotherapy samples and 65 patients without chemotherapy) in Gene Expression Omnibus (GEO) and stomach adenocarcinoma (STAD, containing 360 cancer tissue samples and 50 paired normal tissue samples) in The Cancer Genome Atlas (TCGA) were selected for screening differentially expressed genes (DEGs). Multivariate Cox regression was conducted to screen prognosis-associated genes and its link to patients' prognosis were screened by least absolute shrinkage and selection operator (LASSO) regression analysis. Based on the key genes, a risk scoring equation for the prognosis model was established, and constructed survival prognosis model. The model was tested for predictive ability through training set (TCGA datasets) and validation set (GSE84437). The correlations of the risk score with clinical pathological features, immune score and drug sensitivity score were evaluated.

RESULTS

In total, 179 overlapping genes were obtained by screening DEGs. Univariate Cox analysis revealed 36 prognosis-related genes, and LASSO regression analysis revealed 8 key genes (KCNJ2, GATA5, CLDN1, SERPINE1, FCER2, PMEPA1, TMEM37 and CRTAC1). Kaplan-Meier (K-M) analysis uncovered a relatively short overall survival time in the high-risk group. The model was verified to possess favourable predictive ability. In addition, the nomogram model were demonstrated good predictability with area under the curve (AUC) for 1-5 years in training set were 0.78, 0.78, 0.76, 0.79 and 0.81. The high-risk group was less likely to get benefits from immunotherapy and less sensitive to cisplatin.

CONCLUSIONS

According to the results of our training set and validation set, the risk model based on the eight chemotherapy-related gene signatures predicting prognosis has certain predictive accuracy in predicting the survival of GA patients which can be a promising prognostic parameter for GA. However, its efficacy remains to be proved in clinical practice.

Use and performance of machine learning models for type 2 diabetes prediction in clinical and community care settings: Protocol for a systematic review and meta-analysis of predictive modeling studies

Objective

Machine learning involves the use of algorithms without explicit instructions. Of late, machine learning models have been widely applied for the prediction of type 2 diabetes. However, no evidence synthesis of the performance of these prediction models of type 2 diabetes is available. We aim to identify machine learning prediction models for type 2 diabetes in clinical and community care settings and determine their predictive performance.

Methods

The systematic review of English language machine learning predictive modeling studies in 12 databases will be conducted. Studies predicting type 2 diabetes in predefined clinical or community settings are eligible. Standard CHARMS and TRIPOD guidelines will guide data extraction. Methodological quality will be assessed using a predefined risk of bias assessment tool. The extent of validation will be categorized by Reilly–Evans levels. Primary outcomes include model performance metrics of discrimination ability, calibration, and classification accuracy. Secondary outcomes include candidate predictors, algorithms used, level of validation, and intended use of models. The random-effects meta-analysis of c-indices will be performed to evaluate discrimination abilities. The c-indices will be pooled per prediction model, per model type, and per algorithm. Publication bias will be assessed through funnel plots and regression tests. Sensitivity analysis will be conducted to estimate the effects of study quality and missing data on primary outcome. The sources of heterogeneity will be assessed through meta-regression. Subgroup analyses will be performed for primary outcomes.

Ethics and dissemination

No ethics approval is required, as no primary or personal data are collected. Findings will be disseminated through scientific sessions and peer-reviewed journals.

PROSPERO registration number

CRD42019130886

Women's Suicide in the First-Year Postpartum: A Population-based Study

BACKGROUND

In dominant suicidology there is a long traditionof theorizing that women are protected from suicide, particularly when mothers and during the postpartum. Studies have mostly confirmed the postpartum suicide-protection theory, with low suicide mortality generally observed among postpartum women. A limitation of these studies is that most were conducted in majority European-descent-population countries. A challenge to the more general maternal suicide-protection theory is that in East-Asia women of childbearing age exhibit substantial suicidality, nonfatal and fatal. This study evaluated whether suicide is less likely in first-year postpartum women as compared to women past the first-year postpartum.

METHODS

This population-based, nested case-control study focused on women whose live birth was between 2001 and 2016 in East-Asian Taiwan. To ascertain suicide outcomes, the women were followed until 2017. For each suicide case, four control cases were randomly selected from the Birth Certificate Application dataset, with a 1:4 matching ratio based on age of last live-delivery and parity (one delivery record vs. two or more records) (cases N = 1571; controls N = 6284). Conditional logistic regression analyses were conducted to assess whether suicide was less likely in women in the first-year postpartum relative to women past the first-year postpartum.

RESULTS

The odds ratios of suicide were elevated at 42 days postpartum [Odds Ratio (OR) = 2.06; 95% Confidence Interval (CI) = (1.04, 4.16)], six-months postpartum [OR = 2.28; 95% CI = (1.60, 3.29)] and one-year postpartum [OR = 2.26; 95% CI = (1.76, 2.96)], when controlling for sociodemographic and mental-disorder variables. Suicide was more likely in women who were single at index birth, had lower socioeconomic status, or had a mental disorder history.

CONCLUSION

Our findings suggest that the postpartum stage is not suicide-protective per se. Whether the postpartum stage is associated with suicide protection or suiciderisk appears to depend on context and culture.

Development and application of a dynamic prediction model for esophageal cancer

Background

Current prediction models of esophageal cancer (EC) are limited to predicting at a specific time point, and ignore changes in hazard ratios of predictive variables, known as time-varying effects. Our study aimed to investigate variables with time-varying effects in EC and to develop a prediction model that can update the 5-year predicted dynamic overall survival (DOS) probability during the follow-up period.

Methods

Firstly, the clinicopathological information and survival data of 4,541 patients with EC was obtained from the Surveillance, Epidemiology, and End Results (SEER) database between 2007 and 2011 for modeling. Secondly, the time-varying effect of variables was assessed and the dynamic prediction model was developed based on the proportional baselines landmark supermodel.

Results

Here, we found that age at diagnosis, sex, location of primary tumor, histological type, chemotherapy, surgery, and T stage showed significant time-varying effects on overall survival. Thirdly, the prediction model was validated by an internal SEER validation cohort and a Chinese patient cohort, respectively, and achieved promising results as follows: area under the curve (AUC) =0.733 (internal validation) and 0.864 (external validation). The heuristic shrinkage factor was 0.995. Finally, several clear cases were selected as examples for model application to map the patient’s 5-year DOS curves and to respectively demonstrate the impact of different variables’ time-varying effect on survival.

Conclusions

Overall, our results suggest that the existence of time-varying effect highlights the importance of updating the predicted survival probability during the follow-up period. Moreover, this prediction model can be used to assist doctors in making more-individualized treatment decisions based on a dynamic assessment of patient prognosis.

Prediction of long-term survival after gastrectomy using random survival forests

BACKGROUND

No well validated and contemporaneous tools for personalized prognostication of gastric adenocarcinoma exist. This study aimed to derive and validate a prognostic model for overall survival after surgery for gastric adenocarcinoma using a large national dataset.

METHODS

National audit data from England and Wales were used to identify patients who underwent a potentially curative gastrectomy for adenocarcinoma of the stomach. A total of 2931 patients were included and 29 clinical and pathological variables were considered for their impact on survival. A non-linear random survival forest methodology was then trained and validated internally using bootstrapping with calibration and discrimination (time-dependent area under the receiver operator curve (tAUC)) assessed.

RESULTS

The median survival of the cohort was 69 months, with a 5-year survival of 53.2 per cent. Ten variables were found to influence survival significantly and were included in the final model, with the most important being lymph node positivity, pT stage and achieving an R0 resection. Patient characteristics including ASA grade and age were also influential. On validation the model achieved excellent performance with a 5-year tAUC of 0.80 (95 per cent c.i. 0.78 to 0.82) and good agreement between observed and predicted survival probabilities. A wide spread of predictions for 3-year (14.8-98.3 (i.q.r. 43.2-84.4) per cent) and 5-year (9.4-96.1 (i.q.r. 31.7-73.8) per cent) survival were seen.

CONCLUSIONS

A prognostic model for survival after a potentially curative resection for gastric adenocarcinoma was derived and exhibited excellent discrimination and calibration of predictions.

Artificial Intelligence in Gastric Cancer: Identifying Gastric Cancer Using Endoscopic Images with Convolutional Neural Network

Simple Summary

Gastric cancer (GC) is one of the most newly diagnosed cancers and the fifth leading cause of death globally. Previous studies reported that the detection rate of gastric cancer (EGC) at an earlier stage is low, and the overall false-negative rate with esophagogastroduodenoscopy (EGD) is up to 25.8%, which often leads to inappropriate treatment. Accurate diagnosis of EGC can reduce unnecessary interventions and benefits treatment planning. Convolutional neural network (CNN) models have recently shown promising performance in analyzing medical images, including endoscopy. This study shows that an automated tool based on the CNN model could improve EGC diagnosis and treatment decision.

Abstract

Gastric cancer (GC) is one of the most newly diagnosed cancers and the fifth leading cause of death globally. Identification of early gastric cancer (EGC) can ensure quick treatment and reduce significant mortality. Therefore, we aimed to conduct a systematic review with a meta-analysis of current literature to evaluate the performance of the CNN model in detecting EGC. We conducted a systematic search in the online databases (e.g., PubMed, Embase, and Web of Science) for all relevant original studies on the subject of CNN in EGC published between 1 January 2010, and 26 March 2021. The Quality Assessment of Diagnostic Accuracy Studies-2 was used to assess the risk of bias. Pooled sensitivity, specificity, positive likelihood ratio, negative likelihood ratio, and diagnostic odds ratio were calculated. Moreover, a summary receiver operating characteristic curve (SROC) was plotted. Of the 171 studies retrieved, 15 studies met inclusion criteria. The application of the CNN model in the diagnosis of EGC achieved a SROC of 0.95, with corresponding sensitivity of 0.89 (0.88–0.89), and specificity of 0.89 (0.89–0.90). Pooled sensitivity and specificity for experts endoscopists were 0.77 (0.76–0.78), and 0.92 (0.91–0.93), respectively. However, the overall SROC for the CNN model and expert endoscopists was 0.95 and 0.90. The findings of this comprehensive study show that CNN model exhibited comparable performance to endoscopists in the diagnosis of EGC using digital endoscopy images. Given its scalability, the CNN model could enhance the performance of endoscopists to correctly stratify EGC patients and reduce work load.

Risk Prediction Model of 90-Day Mortality After Esophagectomy for Cancer

Importance

Ninety-day mortality rates after esophagectomy are an indicator of the quality of surgical oncologic management. Accurate risk prediction based on large data sets may aid patients and surgeons in making informed decisions.

Objective

To develop and validate a risk prediction model of death within 90 days after esophagectomy for cancer using the International Esodata Study Group (IESG) database, the largest existing prospective, multicenter cohort reporting standardized postoperative outcomes.

Design, Setting, and Participants

In this diagnostic/prognostic study, we performed a retrospective analysis of patients from 39 institutions in 19 countries between January 1, 2015, and December 31, 2019. Patients with esophageal cancer were randomly assigned to development and validation cohorts. A scoring system that predicted death within 90 days based on logistic regression β coefficients was conducted. A final prognostic score was determined and categorized into homogeneous risk groups that predicted death within 90 days. Calibration and discrimination tests were assessed between cohorts.

Exposures

Esophageal resection for cancer of the esophagus and gastroesophageal junction.

Main Outcomes and Measures

All-cause postoperative 90-day mortality.

Results

A total of 8403 patients (mean [SD] age, 63.6 [9.0] years; 6641 [79.0%] male) were included. The 30-day mortality rate was 2.0% (n = 164), and the 90-day mortality rate was 4.2% (n = 353). Development (n = 4172) and validation (n = 4231) cohorts were randomly assigned. The multiple logistic regression model identified 10 weighted point variables factored into the prognostic score: age, sex, body mass index, performance status, myocardial infarction, connective tissue disease, peripheral vascular disease, liver disease, neoadjuvant treatment, and hospital volume. The prognostic scores were categorized into 5 risk groups: very low risk (score, ≥1; 90-day mortality, 1.8%), low risk (score, 0; 90-day mortality, 3.0%), medium risk (score, -1 to -2; 90-day mortality, 5.8%), high risk (score, -3 to -4: 90-day mortality, 8.9%), and very high risk (score, ≤-5; 90-day mortality, 18.2%). The model was supported by nonsignificance in the Hosmer-Lemeshow test. The discrimination (area under the receiver operating characteristic curve) was 0.68 (95% CI, 0.64-0.72) in the development cohort and 0.64 (95% CI, 0.60-0.69) in the validation cohort.

Conclusions and Relevance

In this study, on the basis of preoperative variables, the IESG risk prediction model allowed stratification of an individual patient's risk of death within 90 days after esophagectomy. These data suggest that this model can help in the decision-making process when esophageal cancer surgery is being considered and in informed consent.

Down-staging depth score could be a survival predictor for locally advanced gastric cancer patients after preoperative chemoradiotherapy

Objective

The predictive effect of preoperative chemoradiotherapy (CRT) is low and difficult in guiding individualized treatment. We examined a surrogate endpoint for long-term outcomes in locally advanced gastric cancer patients after preoperative CRT.

Methods

From April 2012 to April 2019, 95 patients with locally advanced gastric cancer who received preoperative concurrent CRT and who were enrolled in three prospective studies were included. All patients were stage T_3/4N₊. Local control, distant metastasis-free survival (DMFS), disease-free survival (DFS) and overall survival (OS) were evaluated. Clinicopathological factors related to long-term prognosis were analyzed using univariate and multivariate analyses. The down-staging depth score (DDS), which is a novel method of evaluating CRT response, was used to predict long-term outcomes.

Results

The median follow-up period for survivors was 30 months. The area under the curve (AUC) of the receiver operating characteristic (ROC) curve predicted by the DDS was 0.728, which was better than the pathological complete response (pCR), histological response and ypN0. Decision curve analysis further affirmed that DDS had the largest net benefit. The DDS cut-off value was 4. pCR and ypN0 were associated with OS (P=0.026 and 0.049). Surgery and DDS are correlated with DMFS, DFS and OS (surgery: P=0.001, <0.001 and <0.001, respectively; and DDS: P=0.009, 0.013 and 0.032, respectively). Multivariate analysis showed that DDS was an independent prognostic factor of DFS (P=0.021).

Conclusions

DDS is a simple, short-term indicator that was a better surrogate endpoint than pCR, histological response and ypN0 for DFS.

Informing Patients With Esophagogastric Cancer About Treatment Outcomes by Using a Web-Based Tool and Training: Development and Evaluation Study

Background

Due to the increasing use of shared decision-making, patients with esophagogastric cancer play an increasingly important role in the decision-making process. To be able to make well-informed decisions, patients need to be adequately informed about treatment options and their outcomes, namely survival, side effects or complications, and health-related quality of life. Web-based tools and training programs can aid physicians in this complex task. However, to date, none of these instruments are available for use in informing patients with esophagogastric cancer about treatment outcomes.

Objective

This study aims to develop and evaluate the feasibility of using a web-based prediction tool and supporting communication skills training to improve how physicians inform patients with esophagogastric cancer about treatment outcomes. By improving the provision of treatment outcome information, we aim to stimulate the use of information that is evidence-based, precise, and personalized to patient and tumor characteristics and is communicated in a way that is tailored to individual information needs.

Methods

We designed a web-based, physician-assisted prediction tool—Source—to be used during consultations by using an iterative, user-centered approach. The accompanying communication skills training was developed based on specific learning objectives, literature, and expert opinions. The Source tool was tested in several rounds—a face-to-face focus group with 6 patients and survivors, semistructured interviews with 5 patients, think-aloud sessions with 3 medical oncologists, and interviews with 6 field experts. In a final pilot study, the Source tool and training were tested as a combined intervention by 5 medical oncology fellows and 3 esophagogastric outpatients.

Results

The Source tool contains personalized prediction models and data from meta-analyses regarding survival, treatment side effects and complications, and health-related quality of life. The treatment outcomes were visualized in a patient-friendly manner by using pictographs and bar and line graphs. The communication skills training consisted of blended learning for clinicians comprising e-learning and 2 face-to-face sessions. Adjustments to improve both training and the Source tool were made according to feedback from all testing rounds.

Conclusions

The Source tool and training could play an important role in informing patients with esophagogastric cancer about treatment outcomes in an evidence-based, precise, personalized, and tailored manner. The preliminary evaluation results are promising and provide valuable input for the further development and testing of both elements. However, the remaining uncertainty about treatment outcomes in patients and established habits in doctors, in addition to the varying trust in the prediction models, might influence the effectiveness of the tool and training in daily practice. We are currently conducting a multicenter clinical trial to investigate the impact that the combined tool and training have on the provision of information in the context of treatment decision-making.

Combined detection of peripheral blood VEGF and inflammation biomarkers to evaluate the clinical response and prognostic prediction of non-operative ESCC

An association between angiogenesis/inflammation status and tumor has been reported in various types of cancer. This study sought to assess the role of peripheral blood VEGF and some inflammation biomarkers in evaluating clinical response and prognosis in patients with non-operative esophageal squamous cell carcinoma (ESCC). Peripheral blood of 143 patients with non-operative ESCC at our institute was dynamically collected at 5 time points including 1 day before radiotherapy, during radiotherapy (15f), at the end of radiotherapy, 1 month after radiotherapy, and 3 months after radiotherapy. VEGF expression in the peripheral blood was detected and related inflammation biomarkers such as GPS, CAR and CLR were counted. Logistic regression and Cox regression were implemented respectively to analyze the correlation of each predictor with clinical response and prognosis. The performance of combined testing was estimated using AUCs. Based on independent predictors, a nomogram prediction model was established to predict the probabilities of 1- and 2-year PFS of patients. The effectiveness of the nomogram model was characterized by C-index, AUC, calibration curves and DCA. VEGF and CLR levels at the end of radiotherapy were independent predictors of clinical response, while VEGF and GPS levels at 3 months after radiotherapy were independent prognostic predictors. The efficacy of combined detection of VEGF and CLR is superior to the single detection in evaluating clinical response and prognosis. The nomogram showed excellent accuracy in predicting PFS. The combined detection of VEGF and CLR at the end of radiotherapy can be used to evaluate the clinical response of patients with non-operative ESCC, and the combined detection of VEGF and GPS 3 months after radiotherapy can be used to predict the prognosis. Implemented by nomogram model, it is expected to provide practical and reliable method to evaluate the clinical response and prognosis of patients with non-operative ESCC tool.

Kaplan-Meier Curves, Cox Model, and P-Values Are Not Enough for the Prognostic Evaluation of Tumor Markers: Statistical Suggestions for a More Comprehensive Approach

The assessment of prognostic markers is key to the improvement of therapeutic strategies for cancer patients. Some promising markers may fail to be applied in clinical practice, or some useless markers may be applied, because of misleading results ensuing from inadequate planning of the study and/or from an oversimplified statistical analysis. This commentary illustrates and discusses the main issues involved in planning an effective clinical study and the subsequent statistical analysis for the prognostic evaluation of a cancer marker. Another aim is to extend the most applied statistical models (ie, those using Kaplan-Meier and Cox) to enable the choice of the best-suited methods for study endpoints. Specifically, for tumor-centered endpoints like tumor recurrence, the issue of competing risks is highlighted. For markers measured on a continuous numerical scale, a loss of relevant prognostic information may occur by setting arbitrary cutoffs; thus, the methods to analyze the original scale are explained. Furthermore, because the P-value is not a sufficient criterion to assess the usefulness of a marker in clinical practice, measures for evaluating the ability of the marker to discriminate between "good" and "bad" prognoses are illustrated. Several tumor markers are considered both in human and veterinary medicine. Given the similarity between markers for human breast cancer and canine mammary cancer, an application of the statistical methods discussed within the article to a public dataset from human breast cancer patients is shown.

Models for estimating and projecting global, regional and national prevalence and disease burden of asthma: a systematic review

Background

Statistical models are increasingly being used to estimate and project the prevalence and burden of asthma. Given substantial variations in these estimates, there is a need to critically assess the properties of these models and assess their transparency and reproducibility. We aimed to critically appraise the strengths, limitations and reproducibility of existing models for estimating and projecting the global, regional and national prevalence and burden of asthma.

Methods

We undertook a systematic review, which involved searching Medline, Embase, World Health Organization Library and Information Services (WHOLIS) and Web of Science from 1980 to 2017 for modelling studies. Two reviewers independently assessed the eligibility of studies for inclusion and then assessed their strengths, limitations and reproducibility using pre-defined quality criteria. Data were descriptively and narratively synthesised.

Results

We identified 108 eligible studies, which employed a total of 51 models: 42 models were used to derive national level estimates, two models for regional estimates, four models for global and regional estimates and three models for global, regional and national estimates. Ten models were used to estimate the prevalence of asthma, 27 models estimated the burden of asthma – including, health care service utilisation, disability-adjusted life years, mortality and direct and indirect costs of asthma – and 14 models estimated both the prevalence and burden of asthma. Logistic and linear regression models were most widely used for national estimates. Different versions of the DisMod-MR- Bayesian meta-regression models and Cause Of Death Ensemble model (CODEm) were predominantly used for global, regional and national estimates. Most models suffered from a number of methodological limitations – in particular, poor reporting, insufficient quality and lack of reproducibility.

Conclusions

Whilst global, regional and national estimates of asthma prevalence and burden continue to inform health policy and investment decisions on asthma, most models used to derive these estimates lack the required reproducibility. There is a need for better-constructed models for estimating and projecting the prevalence and disease burden of asthma and a related need for better reporting of models, and making data and code available to facilitate replication.

Cancer associated macrophage-like cells and prognosis of esophageal cancer after chemoradiation therapy

BACKGROUND

Cancer Associated Macrophage-Like cells (CAMLs) are polynucleated circulating stromal cells found in the bloodstream of numerous solid-tumor malignancies. Variations within CAML size have been associated with poorer progression free survival (PFS) and overall survival (OS) in a variety of cancers; however, no study has evaluated their clinical significance in esophageal cancer (EC).

METHODS

To examine this significance, we ran a 2 year prospective pilot study consisting of newly diagnosed stage I-III EC patients (n = 32) receiving chemoradiotherapy (CRT). CAML sizes were sequentially monitored prior to CRT (BL), ~ 2 weeks into treatment (T1), and at the first available sample after the completion of CRT (T2).

RESULTS

We found CAMLs in 88% (n = 28/32) of all patient samples throughout the trial, with a sensitivity of 76% (n = 22/29) in pre-treatment screening samples. Improved 2 year PFS and OS was found in patients with CAMLs < 50 μm by the completion of CRT over patients with CAMLs ≥ 50 μm; PFS (HR = 12.0, 95% CI = 2.7-54.1, p = 0.004) and OS (HR = 9.0, 95%CI = 1.9-43.5, p = 0.019).

CONCLUSIONS

Tracking CAML sizes throughout CRT as a minimally invasive biomarker may serve as a prognostic tool in mapping EC progression, and further studies are warranted to determine if presence of these cells prior to treatment suggest diagnostic value for at-risk populations.

Use and performance of machine learning models for type 2 diabetes prediction in community settings: A systematic review and meta-analysis

OBJECTIVE

We aimed to identify machine learning (ML) models for type 2 diabetes (T2DM) prediction in community settings and determine their predictive performance.

METHOD

Systematic review of ML predictive modelling studies in 13 databases since 2009 was conducted. Primary outcomes included metrics of discrimination, calibration, and classification. Secondary outcomes included important variables, level of validation, and intended use of models. Meta-analysis of c-indices, subgroup analyses, meta-regression, publication bias assessments and sensitivity analyses were conducted.

RESULTS

Twenty-three studies (40 prediction models) were included. Studies with high-, moderate-, and low- risk of bias were 3, 14, and 6 respectively. All studies conducted internal validation whereas none conducted external validation of their models. Twenty studies provided classification metrics to varying extents whereas only 7 studies performed model calibration. Eighteen studies reported information on both the variables used for model development and the feature importance. Twelve studies highlighted potential applicability of their models for T2DM screening. Meta-analysis produced a good pooled c-index (0.812). Sources of heterogeneity were identified through subgroup analyses and meta-regression. Issues pertaining to methodological quality and reporting were observed.

CONCLUSIONS

We found evidence of good performance of ML models for T2DM prediction in the community. Improvements to methodology, reporting and validation are needed before they can be used at scale.

Systems Biology of Gastric Cancer: Perspectives on the Omics-Based Diagnosis and Treatment

Gastric cancer is the fifth most diagnosed cancer in the world, affecting more than a million people and causing nearly 783,000 deaths each year. The prognosis of advanced gastric cancer remains extremely poor despite the use of surgery and adjuvant therapy. Therefore, understanding the mechanism of gastric cancer development, and the discovery of novel diagnostic biomarkers and therapeutics are major goals in gastric cancer research. Here, we review recent progress in application of omics technologies in gastric cancer research, with special focus on the utilization of systems biology approaches to integrate multi-omics data. In addition, the association between gastrointestinal microbiota and gastric cancer are discussed, which may offer insights in exploring the novel microbiota-targeted therapeutics. Finally, the application of data-driven systems biology and machine learning approaches could provide a predictive understanding of gastric cancer, and pave the way to the development of novel biomarkers and rational design of cancer therapeutics.

Development and validation of an artificial neural network prognostic model after gastrectomy for gastric carcinoma: An international multicenter cohort study

BACKGROUND

Recently, artificial neural network (ANN) methods have also been adopted to deal with the complex multidimensional nonlinear relationship between clinicopathologic variables and survival for patients with gastric cancer. Using a multinational cohort, this study aimed to develop and validate an ANN-based survival prediction model for patients with gastric cancer.

METHODS

Patients with gastric cancer who underwent gastrectomy in a Chinese center, a Japanese center, and recorded in the Surveillance, Epidemiology, and End Results database, respectively, were included in this study. Multilayer perceptron neural network was used to develop the prediction model. Time-dependent receiver operating characteristic (ROC) curves, area under the curves (AUCs), and decision curve analysis (DCA) were used to compare the ANN model with previous prediction models.

RESULTS

An ANN model with nine input nodes, nine hidden nodes, and two output nodes was constructed. These three cohort's data showed that the AUC of the model was 0.795, 0.836, and 0.850 for 5-year survival prediction, respectively. In the calibration curve analysis, the ANN-predicted survival had a high consistency with the actual survival. Comparison of the DCA and time-dependent ROC between the ANN model and previous prediction models showed that the ANN model had good and stable prediction capability compared to the previous models in all cohorts.

CONCLUSIONS

The ANN model has significantly better discriminative capability and allows an individualized survival prediction. This model has good versatility in Eastern and Western data and has high clinical application value.

External Validation of the Dutch SOURCE Survival Prediction Model in Belgian Metastatic Oesophageal and Gastric Cancer Patients

The SOURCE prediction model predicts individualised survival conditional on various treatments for patients with metastatic oesophageal or gastric cancer. The aim of this study was to validate SOURCE in an external cohort from the Belgian Cancer Registry. Data of Belgian patients diagnosed with metastatic disease between 2004 and 2014 were extracted (n = 4097). Model calibration and discrimination (c-indices) were determined. A total of 2514 patients with oesophageal cancer and 1583 patients with gastric cancer with a median survival of 7.7 and 5.4 months, respectively, were included. The oesophageal cancer model showed poor calibration (intercept: 0.30, slope: 0.42) with an absolute mean prediction error of 14.6%. The mean difference between predicted and observed survival was -2.6%. The concordance index (c-index) of the oesophageal model was 0.64. The gastric cancer model showed good calibration (intercept: 0.02, slope: 0.91) with an absolute mean prediction error of 2.5%. The mean difference between predicted and observed survival was 2.0%. The c-index of the gastric cancer model was 0.66. The SOURCE gastric cancer model was well calibrated and had a similar performance in the Belgian cohort compared with the Dutch internal validation. However, the oesophageal cancer model had not. Our findings underscore the importance of evaluating the performance of prediction models in other populations.

Conditional survival after neoadjuvant chemoradiotherapy and surgery for oesophageal cancer

Background

Conditional survival accounts for the time already survived after surgery and may be of additional informative value. The aim was to assess conditional survival in patients with oesophageal cancer and to create a nomogram predicting the conditional probability of survival after oesophagectomy.

Methods

This retrospective study included consecutive patients with oesophageal cancer who received neoadjuvant chemoradiation followed by oesophagectomy between January 2004 and 2019. Conditional survival was defined as the probability of surviving y years after already surviving for x years. The formula used for conditional survival (CS) was: CS_(x|y) = S_(x + y)/S_(x), where S_(x) represents overall survival at x years. Cox proportional hazards models were used to evaluate predictors of overall survival. A nomogram was constructed to predict 5‐year survival directly after surgery and given survival for 1, 2, 3 and 4 years after surgery.

Results

Some 660 patients were included. Median overall survival was 44·4 (95 per cent c.i. 37·0 to 51·8) months. The probability of achieving 5‐year overall survival after resection increased from 45 per cent directly after surgery to 54, 65, 79 and 88 per cent given 1, 2, 3 and 4 years already survived respectively. Cardiac co‐morbidity, cN category, ypT category, ypN category, chyle leakage and pulmonary complications were independent predictors of survival. The nomogram predicted 5‐year survival using these predictors and number of years already survived.

Conclusion

The probability of achieving 5‐year overall survival after oesophagectomy for cancer increases with each additional year survived. The proposed nomogram predicts survival in patients after oesophagectomy, taking the years already survived into account.

Prognostic Models for Predicting Overall Survival in Patients with Primary Gastric Cancer: A Systematic Review

Background

This study was designed to review the methodology and reporting of gastric cancer prognostic models and identify potential problems in model development.

Methods

This systematic review was conducted following the CHARMS checklist. MEDLINE and EMBASE were searched. Information on patient characteristics, methodological details, and models' performance was extracted. Descriptive statistics was used to summarize the methodological and reporting quality.

Results

In total, 101 model developments and 32 external validations were included. The median (range) of training sample size, number of death, and number of final predictors were 360 (29 to 15320), 193 (14 to 9560), and 5 (2 to 53), respectively. Ninety-one models were developed from routine clinical data. Statistical assumptions were reported to be checked in only nine models. Most model developments (94/101) used complete-case analysis. Discrimination and calibration were not reported in 33 and 55 models, respectively. The majority of models (81/101) have never been externally validated. None of the models have been evaluated regarding clinical impact.

Conclusions

Many prognostic models have been developed, but their usefulness in clinical practice remains uncertain due to methodological shortcomings, insufficient reporting, and lack of external validation and impact studies.

Impact

Future research should improve methodological and reporting quality and emphasize more on external validation and impact assessment.

The Use of (Network) Meta-Analysis in Clinical Oncology

Meta-analysis is important in oncological research to provide a more reliable answer to a clinical research question that was assessed in multiple studies but with inconsistent results. Pair-wise meta-analysis can be applied when comparing two treatments at once, whereas it is possible to compare multiple treatments at once with network meta-analysis (NMA). After careful systematic review of the literature and quality assessment of the identified studies, there are several assumptions in the use of meta-analysis. First, the added value of meta-analysis should be evaluated by examining the comparability of study populations. Second, the appropriate comparator in meta-analysis should be chosen according to the types of comparisons made in individual studies: (1) Experimental and comparator arms are different treatments (A vs. B); (2) Substitution of a conventional treatment by an experimental treatment (A+B vs. A+C); or (3) Addition of an experimental treatment (A+B vs. B). Ideally there is one common comparator treatment, but when there are multiple common comparators, the most efficacious comparator is preferable. Third, treatments can only be adequately pooled in meta-analysis or merged into one treatment node in NMA when considering likewise mechanism of action and similar setting in which treatment is indicated. Fourth, for both pair-wise meta-analysis and NMA, adequate assessment of heterogeneity should be performed and sub-analysis and sensitivity analysis can be applied to objectify a possible confounding factor. Network inconsistency, as statistical manifestation of violating the transitivity assumption, can best be evaluated by node-split modeling. NMA has advantages over pair-wise meta-analysis, such as clarification of inconsistent outcomes from multiple studies including multiple common comparators and indirect effect calculation of missing direct comparisons between important treatments. Also, NMA can provide increased statistical power and cross-validation of the observed treatment effect of weak connections with reasonable network connectivity and sufficient sample-sizes. However, inappropriate use of NMA can cause misleading results, and may emerge when there is low network connectivity, and therefore low statistical power. Furthermore, indirect evidence is still observational and should be interpreted with caution. NMA should therefore preferably be conducted and interpreted by both expert clinicians in the field and an experienced statistician. Finally, the use of meta-analysis can be extended to other areas, for example the identification of prognostic and predictive factors. Also, the integration of evidence from both meta-analysis and expert opinion can improve the construction of prognostic models in real-world databases.

Meta-Enrichment Analyses to Identify Advanced Gastric Cancer Patients Who Achieve a Higher Response to S-1/Cisplatin

The Multicenter phase III comparison of cisplatin/S-1 with cisplatin/infusional fluorouracil in advanced gastric or gastroesophageal adenocarcinoma study (FLAGS) and the Diffuse Gastric and Esophagogastric Junction Cancer S-1 Trial (DIGEST) have shown that patients with advanced gastric cancer treated with S-1/Cisplatin (CS) have similar overall survival (OS) compared to 5-fluorouracil/cisplatin (CF). The purpose of this analysis was to identify patients who may specifically benefit from CS using meta-enrichment analysis of the combined two datasets. Eleven clinico-pathological factors were selected and a high response enrichable population was determined. The efficacy of CS in the combined data set of 1365 patients (n = 1019 from FLAGS and n = 346 from DIGEST) was analyzed. We identified 683 patients (n = 374 from CS, n = 309 from CF) as the high response enrichable population who were classified as those with Eastern Cooperative Oncology Group Performance Status (ECOG PS) 1, more than two metastatic sites and low neutrophil-lymphocyte ratio (log(NL ratio)). In the high response enrichable population, the median OS in the CS group was 241 days compared to 210 days in the CF group (hazard ratio 0.776; 95% confidence interval 0.658 to 0.915; p-value 0.004). Through meta-enrichment analysis, the high response enrichable population to CS was identified. Our findings show the clinical importance of selecting the appropriate treatment based on specific patient characteristics.

Prognostic and Predictive Factors for the Curative Treatment of Esophageal and Gastric Cancer in Randomized Controlled Trials: A Systematic Review and Meta-Analysis

BACKGROUND

An overview of promising prognostic variables and predictive subgroups concerning the curative treatment of esophageal and gastric cancer from randomized controlled trials (RCTs) is lacking. Therefore, we conducted a systematic review and meta-analysis.

METHODS

PubMed, EMBASE, CENTRAL, and ASCO/ESMO conferences were searched up to March 2019 for RCTs on the curative treatment of esophageal or gastric cancer with data on prognostic and/or predictive factors for overall survival. Prognostic factors were deemed potentially clinically relevant according to the following criteria; (1) statistically significant (p < 0.05) in a multivariate analysis, (2) reported in at least 250 patients, and (3) p < 0.05, in ≥ 33% of the total number of patients in RCTs reporting this factor. Predictive factors were potentially clinically-relevant if (1) the p-value for interaction between subgroups was <0.20 and (2) the hazard ratio in one of the subgroups was significant (p < 0.05).

RESULTS

For gastric cancer, 39 RCTs were identified (n = 13,530 patients) and, for esophageal cancer, 33 RCTs were identified (n = 8618 patients). In total, we identified 23 potentially clinically relevant prognostic factors for gastric cancer and 16 for esophageal cancer. There were 15 potentially clinically relevant predictive factors for gastric cancer and 10 for esophageal cancer.

CONCLUSION

The identified prognostic and predictive factors can be included and analyzed in future RCTs and be of guidance for nomograms. Further validation should be performed in large patient cohorts.

Systematic review of clinical prediction models for survival after surgery for resectable pancreatic cancer

BACKGROUND

As more therapeutic options for pancreatic cancer are becoming available, there is a need to improve outcome prediction to support shared decision-making. A systematic evaluation of prediction models in resectable pancreatic cancer is lacking.

METHODS

This systematic review followed the CHARMS and PRISMA guidelines. PubMed, Embase and Cochrane Library databases were searched up to 11 October 2017. Studies reporting development or validation of models predicting survival in resectable pancreatic cancer were included. Models without performance measures, reviews, abstracts or more than 10 per cent of patients not undergoing resection in postoperative models were excluded. Studies were appraised critically.

RESULTS

After screening 4403 studies, 22 (44 319 patients) were included. There were 19 model development/update studies and three validation studies, altogether concerning 21 individual models. Two studies were deemed at low risk of bias. Eight models were developed for the preoperative setting and 13 for the postoperative setting. Most frequently included parameters were differentiation grade (11 of 21 models), nodal status (8 of 21) and serum albumin (7 of 21). Treatment-related variables were included in three models. The C-statistic/area under the curve values ranged from 0·57 to 0·90. Based on study design, validation methods and the availability of web-based calculators, two models were identified as the most promising.

CONCLUSION

Although a large number of prediction models for resectable pancreatic cancer have been reported, most are at high risk of bias and have not been validated externally. This overview of prognostic factors provided practical recommendations that could help in designing easily applicable prediction models to support shared decision-making.

SOURCE: A Registry-Based Prediction Model for Overall Survival in Patients with Metastatic Oesophageal or Gastric Cancer

Prediction models are only sparsely available for metastatic oesophagogastric cancer. Because treatment in this setting is often preference-based, decision-making with the aid of a prediction model is wanted. The aim of this study is to construct a prediction model, called SOURCE, for the overall survival in patients with metastatic oesophagogastric cancer. Data from patients with metastatic oesophageal (n = 8010) or gastric (n = 4763) cancer diagnosed during 2005⁻2015 were retrieved from the nationwide Netherlands cancer registry. A multivariate Cox regression model was created to predict overall survival for various treatments. Predictor selection was performed via the Akaike Information Criterion and a Delphi consensus among experts in palliative oesophagogastric cancer. Validation was performed according to a temporal internal-external scheme. The predictive quality was assessed with the concordance-index (c-index) and calibration. The model c-indices showed consistent discriminative ability during validation: 0.71 for oesophageal cancer and 0.68 for gastric cancer. The calibration showed an average slope of 1.0 and intercept of 0.0 for both tumour locations, indicating a close agreement between predicted and observed survival. With a fair c-index and good calibration, SOURCE provides a solid foundation for further investigation in clinical practice to determine its added value in shared decision making.

Prognostic and predictive factors for overall survival in metastatic oesophagogastric cancer: A systematic review and meta-analysis

BACKGROUND

Consistent evidence on prognostic and predictive factors for advanced oesophagogastric cancer is lacking. Therefore, we performed a systematic review and meta-analysis.

METHODS

We searched PubMed, Embase and the Cochrane Central Register of Controlled Trials (CENTRAL) databases for phase II/III randomised controlled trials (RCTs) until February 2017 on palliative systemic therapy for advanced oesophagogastric cancer that reported prognostic or predictive factors for overall survival (PROSPERO-CRD42014015177). Prognostic factors were identified from multivariate regression analyses in study reports. Factors were considered potentially clinically relevant if statistically significant (P ≤ 0.05) in multivariate analysis in ≥50% of the total number of patients in the pooled sample of the RCTs and were reported with a pooled sample size of ≥600 patients in the first-line or ≥300 patients in the beyond first-line setting. Predictive factors were identified from time-to-event stratified treatment comparisons and deemed potentially clinically relevant if the P-value for interaction between subgroups was ≤0.20 and the hazard ratio in one of the subgroups was significant (P ≤ 0.05).

RESULTS

Forty-six original RCTs were included (n = 15,392 patients) reporting on first-line (n = 33) and beyond first-line therapy (n = 13). Seventeen prognostic factors for overall survival in the first-line and four in the beyond first-line treatment setting were potentially clinically relevant. Twenty-one predictive factors in first-line and nine in beyond first-line treatment setting were potentially relevant regarding treatment efficacy.

CONCLUSIONS

The prognostic and predictive factors identified in this systematic review can be used to characterise patients in clinical practice, be included in future trial designs, enrich prognostic tools and generate hypotheses to be tested in future research to promote patient-centred treatment.