Assessing the Clinical Impact of Risk Prediction Models With Decision Curves: Guidance for Correct Interpretation and Appropriate Use

Combining Asian-European Genome-Wide Association Studies of Colorectal Cancer Improves Risk Prediction Across Race and Ethnicity

Polygenic risk scores (PRS) have great potential to guide precision colorectal cancer (CRC) prevention by identifying those at higher risk to undertake targeted screening. However, current PRS using European ancestry data have sub-optimal performance in non-European ancestry populations, limiting their utility among these populations. Towards addressing this deficiency, we expanded PRS development for CRC by incorporating Asian ancestry data (21,731 cases; 47,444 controls) into European ancestry training datasets (78,473 cases; 107,143 controls). The AUC estimates (95% CI) of PRS were 0.63(0.62-0.64), 0.59(0.57-0.61), 0.62(0.60-0.63), and 0.65(0.63-0.66) in independent datasets including 1,681-3,651 cases and 8,696-115,105 controls of Asian, Black/African American, Latinx/Hispanic, and non-Hispanic White, respectively. They were significantly better than the European-centric PRS in all four major US racial and ethnic groups (p-values<0.05). Further inclusion of non-European ancestry populations, especially Black/African American and Latinx/Hispanic, is needed to improve the risk prediction and enhance equity in applying PRS in clinical practice.

Construction of an original anoikis-related prognostic model closely related to immune infiltration in gastric cancer

Background: Anoikis is considered as a particular type of programmed cell death, the weakness or resistance of which contributes greatly to the development and progression of most malignant solid tumors. However, the latent impact of anoikis-related genes (ARGs) on gastric cancer (GC) is still ambiguous. Based on these, this study established an anoikis-related prognostic model of GC to identify the prognosis of patients and provide more effective treatment in clinical practice. Methods: First, we extracted four public datasets containing the gene expression and clinicopathological information of GC, which were worked as the training and validating sets, separately. Then, an anoikis-related survival-predicted model of GC was developed via Lasso and COX regression analyses and verified by using the Kaplan-Meier (KM) curve and receiver operating characteristic (ROC) curve analyses. Next, we assigned GC patients to two groups characterized by the risk score calculated and analyzed somatic mutation, functional pathways, and immune infiltration between the different two groups. Finally, a unique nomogram was offered to clinicians to forecast the personal survival probability of GC patients. Results: Based on seven anoikis-related markers screened and identified, a carcinogenic model of risk score was produced. Patients placed in the high-score group suffered significantly worse overall survival (OS) in four cohorts. Additionally, the model revealed a high sensitivity and specificity to prognosticate the prognoses of GC patients [area under the ROC curve (AUC) at 5-year = 0.713; GSE84437, AUC at 5-year = 0.639; GSE15459, AUC at 5-year = 0.672; GSE62254, AUC at 5-year = 0.616]. Apart from the excellent predictive performance, the model was also identified as an independent prediction factor from other clinicopathological characteristics. Combining anoikis-related prognostic model with GC clinical features, we built a more comprehensive nomogram to foresee the likelihood of survival of GC patients in a given year, showing a well-accurate prediction performance. Conclusion: In summary, this study created a new anoikis-related signature for GC, which has potentially provided new critical insights into survival prediction and individualized therapy development.

Assessing Performance and Clinical Usefulness in Prediction Models With Survival Outcomes: Practical Guidance for Cox Proportional Hazards Models

Risk prediction models need thorough validation to assess their performance. Validation of models for survival outcomes poses challenges due to the censoring of observations and the varying time horizon at which predictions can be made. This article describes measures to evaluate predictions and the potential improvement in decision making from survival models based on Cox proportional hazards regression.

As a motivating case study, the authors consider the prediction of the composite outcome of recurrence or death (the "event") in patients with breast cancer after surgery. They developed a simple Cox regression model with 3 predictors, as in the Nottingham Prognostic Index, in 2982 women (1275 events over 5 years of follow-up) and externally validated this model in 686 women (285 events over 5 years). Improvement in performance was assessed after the addition of progesterone receptor as a prognostic biomarker.

The model predictions can be evaluated across the full range of observed follow-up times or for the event occurring by the end of a fixed time horizon of interest. The authors first discuss recommended statistical measures that evaluate model performance in terms of discrimination, calibration, or overall performance. Further, they evaluate the potential clinical utility of the model to support clinical decision making according to a net benefit measure. They provide SAS and R code to illustrate internal and external validation.

The authors recommend the proposed set of performance measures for transparent reporting of the validity of predictions from survival models.

Identification of diagnostic markers related to oxidative stress and inflammatory response in diabetic kidney disease by machine learning algorithms: Evidence from human transcriptomic data and mouse experiments

INTRODUCTION

Diabetic kidney disease (DKD) is a long-term complication of diabetes and causes renal microvascular disease. It is also one of the main causes of end-stage renal disease (ESRD), which has a complex pathophysiological process. Timely prevention and treatment are of great significance for delaying DKD. This study aimed to use bioinformatics analysis to find key diagnostic markers that could be possible therapeutic targets for DKD.

METHODS

We downloaded DKD datasets from the Gene Expression Omnibus (GEO) database. Overexpression enrichment analysis (ORA) was used to explore the underlying biological processes in DKD. Algorithms such as WGCNA, LASSO, RF, and SVM_RFE were used to screen DKD diagnostic markers. The reliability and practicability of the the diagnostic model were evaluated by the calibration curve, ROC curve, and DCA curve. GSEA analysis and correlation analysis were used to explore the biological processes and significance of candidate markers. Finally, we constructed a mouse model of DKD and diabetes mellitus (DM), and we further verified the reliability of the markers through experiments such as PCR, immunohistochemistry, renal pathological staining, and ELISA.

RESULTS

Biological processes, such as immune activation, T-cell activation, and cell adhesion were found to be enriched in DKD. Based on differentially expressed oxidative stress and inflammatory response-related genes (DEOIGs), we divided DKD patients into C1 and C2 subtypes. Four potential diagnostic markers for DKD, including tenascin C, peroxidasin, tissue inhibitor metalloproteinases 1, and tropomyosin (TNC, PXDN, TIMP1, and TPM1, respectively) were identified using multiple bioinformatics analyses. Further enrichment analysis found that four diagnostic markers were closely related to various immune cells and played an important role in the immune microenvironment of DKD. In addition, the results of the mouse experiment were consistent with the bioinformatics analysis, further confirming the reliability of the four markers.

CONCLUSION

In conclusion, we identified four reliable and potential diagnostic markers through a comprehensive and systematic bioinformatics analysis and experimental validation, which could serve as potential therapeutic targets for DKD. We performed a preliminary examination of the biological processes involved in DKD pathogenesis and provide a novel idea for DKD diagnosis and treatment.

Evolution of Making Clinical Predictions in Neurosurgery

Prediction of clinical outcomes is an essential task for every physician. Physicians may base their clinical prediction of an individual patient on their intuition and on scientific material such as studies presenting population risks and studies reporting on risk factors (prognostic factors). A relatively new and more informative approach for making clinical predictions relies on the use of statistical models that simultaneously consider multiple predictors that provide an estimate of the patient's absolute risk of an outcome. There is a growing body of literature in the neurosurgical field reporting on clinical prediction models. These tools have high potential in supporting (not replacing) neurosurgeons with their prediction of a patient's outcome. If used sensibly, these tools pave the way for more informed decision-making with or for individual patients. Patients and their significant others want to know their risk of the anticipated outcome, how it is derived, and the uncertainty associated with it. Learning from these prediction models and communicating the output to others has become an increasingly important skill neurosurgeons have to master. This article describes the evolution of making clinical predictions in neurosurgery, synopsizes key phases for the generation of a useful clinical prediction model, and addresses some considerations when deploying and communicating the results of a prediction model. The paper is illustrated with multiple examples from the neurosurgical literature, including predicting arachnoid cyst rupture, predicting rebleeding in patients suffering from aneurysmal subarachnoid hemorrhage, and predicting survival in glioblastoma patients.

A novel risk stratification model for STEMI after primary PCI: global longitudinal strain and deep neural network assisted myocardial contrast echocardiography quantitative analysis

Background

In ST-segment elevation myocardial infarction (STEMI) with the restoration of TIMI 3 flow by percutaneous coronary intervention (PCI), visually defined microvascular obstruction (MVO) was shown to be the predictor of poor prognosis, but not an ideal risk stratification method. We intend to introduce deep neural network (DNN) assisted myocardial contrast echocardiography (MCE) quantitative analysis and propose a better risk stratification model.

Methods

194 STEMI patients with successful primary PCI with at least 6 months follow-up were included. MCE was performed within 48 h after PCI. The major adverse cardiovascular events (MACE) were defined as cardiac death, congestive heart failure, reinfarction, stroke, and recurrent angina. The perfusion parameters were derived from a DNN-based myocardial segmentation framework. Three patterns of visual microvascular perfusion (MVP) qualitative analysis: normal, delay, and MVO. Clinical markers and imaging features, including global longitudinal strain (GLS) were analyzed. A calculator for risk was constructed and validated with bootstrap resampling.

Results

The time-cost for processing 7,403 MCE frames is 773 s. The correlation coefficients of microvascular blood flow (MBF) were 0.99 to 0.97 for intra-observer and inter-observer variability. 38 patients met MACE in 6-month follow-up. We proposed A risk prediction model based on MBF [HR: 0.93 (0.91-0.95)] in culprit lesion areas and GLS [HR: 0.80 (0.73-0.88)]. At the best risk threshold of 40%, the AUC was 0.95 (sensitivity: 0.84, specificity: 0.94), better than visual MVP method (AUC: 0.70, Sensitivity: 0.89, Specificity: 0.40, IDI: -0.49). The Kaplan-Meier curves showed that the proposed risk prediction model allowed for better risk stratification.

Conclusion

The MBF + GLS model allowed more accurate risk stratification of STEMI after PCI than visual qualitative analysis. The DNN-assisted MCE quantitative analysis is an objective, efficient and reproducible method to evaluate microvascular perfusion.

Identification of a novel reactive oxygen species (ROS)-related genes model combined with RT-qPCR experiments for prognosis and immunotherapy in gastric cancer

Reactive oxygen species play a crucial role in the prognosis and tumor microenvironment (TME) of malignant tumors. An ROS-related signature was constructed in gastric cancer (GC) samples from TCGA database. ROS-related genes were obtained from the Molecular Signatures Database. Consensus clustering was used to establish distinct ROS-related subtypes related to different survival and immune cell infiltration patterns. Sequentially, prognostic genes were identified in the ROS-related subtypes, which were used to identify a stable ROS-related signature that predicted the prognosis of GC. Correlation analysis revealed the significance of immune cell iniltration, immunotherapy, and drug sensitivity in gastric cancers with different risks. The putative molecular mechanisms of the different gastric cancer risks were revealed by functional enrichment analysis. A robust nomogram was established to predict the outcome of each gastric cancer. Finally, we verified the expression of the genes involved in the model using RT-qPCR. In conclusion, the ROS-related signature in this study is a novel and stable biomarker associated with TME and immunotherapy responses.

The significance of time interval between perioperative SOX/XELOX chemotherapy and clinical decision model in gastric cancer

Introduction

To investigate the influences of time interval between multimodality therapies on survival for locally advanced gastric cancer (LAGC) patients, 627 patients were included in a retrospective study, and 350 who received neoadjuvant chemotherapy (NACT) based on SOX (S-1 plus Oxaliplatin)/XELOX (Capecitabine plus Oxaliplatin) treatment, radical surgery, and adjuvant chemotherapy (AC) from 2005.01 to 2018.06 were eligible for analyses.

Methods

Three factors were used to assess influences, including time interval from NACT accomplishment to AC initiation (PECTI), time to surgery after NACT accomplishment (TTS), and time to adjuvant chemotherapy after surgery (TAC).

Results

Concerning PECTIs, 99 (28.29%) experienced it within 9 weeks, 188 (53.71%) within 9-13 weeks, 63 (18.00%) over 13 weeks. Patients' 5-year overall survival (OS) significantly decreased as trichotomous PECTI increased (78.6% vs 66.7% vs 55.7%, P = .02). Analogously, there was a significant decrease for dichotomous TTS (within vs over 5 weeks) in OS (P = .03) and progression free survival (PFS) (P = .01) but not for dichotomous TAC (within vs over 6 weeks) in OS and PFS (P = .40). Through multivariate Cox analyses, patients with PECTI over 13 weeks had significantly worse OS (P = .03) and PFS (P = .02). Furthermore, extended TTS had significantly worse OS and PFS but insignificantly worse OS and PFS than extended TAC. Therefore, gastric patients receiving perioperative SOX/XELOX chemotherapy and surgery with extended PECTI over 9 weeks or TTS over 5 weeks would have a negative correlation with PFS and OS, and worse when PECTI over 13 weeks. Nomograms (including PECTI, ypT, ypN, Area Under Curve (AUC) = 0.81) could predict patient survival probability and guide intervention with net benefit.

Discussion

In control of PECTI, TTS could be extended appropriately, and shortened TAC might make a remedy, and delayed TAC might be allowed when TTS was shortened.

Low-dose glucocorticoids withdrawn in systemic lupus erythematosus: a desirable and attainable goal

OBJECTIVES

To assess the risk of flare in systemic lupus erythematosus (SLE) patients after low-dose glucocorticoid (GC) discontinuation and to evaluate the risk factors of flare.

METHODS

SLE patients who ever discontinued GCs were identified from the Peking University First Hospital SLE cohort. The disease flare profile after GC discontinuation was analysed. The flare rate was analysed using Kaplan-Meier analysis. Cox regression was used to determine the effects of variables on SLE flare. A prognostic nomogram using Cox proportional hazards regression modelling was developed.

RESULTS

A total of 132 SLE patients were eligible for the final analysis. They were followed up for a median of 21.8 months (interquartile range 9.01-36.7). The cumulative probability of flare after GC discontinuation was 8.3% at 6 months, 16.8% at 1 years and 27.5% at 2 years. In multivariate Cox analysis, hypocomplementemia and serologically active clinically quiescent (SACQ) were independent risk factors of flare [hazard ratio (HR0 2.53 (95% CI 1.32, 4.88); HR 3.17 (95% CI 1.44, 6.97), respectively]. Age ≥40 years at GC withdrawal and hydroxychloroquine (HCQ) usage were independent protective factors of flare [HR 0.53 (95% CI 0.29, 0.99); HR 0.32 (95% CI 0.17, 0.62), respectively]. The protective effect of HCQ was dosage related. From the perspective of different tapering strategies embodied as the duration from prednisone 5 mg/day to complete discontinuation, a slower tapering strategy (12-24 months) significantly reduced the risk of flare compared with a faster tapering strategy (<3 months) [HR 0.30 (95% CI 0.11, 0.82), P = 0.019]. The prognostic nomogram including the aforementioned factors effectively predicted the 1 and 2 year probability of being flare-free.

CONCLUSION

Low-dose GC is feasibly discontinued in real-life settings. SACQ and younger age are potential risk factors of SLE flare, while HCQ use and slow GC tapering to withdrawal can reduce relapse. The visualized model we developed may help to predict the risk of flare among SLE patients who discontinued GC.

Reduced Skeletal Muscle Mass Is Associated with an Increased Risk of Asthma Control and Exacerbation

BACKGROUND

Skeletal muscle mass (SMM) has been suggested to be associated with multiple health-related outcomes. However, the potential influence of SMM on asthma has not been largely explored.

OBJECTIVE

To study the association between SMM and clinical features of asthma, including asthma control and exacerbation, and to construct a model based on SMM to predict the risk of asthma exacerbation (AEx).

METHODS

In this prospective cohort study, we consecutively recruited patients with asthma (n = 334), classified as the SMM ^Normal group (n = 223), SMM ^Low group (n = 88), and SMM ^High group (n = 23). We investigated the association between SMM and clinical asthma characteristics and explored the association between SMM and asthma control and AEx within a 12-month follow-up period. Based on SMM, an exacerbation prediction model was developed, and the overall performance was externally validated in an independent cohort (n = 157).

RESULTS

Compared with the SMM ^Normal group, SMM ^Low group exhibited more airway obstruction and worse asthma control, while SMM ^High group had a reduced eosinophil percentage in induced sputum. Furthermore, SMM ^Low group was at a significantly increased risk of moderate-to-severe exacerbation compared with the SMM ^Normal group (relative risk _adjusted 2.02 [95% confidence interval (CI), 1.35-2.68]; p = 0.002). In addition, a model involving SMM was developed which predicted AEx (area under the curve: 0.750, 95% CI: 0.691-0.810).

CONCLUSIONS

Low SMM was an independent risk factor for future AEx. Furthermore, a model involving SMM for predicting the risk of AEx in patients with asthma indicated that assessment of SMM has potential clinical implications for asthma management.

Classification of solid pulmonary nodules using a machine-learning nomogram based on 18F-FDG PET/CT radiomics integrated clinicobiological features

Background

To develop and validate an ¹⁸F-fluorodeoxyglucose (FDG) positron emission tomography/computed tomography (PET/CT) and clinico-biological features-based nomogram for distinguishing solid benign pulmonary nodules (BPNs) from malignant pulmonary nodules (MPNs).

Methods

A total of 280 patients with BPN (n=128) or MPN (n=152) were collected retrospectively and randomized into the training set (n=196) and validation set (n=84). Pretherapeutic clinicobiological markers, PET/CT metabolic features and radiomic features were analyzed and selected to develop prediction models by the machine-learning method [Least Absolute Shrinkage and Selection Operator (LASSO) regression]. These prediction models were validated using the area under the curve (AUC) of the receiver-operator characteristic (ROC) analysis and decision curve analysis (DCA). Then, the factors of the model with the optimal predictive efficiency were used to constructed a nomogram to provide a visually quantitative tool for distinguishing BPN from MPN patients.

Results

We developed 3 independent models (Clinical Model, Radiomics Model and Combined Model) to distinguish patients with BPN from those with MPN in the training set. The Combined Model was validated to hold the optimal efficiency and clinical utility with the lowest false positive rate (FPR) in classifying the solid pulmonary nodules in two sets (AUCs of 0.91 and 0.94, FPRs of 18.68% and 5.41%, respectively; P<0.05). Thus, the quantitative nomogram was developed based on the Combined Model, and a good consistency between the predictions and the actual observations was validated by the calibration curves.

Conclusions

This study presents a machine-learning nomogram integrated clinico-biologico-radiological features that can improve the efficiency and reduce the FPR in the noninvasive differentiation of BPN from MPN.

Development of Machine Learning Algorithms Incorporating Electronic Health Record Data, Patient-Reported Outcomes, or Both to Predict Mortality for Outpatients With Cancer

PURPOSE

Machine learning (ML) algorithms that incorporate routinely collected patient-reported outcomes (PROs) alongside electronic health record (EHR) variables may improve prediction of short-term mortality and facilitate earlier supportive and palliative care for patients with cancer.

METHODS

We trained and validated two-phase ML algorithms that incorporated standard PRO assessments alongside approximately 200 routinely collected EHR variables, among patients with medical oncology encounters at a tertiary academic oncology and a community oncology practice.

RESULTS

Among 12,350 patients, 5,870 (47.5%) completed PRO assessments. Compared with EHR- and PRO-only algorithms, the EHR + PRO model improved predictive performance in both tertiary oncology (EHR + PRO v EHR v PRO: area under the curve [AUC] 0.86 [0.85-0.87] v 0.82 [0.81-0.83] v 0.74 [0.74-0.74]) and community oncology (area under the curve 0.89 [0.88-0.90] v 0.86 [0.85-0.88] v 0.77 [0.76-0.79]) practices.

CONCLUSION

Routinely collected PROs contain added prognostic information not captured by an EHR-based ML mortality risk algorithm. Augmenting an EHR-based algorithm with PROs resulted in a more accurate and clinically relevant model, which can facilitate earlier and targeted supportive care for patients with cancer.

A nomogram based on shear wave elastography for assessment of renal fibrosis in patients with chronic kidney disease

BACKGROUND

Non-invasive evaluation of renal fibrosis is still challenging. This study aimed to establish a nomogram based on shear wave elastography (SWE) and clinical features for the assessment of the severity of renal fibrosis in patients with chronic kidney disease (CKD).

METHODS

One hundred and sixty-two patients with CKD who underwent kidney biopsy and SWE examination were prospectively enrolled between April 2019 and December 2021. Patients were classified into mildly or moderately-severely impaired group based on pathology results. All patients were randomly divided into a training (n = 113) or validation cohort (n = 49). Least absolute shrinkage and selection operator (LASSO) algorithm was used for data dimensionality reduction and feature selection. Then, a diagnostic nomogram incorporating the selected features was constructed using multivariable logistic regression analysis. Nomogram performance was evaluated for discrimination, calibration, and clinical utility in training and validation cohorts.

RESULTS

The established SWE nomogram, which integrated SWE value, hypertension, and estimated glomerular filtration rate, showed fine calibration and discrimination in both training (area under the receiver operator characteristic curve (AUC) = 0.94; 95% confidence interval (CI) 0.89-0.98) and validation cohorts (AUC = 0.84; 95% CI 0.71-0.96). Significant improvement in net reclassification and integrated discrimination indicated that the SWE value is a valuable biomarker to assess moderate-severe renal impairment. Furthermore, decision curve analysis revealed that the SWE nomogram has clinical value.

CONCLUSION

The proposed SWE nomogram showed favorable performance in determining individualized risk of moderate-severe renal pathological impairment in patients with CKD, which will help to facilitate clinical decision-making.

Non-invasive model for predicting esophageal varices based on liver and spleen volume

BACKGROUND

Upper endoscopy is the gold standard for predicting esophageal varices in China. Guidelines and consensus suggest that patients with liver cirrhosis should undergo periodic upper endoscopy, most patients undergo their first upper endoscopy when esophageal variceal bleeds. Therefore, it is important to develop a non-invasive model to early diagnose esophageal varices.

AIM

To develop a non-invasive predictive model for esophageal varices based on liver and spleen volume in viral cirrhosis patients.

METHODS

We conducted a cross-sectional study based on viral cirrhosis crowd in the Second Affiliated Hospital of Xi'an Jiaotong University. By collecting the basic information and clinical data of the participants, we derived the independent risk factors and established the prediction model of esophageal varices. The established model was compared with other models. Area under the receiver operating characteristic curve, calibration plot and decision curve analysis were used to test the discriminating ability, calibration ability and clinical practicability in both the internal and external validation.

RESULTS

The portal vein diameter, the liver and spleen volume, and volume change rate were the independent risk factors of esophageal varices. We successfully used the factors to establish the predictive model [area under the curve (AUC) 0.87, 95%CI: 0.80-0.95], which showed better predictive value than other models. The model showed good discriminating ability, calibration ability and the clinical practicability in both modelling group and external validation group.

CONCLUSION

The developed non-invasive predictive model can be used as an effective tool for predicting esophageal varices in viral cirrhosis patients.

Development and validation of a predictive model of the impact of single nucleotide polymorphisms in the ICAM-1 gene on the risk of ischemic cardiomyopathy

OBJECTIVE

Previous research has linked single nucleotide polymorphisms (SNPs) in the ICAM-1 gene to an increased risk of developing ischemic cardiomyopathy (ICM); however, a diagnostic model of ICM according to the ICAM-1 variant has not yet been developed. Therefore, this study aimed to explore the correlation between SNPs in ICAM-1 and the presence of ICM, along with developing a diagnostic model for ICM based on the variants of the ICAM-1 gene.

METHOD

This study recruited a total of 252 patients with ICM and 280 healthy controls. In addition, all the participants were genotyped for SNPs in the ICAM-1 gene by polymerase chain reaction-restriction fragment length polymorphism (PCR-RFLP). Using the training dataset of 371 people, we constructed a nomogram model based on ICAM-1 gene variants and clinical variables. To optimize the feature choice for the ICM risk model, a least absolute shrinkage and selection operator (LASSO) regression model was adopted. We also employed multivariable logistic regression analysis to build a prediction model by integrating the clinical characteristics chosen in the LASSO regression model. Following the receiver operating characteristic (ROC), a calibration plot and decision curve analysis (DCA) were used to evaluate the discrimination, calibration, and clinical usefulness of the predictive model.

RESULT

The predictors involved in the prediction nomogram included age, smoking, diabetes, low-density lipoprotein-cholesterol, hemoglobin, N-terminal pro-B-type natriuretic peptide, ejection fraction, and the rs5491 SNP. The nomogram model exhibited good discrimination ability, with the AUC value of ROC of 0.978 (95%CI: 0.967-0.989, P < 0.001) in the training group and 0.983 (95% CI: 0.969-0.998, P < 0.001) in the validation group. The Hosmer-Lemeshow test demonstrated good model calibration with consistency (P training group = 0.937; P validation group = 0.910). The DCA showed that the ICM nomogram was clinically beneficial, with the threshold probabilities ranging from 0.0 to 1.0.

CONCLUSION

The AT genotype in rs5491 of the ICAM-1 gene was associated with having a higher frequency of ICM. Individuals carrying the mutant AT genotype showed a 5.816-fold higher frequency of ICM compared with those with the AA genotype. ICM patients with the AT genotype also had a higher rate of cardiogenic death. We, therefore, developed a nomogram model that could offer an individualized prediction of ICM risk factors.

Predictive machine learning model for microvascular invasion identification in hepatocellular carcinoma based on the LI-RADS system

Purposes

This study aimed to establish a predictive model of microvascular invasion (MVI) in hepatocellular carcinoma (HCC) by contrast-enhanced computed tomography (CT), which relied on a combination of machine learning approach and imaging features covering Liver Imaging and Reporting and Data System (LI-RADS) features.

Methods

The retrospective study included 279 patients with surgery who underwent preoperative enhanced CT. They were randomly allocated to training set, validation set, and test set (167 patients vs. 56 patients vs. 56 patients, respectively). Significant imaging findings for predicting MVI were identified through the Least Absolute Shrinkage and Selection Operator (LASSO) logistic regression method. Predictive models were performed by machine learning algorithm, support vector machine (SVM), in the training set and validation set, and evaluated in the test set. Further, a combined model adding clinical findings to the radiologic model was developed. Based on the LI-RADS category, subgroup analyses were conducted.

Results

We included 116 patients with MVI which were diagnosed through pathological confirmation. Six imaging features were selected about MVI prediction: four LI-RADS features (corona enhancement, enhancing capsule, non-rim aterial phase hyperehancement, tumor size) and two non-LI-RADS features (internal arteries, non-smooth tumor margin). The radiological feature with the best accuracy was corona enhancement followed by internal arteries and tumor size. The accuracies of the radiological model and combined model were 0.725–0.714 and 0.802–0.732 in the training set, validation set, and test set, respectively. In the LR-4/5 subgroup, a sensitivity of 100% and an NPV of 100% were obtained by the high-sensitivity threshold. A specificity of 100% and a PPV of 100% were acquired through the high specificity threshold in the LR-M subgroup.

Conclusion

A combination of LI-RADS features and non-LI-RADS features and serum alpha-fetoprotein value could be applied as a preoperative biomarker for predicting MVI by the machine learning approach. Furthermore, its good performance in the subgroup by LI-RADS category may help optimize the management of HCC patients.

Does the SORG Orthopaedic Research Group Hip Fracture Delirium Algorithm Perform Well on an Independent Intercontinental Cohort of Patients With Hip Fractures Who Are 60 Years or Older?

BACKGROUND

Postoperative delirium in patients aged 60 years or older with hip fractures adversely affects clinical and functional outcomes. The economic cost of delirium is estimated to be as high as USD 25,000 per patient, with a total budgetary impact between USD 6.6 to USD 82.4 billion annually in the United States alone. Forty percent of delirium episodes are preventable, and accurate risk stratification can decrease the incidence and improve clinical outcomes in patients. A previously developed clinical prediction model (the SORG Orthopaedic Research Group hip fracture delirium machine-learning algorithm) is highly accurate on internal validation (in 28,207 patients with hip fractures aged 60 years or older in a US cohort) in identifying at-risk patients, and it can facilitate the best use of preventive interventions; however, it has not been tested in an independent population. For an algorithm to be useful in real life, it must be valid externally, meaning that it must perform well in a patient cohort different from the cohort used to "train" it. With many promising machine-learning prediction models and many promising delirium models, only few have also been externally validated, and even fewer are international validation studies.

QUESTION/PURPOSE

Does the SORG hip fracture delirium algorithm, initially trained on a database from the United States, perform well on external validation in patients aged 60 years or older in Australia and New Zealand?

METHODS

We previously developed a model in 2021 for assessing risk of delirium in hip fracture patients using records of 28,207 patients obtained from the American College of Surgeons National Surgical Quality Improvement Program. Variables included in the original model included age, American Society of Anesthesiologists (ASA) class, functional status (independent or partially or totally dependent for any activities of daily living), preoperative dementia, preoperative delirium, and preoperative need for a mobility aid. To assess whether this model could be applied elsewhere, we used records from an international hip fracture registry. Between June 2017 and December 2018, 6672 patients older than 60 years of age in Australia and New Zealand were treated surgically for a femoral neck, intertrochanteric hip, or subtrochanteric hip fracture and entered into the Australian & New Zealand Hip Fracture Registry. Patients were excluded if they had a pathological hip fracture or septic shock. Of all patients, 6% (402 of 6672) did not meet the inclusion criteria, leaving 94% (6270 of 6672) of patients available for inclusion in this retrospective analysis. Seventy-one percent (4249 of 5986) of patients were aged 80 years or older, after accounting for 5% (284 of 6270) of missing values; 68% (4292 of 6266) were female, after accounting for 0.06% (4 of 6270) of missing values, and 83% (4690 of 5661) of patients were classified as ASA III/IV, after accounting for 10% (609 of 6270) of missing values. Missing data were imputed using the missForest methodology. In total, 39% (2467 of 6270) of patients developed postoperative delirium. The performance of the SORG hip fracture delirium algorithm on the validation cohort was assessed by discrimination, calibration, Brier score, and a decision curve analysis. Discrimination, known as the area under the receiver operating characteristic curves (c-statistic), measures the model's ability to distinguish patients who achieved the outcomes from those who did not and ranges from 0.5 to 1.0, with 1.0 indicating the highest discrimination score and 0.50 the lowest. Calibration plots the predicted versus the observed probabilities, a perfect plot has an intercept of 0 and a slope of 1. The Brier score calculates a composite of discrimination and calibration, with 0 indicating perfect prediction and 1 the poorest.

RESULTS

The SORG hip fracture algorithm, when applied to an external patient cohort, distinguished between patients at low risk and patients at moderate to high risk of developing postoperative delirium. The SORG hip fracture algorithm performed with a c-statistic of 0.74 (95% confidence interval 0.73 to 0.76). The calibration plot showed high accuracy in the lower predicted probabilities (intercept -0.28, slope 0.52) and a Brier score of 0.22 (the null model Brier score was 0.24). The decision curve analysis showed that the model can be beneficial compared with no model or compared with characterizing all patients as at risk for developing delirium.

CONCLUSION

Algorithms developed with machine learning are a potential tool for refining treatment of at-risk patients. If high-risk patients can be reliably identified, resources can be appropriately directed toward their care. Although the current iteration of SORG should not be relied on for patient care, it suggests potential utility in assessing risk. Further assessment in different populations, made easier by international collaborations and standardization of registries, would be useful in the development of universally valid prediction models. The model can be freely accessed at: https://sorg-apps.shinyapps.io/hipfxdelirium/ .

LEVEL OF EVIDENCE

Level III, therapeutic study.

Predicting persistent organ failure on admission in patients with acute pancreatitis: development and validation of a mobile nomogram

BACKGROUND

Early prediction of persistent organ failure (POF) is important for triage and timely treatment of patients with acute pancreatitis (AP).

METHODS

All AP patients were consecutively admitted within 48 h of symptom onset. A nomogram was developed to predict POF on admission using data from a retrospective training cohort, validated by two prospective cohorts. The clinical utility of the nomogram was defined by concordance index (C-index), decision curve analysis (DCA), and clinical impact curve (CIC), while the performance by post-test probability.

RESULTS

There were 816, 398, and 880 patients in the training, internal and external validation cohorts, respectively. Six independent predictors determined by logistic regression analysis were age, respiratory rate, albumin, lactate dehydrogenase, oxygen support, and pleural effusion and were included in the nomogram (web-based calculator: https://shina.shinyapps.io/DynNomapp/). This nomogram had reasonable predictive ability (C-indexes 0.88/0.91/0.81 for each cohort) and promising clinical utility (DCA and CIC). The nomogram had a positive likelihood ratio and post-test probability of developing POF in the training, internal and external validation cohorts of 4.26/31.7%, 7.89/39.1%, and 2.75/41%, respectively, superior or equal to other prognostic scores.

CONCLUSIONS

This nomogram can predict POF of AP patients and should be considered for clinical practice and trial allocation.

Post-operative anastomotic leakage and collagen changes in patients with rectal cancer undergoing neoadjuvant chemotherapy vs chemoradiotherapy

Background

A significant difference in the anastomotic leakage (AL) rate has been observed between patients with locally advanced rectal cancer who have undergone preoperative chemotherapy and those undergoing preoperative chemoradiotherapy. This study aimed to quantitatively analyse collagen structural changes caused by preoperative chemoradiotherapy and illuminate the relationship between collagen changes and AL.

Methods

Anastomotic distal and proximal "doughnut" specimens from the Sixth Affiliated Hospital of Sun Yat-sen University (Guangzhou, China) were quantitatively assessed for collagen structural changes between patients with and without preoperative radiotherapy using multiphoton imaging. Then, patients treated with preoperative chemoradiotherapy were used as a training cohort to construct an AL-SVM classifier by the Mann-Whitney U test and support vector machine (SVM). An independent test cohort from the Fujian Province Cancer Hospital (Fuzhou, China) was used to validate the AL-SVM classifier.

Results

A total of 207 patients were included from the Sixth Affiliated Hospital of Sun Yat-sen University. The AL rate in the preoperative chemoradiotherapy group (n = 107) was significantly higher than that in the preoperative chemotherapy group (n = 100) (21.5% vs 7.0%, P = 0.003). A fully quantitative analysis showed notable morphological and spatial distribution feature changes in collagen in the preoperative chemoradiotherapy group. Then, the patients who received preoperative chemoradiotherapy were used as a training cohort to construct the AL-SVM classifier based on five collagen features and the tumor distance from the anus. The AL-SVM classifier showed satisfactory discrimination and calibration with areas under the curve of 0.907 and 0.856 in the training and test cohorts, respectively.

Conclusions

The collagen structure may be notably altered by preoperative radiotherapy. The AL-SVM classifier was useful for the individualized prediction of AL in rectal cancer patients undergoing preoperative chemoradiotherapy.

PredictCBC-2.0: a contralateral breast cancer risk prediction model developed and validated in ~ 200,000 patients

BACKGROUND

Prediction of contralateral breast cancer (CBC) risk is challenging due to moderate performances of the known risk factors. We aimed to improve our previous risk prediction model (PredictCBC) by updated follow-up and including additional risk factors.

METHODS

We included data from 207,510 invasive breast cancer patients participating in 23 studies. In total, 8225 CBC events occurred over a median follow-up of 10.2 years. In addition to the previously included risk factors, PredictCBC-2.0 included CHEK2 c.1100delC, a 313 variant polygenic risk score (PRS-313), body mass index (BMI), and parity. Fine and Gray regression was used to fit the model. Calibration and a time-dependent area under the curve (AUC) at 5 and 10 years were assessed to determine the performance of the models. Decision curve analysis was performed to evaluate the net benefit of PredictCBC-2.0 and previous PredictCBC models.

RESULTS

The discrimination of PredictCBC-2.0 at 10 years was higher than PredictCBC with an AUC of 0.65 (95% prediction intervals (PI) 0.56-0.74) versus 0.63 (95%PI 0.54-0.71). PredictCBC-2.0 was well calibrated with an observed/expected ratio at 10 years of 0.92 (95%PI 0.34-2.54). Decision curve analysis for contralateral preventive mastectomy (CPM) showed the potential clinical utility of PredictCBC-2.0 between thresholds of 4 and 12% 10-year CBC risk for BRCA1/2 mutation carriers and non-carriers.

CONCLUSIONS

Additional genetic information beyond BRCA1/2 germline mutations improved CBC risk prediction and might help tailor clinical decision-making toward CPM or alternative preventive strategies. Identifying patients who benefit from CPM, especially in the general breast cancer population, remains challenging.

Identification and Development of an Age-Related Classification and Signature to Predict Prognosis and Immune Landscape in Osteosarcoma

Background. In childhood and adolescence, the prevailing bone tumor is osteosarcoma associated with frequent recurrence and lung metastasis. This research focused on predicting the survival and immune landscape of osteosarcoma by developing a prognostic signature and establishing aging-related genes (ARGs) subtypes. Methods. The training group comprised of the transcriptomic and associated clinical data of 84 patients with osteosarcoma accessed at the TARGET database and the validation group consisted of 53 patients from GSE21257. The aging-related subtypes were identified using unsupervised consensus clustering analysis. The ARG signature was developed utilizing multivariate Cox analysis and LASSO regression. The prognostic value was assessed using the univariate and multivariate Cox analyses, Kaplan-Meier plotter, time-dependent ROC curve, and nomogram. The functional enrichment analyses were performed by GSEA, GO, and KEGG analysis, while the ssGSEA, ESTIMATE, and CIBERSORT analyses were conducted to reveal the immune landscape in osteosarcoma. Results. The two clusters of osteosarcoma patients formed based on 543 ARGs, depicted a considerable difference in the tumor microenvironment, and the overall survival and immune cell infiltration rate varied as well. Among these, the selected 23 ARGs were utilized for the construction of an efficient predictive prognostic signature for the overall survival prediction. The testing in the validation group of osteosarcoma patients confirmed the status of the high-risk score as an independent indicator for poor prognosis, which was already identified as such using the univariate and multivariate Cox analyses. Furthermore, the ARG signature could distinguish different immune-related functions, infiltration status of immune cells, and tumor microenvironment, as well as predict the immunotherapy response of osteosarcoma patients. Conclusion. The aging-related subtypes were identified and a prognostic signature was developed in this research, which determined different prognoses and allowed for treatment of osteosarcoma patients to be tailored. Additionally, the immunotherapeutic response of individuals with osteosarcoma could also be predicted by the ARG signature.

Feasibility of a CT-based lymph node radiomics nomogram in detecting lymph node metastasis in PDAC patients

Objectives

To investigate the potential value of a contrast enhanced computed tomography (CECT)-based radiological-radiomics nomogram combining a lymph node (LN) radiomics signature and LNs’ radiological features for preoperative detection of LN metastasis in patients with pancreatic ductal adenocarcinoma (PDAC).

Materials and methods

In this retrospective study, 196 LNs in 61 PDAC patients were enrolled and divided into the training (137 LNs) and validation (59 LNs) cohorts. Radiomic features were extracted from portal venous phase images of LNs. The least absolute shrinkage and selection operator (LASSO) regression algorithm with 10-fold cross-validation was used to select optimal features to determine the radiomics score (Rad-score). The radiological-radiomics nomogram was developed by using significant predictors of LN metastasis by multivariate logistic regression (LR) analysis in the training cohort and validated in the validation cohort independently. Its diagnostic performance was assessed by receiver operating characteristic curve (ROC), decision curve (DCA) and calibration curve analyses.

Results

The radiological model, including LN size, and margin and enhancement pattern (three significant predictors), exhibited areas under the curves (AUCs) of 0.831 and 0.756 in the training and validation cohorts, respectively. Nine radiomic features were used to construct a radiomics model, which showed AUCs of 0.879 and 0.804 in the training and validation cohorts, respectively. The radiological-radiomics nomogram, which incorporated the LN Rad-score and the three LNs’ radiological features, performed better than the Rad-score and radiological models individually, with AUCs of 0.937 and 0.851 in the training and validation cohorts, respectively. Calibration curve analysis and DCA revealed that the radiological-radiomics nomogram showed satisfactory consistency and the highest net benefit for preoperative diagnosis of LN metastasis.

Conclusions

The CT-based LN radiological-radiomics nomogram may serve as a valid and convenient computer-aided tool for personalized risk assessment of LN metastasis and help clinicians make appropriate clinical decisions for PADC patients.

Development and validation of safety and efficacy-associated risk calculator for hepatocellular carcinoma in the elderly after resection (SEARCHER): A multi-institutional observational study

BACKGROUND

Increased life expectancy and improved perioperative management have resulted in increased utilization of hepatectomy for hepatocellular carcinoma (HCC) among elderly patients. However, individualized model for predicting the surgical safety and efficacy is lacking. The present study aimed to develop a safety and efficacy-associated risk calculator for HCC in the elderly after resection (SEARCHER).

METHODS

From an international multicenter database, elderly patients who underwent curative-intent hepatectomy for HCC were stratified by patient age: 65-69 years, 70-74 years, 75-79 years, and ≥80 years. Short- and long-term outcomes among the 4 groups were compared. Univariate and multivariate analyses of risk factors of postoperative major morbidity, cancer-specific survival (CSS) and overall survival (OS) were performed in the training cohort. A nomogram-based online calculator was then constructed and validated in the validation cohort.

RESULTS

With increasing age, the risk of postoperative major morbidity and worse OS increased (P = 0.001 and 0.020), but not postoperative mortality and CSS (P = 0.577 and 0.890) among patients across the 4 groups. Based on three nomograms to predict major morbidity, CSS and OS, the SEARCHER model was constructed and made available at https://elderlyhcc.shinyapps.io/SEARCHER. The model demonstrated excellent calibration and optimal performance in both the training and validation cohorts, and performed better than the several commonly-used conventional scoring and staging systems of HCC.

CONCLUSIONS

With higher potential postoperative major morbidity and worse OS as patients age, the decision of whether to perform a hepatectomy for HCC needs to be comprehensively considered in the elderly. The proposed SEARCHER model demonstrated good performance to individually predict safety and efficacy of hepatectomy in elderly patients with HCC.

A Web-Based Prediction Model for Estimating the Probability of Post-hepatectomy Major Complications in Patients with Hepatocellular Carcinoma: A Multicenter Study from a Hepatitis B Virus-Endemic Area

BACKGROUND

The identification of patients at high risk of developing postoperative complications is important to improve surgical safety. We sought to develop an individualized tool to predict post-hepatectomy major complications in hepatitis B virus (HBV)-infected patients with hepatocellular carcinoma (HCC).

METHODS

A multicenter database of patients undergoing hepatectomy for HCC were analyzed; 2/3 and 1/3 of patients were assigned to the training and validation cohorts, respectively. Independent risks of postoperative 30-day major complications (Clavien-Dindo grades III-V) were identified and used to construct a web-based prediction model, which predictive accuracy was assessed using C-index and calibration curves, which was further validated by the validation cohort and compared with conventional scores.

RESULTS

Among 2762 patients, 391 (14.2%) developed major complications after hepatectomy. Diabetes mellitus, concurrent hepatitis C virus infection, HCC beyond the Milan criteria, cirrhosis, preoperative HBV-DNA level, albumin-bilirubin (ALBI), and aspartate transaminase to platelet ratio index (APRI) were identified as independent predictors of developing major complications, which were used to construct the online calculator ( http://www.asapcalculate.top/Cal11_en.html ). This model demonstrated good calibration and discrimination, with the C-indexes of 0.752 and 0.743 in the training and validation cohorts, respectively, which were significantly higher than those conventional scores (the training and validation cohorts: 0.565 ~ 0.650 and 0.568 ~ 0.614, all P < 0.001).

CONCLUSIONS

A web-based prediction model was developed to predict the probability of post-hepatectomy major complications in an individual HBV-infected patient with HCC. It can be used easily in the real-world clinical setting to help management-related decision-making and early warning, especially in areas with endemic HBV infection.

Development and validation of a prognostic nomogram for bone metastasis from lung cancer: A large population-based study

Background

Bone is one of the most common metastatic sites of advanced lung cancer, and the median survival time is significantly shorter than that of patients without metastasis. This study aimed to identify prognostic factors associated with survival and construct a practical nomogram to predict overall survival (OS) in lung cancer patients with bone metastasis (BM).

Methods

We extracted the patients with BM from lung cancer between 2011 and 2015 from the Surveillance, Epidemiology, and End Result (SEER) database. Univariate and multivariate Cox regressions were performed to identify independent prognostic factors for OS. The variables screened by multivariate Cox regression analysis were used to construct the prognostic nomogram. The performance of the nomogram was assessed by receiver operating characteristic (ROC) curve, concordance index (C-index), and calibration curves, and decision curve analysis (DCA) was used to assess its clinical applicability.

Results

A total of 7861 patients were included in this study and were randomly divided into training (n=5505) and validation (n=2356) cohorts using R software in a ratio of 7:3. Cox regression analysis showed that age, sex, race, grade, tumor size, histological type, T stage, N stage, surgery, brain metastasis, liver metastasis, chemotherapy and radiotherapy were independent prognostic factors for OS. The C-index was 0.723 (95% CI: 0.697-0.749) in the training cohorts and 0.738 (95% CI: 0.698-0.778) in the validation cohorts. The AUC of both the training cohorts and the validation cohorts at 3-month (0.842 vs 0.859), 6-month (0.793 vs 0.814), and 1-year (0.776 vs 0.788) showed good predictive performance, and the calibration curves also demonstrated the reliability and stability of the model.

Conclusions

The nomogram associated with the prognosis of BM from lung cancer was a reliable and practical tool, which could provide risk assessment and clinical decision-making for individualized treatment of patients.

Identification of DNA methylation-regulated genes as potential biomarkers for coronary heart disease via machine learning in the Framingham Heart Study

Background

DNA methylation-regulated genes have been demonstrated as the crucial participants in the occurrence of coronary heart disease (CHD). The machine learning based on DNA methylation-regulated genes has tremendous potential for mining non-invasive predictive biomarkers and exploring underlying new mechanisms of CHD.

Results

First, the 2085 age-gender-matched individuals in Framingham Heart Study (FHS) were randomly divided into training set and validation set. We then integrated methylome and transcriptome data of peripheral blood leukocytes (PBLs) from the training set to probe into the methylation and expression patterns of CHD-related genes. A total of five hub DNA methylation-regulated genes were identified in CHD through dimensionality reduction, including ATG7, BACH2, CDKN1B, DHCR24 and MPO. Subsequently, methylation and expression features of the hub DNA methylation-regulated genes were used to construct machine learning models for CHD prediction by LightGBM, XGBoost and Random Forest. The optimal model established by LightGBM exhibited favorable predictive capacity, whose AUC, sensitivity, and specificity were 0.834, 0.672, 0.864 in the validation set, respectively. Furthermore, the methylation and expression statuses of the hub genes were verified in monocytes using methylation microarray and transcriptome sequencing. The methylation statuses of ATG7, DHCR24 and MPO and the expression statuses of ATG7, BACH2 and DHCR24 in monocytes of our study population were consistent with those in PBLs from FHS.

Conclusions

We identified five DNA methylation-regulated genes based on a predictive model for CHD using machine learning, which may clue the new epigenetic mechanism for CHD.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13148-022-01343-2.

Improving breast cancer diagnostics with deep learning for MRI

Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) has a high sensitivity in detecting breast cancer but often leads to unnecessary biopsies and patient workup. We used a deep learning (DL) system to improve the overall accuracy of breast cancer diagnosis and personalize management of patients undergoing DCE-MRI. On the internal test set (n = 3936 exams), our system achieved an area under the receiver operating characteristic curve (AUROC) of 0.92 (95% CI: 0.92 to 0.93). In a retrospective reader study, there was no statistically significant difference (P = 0.19) between five board-certified breast radiologists and the DL system (mean ΔAUROC, +0.04 in favor of the DL system). Radiologists' performance improved when their predictions were averaged with DL's predictions [mean ΔAUPRC (area under the precision-recall curve), +0.07]. We demonstrated the generalizability of the DL system using multiple datasets from Poland and the United States. An additional reader study on a Polish dataset showed that the DL system was as robust to distribution shift as radiologists. In subgroup analysis, we observed consistent results across different cancer subtypes and patient demographics. Using decision curve analysis, we showed that the DL system can reduce unnecessary biopsies in the range of clinically relevant risk thresholds. This would lead to avoiding biopsies yielding benign results in up to 20% of all patients with BI-RADS category 4 lesions. Last, we performed an error analysis, investigating situations where DL predictions were mostly incorrect. This exploratory work creates a foundation for deployment and prospective analysis of DL-based models for breast MRI.

The Application Value of MRI T 2 ∗ WI Radiomics Nomogram in Discriminating Hepatocellular Carcinoma from Intrahepatic Cholangiocarcinoma

Objective. To establish and validate an MRI T $2^{\ast }$ WI-based radiomics nomogram model and to discriminate hepatocellular carcinoma (HCC) from intrahepatic cholangiocarcinoma (ICCA). Methods. 174 patients were retrospectively collected, who were diagnosed with primary hepatic carcinoma by surgery or puncture pathology and received preoperative MRI scans including T $2^{\ast }$ WI scans. There were 113 cases of HCC and 61 cases of mass-type ICCA. T $2^{\ast }$ WI was used for feature extraction, the extent of the lesions was manually outlined at the largest lesions layer of the T $2^{\ast }$ WI, and the feature dimension reduction was performed by the mRMR and LASSO to obtain the optimal feature set. The radiomics features and clinical risk factors were combined to establish the radiomics nomogram model. In both training and validation groups, calibration curves and ROC curves were applied to validate the efficacy of the established model. Finally, calibration curves were applied to assess the degree of fitting and DCA to assess the clinical utility of the established model. Results. The radiomics model had the AUC of 0.90 (95% CI, 0.85–0.96) and 0.91 (95% CI, 0.83–0.99) in the training and validation groups, respectively; the AUC of the radiomics nomogram was 0.97 (95% CI, 0.94–0.99) in the training group and 0.95 (95% CI, 0.95–0.99) in the validation group. DCA suggested the clinical application value of the nomogram model. Conclusion. Radiomics nomogram model based on MRI T $2^{\ast }$ WI scan without enhancement can be used to discriminate HCC from ICCA.

Novel immune scoring dynamic nomograms based on B7-H3, B7-H4, and HHLA2: Potential prediction in survival and immunotherapeutic efficacy for gallbladder cancer

Background

Gallbladder cancer (GBC) is a mortal malignancy with limited therapeutic strategies. We aimed to develop novel immune scoring systems focusing on B7-H3, B7-H4, and HHLA2. We further investigated their potential clinical effects in predicting survival and immunotherapeutic efficacy for GBC.

Methods

This was a retrospective cohort study in a single center that explored the expression characteristics of B7-H3, B7-H4, and HHLA2. The immune scoring nomograms for prognostic were developed via logistic regression analyses. Their performance was evaluated using the Harrell concordance index (C-index) and decision curves analysis (DCA), and validated with calibration curves.

Results

B7-H3, B7-H4, and HHLA2 manifested with a relatively high rate of co-expression patterns in GBC tissues. They were associated with worse clinicopathological stage, suppression of immune microenvironment, and unfavorable prognosis in postoperative survival. B7 stratification established based on B7-H3, B7-H4, and HHLA2 was an independent prognostic predictor (p&lt;0.05 in both groups). Moreover, immune stratification was also successfully constructed based on B7 stratification and the density of CD8+ TILs (all p&lt;0.001). The prediction models were developed based on B7-/or immune stratification combined with the TNM/or Nevin staging system. These novel models have excellent discrimination ability in predicting survival and immunotherapeutic efficacy for GBC patients by DCA and clinical impact plots. Finally, dynamic nomograms were developed for the most promising clinical prediction models (B7-TNM model and Immune-TNM model) to facilitate prediction.

Conclusions

Immune scoring systems focusing on B7-H3, B7-H4, and HHLA2 may effectively stratify the prognosis of GBC. Prognostic nomograms based on novel immune scoring systems may potentially predict survival and immunotherapeutic efficacy in GBC. Further valid verification is necessary.

Clinically Significant Risk Thresholds in the Management of Primary Cutaneous Melanoma: A Survey of Melanoma Experts

BACKGROUND

Risk-based thresholds to guide management are undefined in the treatment of primary cutaneous melanoma but are essential to advance the field from traditional stage-based treatment to more individualized care.

METHODS

To estimate treatment risk thresholds, hypothetical clinical melanoma scenarios were developed and a stratified random sample was distributed to expert melanoma clinicians via an anonymous web-based survey. Scenarios provided a defined 5-year risk of recurrence and asked for recommendations regarding clinical follow-up, imaging, and adjuvant therapy. Marginal probability of response across the spectrum of 5-year recurrence risk was estimated. The risk at which 50% of respondents recommended a treatment was defined as the risk threshold.

RESULTS

The overall response rate was 56% (89/159). Three separate multivariable models were constructed to estimate the recommendations for clinical follow-up more than twice/year, for surveillance cross-sectional imaging at least once/year, and for adjuvant therapy. A 36% 5-year risk of recurrence was identified as the threshold for recommending clinical follow-up more than twice/year. The thresholds for recommending cross-sectional imaging and adjuvant therapy were 30 and 59%, respectively. Thresholds varied with the age of the hypothetical patient: at younger ages they were constant but increased rapidly at ages 60 years and above.

CONCLUSIONS

To our knowledge, these data provide the first estimates of clinically significant treatment thresholds for patients with cutaneous melanoma based on risk of recurrence. Future refinement and adoption of thresholds would permit assessment of the clinical utility of novel prognostic tools and represents an early step toward individualizing treatment recommendations.

The value of combined PET/MRI, CT and clinical metabolic parameters in differentiating lung adenocarcinoma from squamous cell carcinoma

Objective

This study aimed to study the diagnostic efficacy of positron emission tomography (PET)/magnetic resonance imaging (MRI), computed tomography (CT) and clinical metabolic parameters in predicting the histological classification of lung adenocarcinoma (ADC) and squamous cell carcinoma (SCC).

Methods

PET/MRI, CT and clinical metabolic data of 80 patients with lung ADC or SCC were retrospectively collected. According to the pathological results from surgery or fiberscopy, the patients were diagnosed with lung ADC (47 cases) or SCC (33 cases). All 80 patients were divided into a training group (64 cases), an internal testing group (8 cases) and an external testing group (8 cases) in the ratio of 8:1:1. Nine models were constructed by integrating features from different modalities. The Gaussian classifier was used to differentiate ADC and SCC. The prediction ability was evaluated using the receiver operating characteristic curve. The area under the curve (AUC) of the models was compared using Delong’s test. Based on the best composite model, a nomogram was established and evaluated with a calibration curve, decision curve and clinical impact curve.

Results

The composite model (PET/MRI + CT + Clinical) owned the highest AUC values in the training, internal testing and external testing sets, respectively. In the training set, significant differences in the AUC were found between the composite model and other models except for the PET/MRI + CT model. The calibration curves showed good consistency between the predicted output and actual disease. The decision curve analysis and clinical impact curves demonstrated that the composite model increased the clinical net benefit for predicting lung cancer subtypes.

Conclusion

The composite prediction model of PET/MRI + CT + Clinical better distinguished ADC from SCC pathological subtypes preoperatively and achieved clinical benefits, thus providing an accurate clinical diagnosis.

Development and validation of outcome prediction models for acute kidney injury patients undergoing continuous renal replacement therapy

Object

This study aimed to develop and validate a set of practical predictive tools that reliably estimate the 28-day prognosis of acute kidney injury patients undergoing continuous renal replacement therapy.

Methods

The clinical data of acute kidney injury patients undergoing continuous renal replacement therapy were extracted from the Medical Information Mart for Intensive Care IV database with structured query language and used as the development cohort. An all-subset regression was used for the model screening. Predictive models were constructed via a logistic regression, and external validation of the models was performed using independent external data.

Results

Clinical prediction models were developed with clinical data from 1,148 patients and validated with data from 121 patients. The predictive model based on seven predictors (age, vasopressor use, red cell volume distribution width, lactate, white blood cell count, platelet count, and phosphate) exhibited good predictive performance, as indicated by a C-index of 0.812 in the development cohort, 0.811 in the internal validation cohort and 0.768 in the external validation cohort.

Conclusions

The model reliably predicted the 28-day prognosis of acute kidney injury patients undergoing continuous renal replacement therapy. The predictive items are readily available, and the web-based prognostic calculator (https://libo220284.shinyapps.io/DynNomapp/) can be used as an adjunctive tool to support the management of patients.

Resilience index improves prediction of 1-year decreased quality of life in breast cancer

BACKGROUND

Resilience is important in cancer survivorship and has great potential to predict long-term quality of life (QoL) in breast cancer. The study was designed to develop a new prediction model to estimate pretest probability (PTP) of 1-year decreased QoL combing Resilience Index (RI) and conventional risk factors.

METHODS

RI was extracted from 10-item Resilience Scale Specific to Cancer (RS-SC-10) based on the Principal Component Analysis (PCA). Patients were enrolled from Be Resilient to Breast Cancer (BRBC) and the prediction model was developed based on a sample of 506 consecutive patients and validated in an internal cohort (N1 = 314) and two external cohorts (N2 = 223 and N3 = 189). Integrated Discrimination Improvement (IDI) and Net Reclassification Improvement (NRI) were utilized to estimate the incremental value of RI.

RESULTS

RI improved prediction above conventional risk factors. AUC increased from 0.745 to 0.862 while IDI and NRI were 8.39% and 18.44% respectively (P < 0.0001 for all). Five predictors were included in the final model: RI, age, N stage, M stage, and baseline QoL. The new model demonstrated good calibration ability in the internal and external cohorts resulting in C-indexes of 0.862 (95%CI, 0.815-0.909), 0.828 (95%CI, 0.745-0.910), 0.880 (95%CI, 0.816-0.944), and 0.869 (95%CI, 0.796-0.941).

CONCLUSION

RI contributed to a more accurate estimation for PTP of 1-year decreased QoL above conventional risk factors and could help optimize decision making of treatment for breast cancer.

IMPLICATIONS FOR CANCER SURVIVORS

A promising prognostic indicator of RI could improve QoL-related management in Chinese patients with breast cancer.

Biomarker-based risk model to predict cardiovascular events in patients with acute coronary syndromes - Results from BIPass registry

BACKGROUND

Risk models integrating new biomarkers to predict cardiovascular events in acute coronary syndromes (ACS) are lacking. Therefore, we evaluated the prognostic value of biomarkers in addition to clinical predictors and developed a biomarker-based risk model for major adverse cardiovascular events (MACE) within 12 months after hospital admission with ACS.

METHODS

Patients (n = 4407) consecutively enrolled from November, 2017 to October, 2019 in three hospitals of a prospective Chinese registry (BIomarker-based Prognostic Assessment for Patients with Stable Angina and Acute Coronary Syndromes, BIPass) were designated as the risk model development cohort. Validation was performed in 1409 patients enrolled in two independent hospitals. Cox proportional hazards regression analysis was used to generate a risk prediction model and evaluate the incremental prognostic value of each biomarker.

FINDINGS

Over 12 months, 196 patients experienced MACE (5.1%/year). Among twelve candidate biomarkers, N-terminal pro-B-type natriuretic peptide (NT-proBNP) measured at baseline showed the most prognostic capability independent of clinical predictors. The developed BIPass risk model included age, hypertension, previous myocardial infarction, stroke, Killip class, heart rate, and NT-proBNP. It displayed improved discrimination (C-statistic 0.79, 95% CI 0.73-0.85), calibration (GOF = 9.82, p = 0.28) and clinical decision curve in the validation cohort, outperforming the GRACE and TIMI risk scores. Cumulative rates for MACE demonstrated good separation in the BIPass predicted low, intermediate, and high-risk groups.

INTERPRETATION

The BIPass risk model, integrating clinical variables and NT-proBNP, is useful for predicting 12-month MACE in ACS. It effectively identifies a gradient risk of cardiovascular events to aid personalized care.

FUNDING

National Key R&D Program of China (2017YFC0908700, 2020YFC0846600), National S&T Fundamental Resources Investigation Project (2018FY100600, 2018FY100602), Taishan Pandeng Scholar Program of Shandong Province (tspd20181220), Taishan Young Scholar Program of Shandong Province (tsqn20161065, tsqn201812129), Youth Top-Talent Project of National Ten Thousand Talents Plan and Qilu Young Scholar Program.

A Novel Tool to Predict the Overall Survival of High-Grade Osteosarcoma Patients after Neoadjuvant Chemotherapy: A Large Population-Based Cohort Study

Background. The goal of this study was to discover clinical factors linked to overall survival in patients with high-grade osteosarcoma who had received neoadjuvant therapy and to develop a prognostic nomogram and risk classification system. Methods. A total of 762 patients with high-grade osteosarcoma were included in this study. In the training cohort, Cox regression analysis models were used to find prognostic variables that were independently linked with overall survival. To predict overall survival at 3, 5, and 8 years, a nomogram is created. In addition, in both the internal and external validation cohorts, receiver operating characteristic curves, calibration curves, and decision curve analysis (DCA) were utilized to assess the prediction model’s performance. Results. The age, size of the tumor, and the stage of the disease are all important predictive variables for overall survival. The training and validation cohorts have C-indexes of 0.699 and 0.669, respectively. At the same time, the area under the curve values for both cohorts also showed that the nomogram had good discriminatory power. The calibration curve demonstrated the good performance and predictive accuracy of the model. The DCA results suggest that the nomogram has a wide range of therapeutic applications. Furthermore, a new risk classification system based on the nomogram was established, which allows all patients to be classified into three subgroups as high, middle, and low risk of death. Conclusion. The prognostic nomogram constructed in this study may provide a better precise prognostic prediction for patients with high-grade osteosarcoma after neoadjuvant chemotherapy.

Serine Protease Inhibitor Kazal Type 1, A Potential Biomarker for the Early Detection, Targeting, and Prediction of Response to Immune Checkpoint Blockade Therapies in Hepatocellular Carcinoma

Background

We aimed to characterize serine protease inhibitor Kazal type 1 (SPINK1) as a gene signature for the early diagnosis, molecular targeting, and prediction of immune checkpoint blockade (ICB) treatment response of hepatocellular carcinoma (HCC).

Methods

The transcriptomics, proteomics, and phenotypic analyses were performed separately or in combination.

Results

We obtained the following findings on SPINK1. Firstly, in the transcriptomic training dataset, which included 279 stage I and II tumor samples (out of 1,884 stage I–IV HCC specimens) and 259 normal samples, significantly higher area under curve (AUC) values and increased integrated discrimination improvement (IDI) and net reclassification improvement (NRI) were demonstrated for HCC discrimination in SPINK1-associated models compared with those of alpha-fetoprotein (AFP). The calibration of both SPINK1-related curves fitted significantly better than that of AFP. In the two independent transcriptomic validation datasets, which included 201, 103 stage I-II tumor and 192, 169 paired non-tumor specimens, respectively, the obtained results were consistent with the above-described findings. In the proteomic training dataset, which included 98 stage I and II tumor and 165 normal tissue samples, the analyses also revealed better AUCs and increased IDI and NRI in the aforementioned SPINK1-associated settings. A moderate calibration was shown for both SPINK1-associated models relative to the poor results of AFP. Secondly, in the in vitro and/or in vivo murine models, the wet-lab experiments demonstrated that SPINK1 promoted the proliferation, clonal formation, migration, chemoresistance, anti-apoptosis, tumorigenesis, and metastasis of HCC cells, while the anti-SPINK1 antibody inhibited the growth of the cells, suggesting that SPINK1 has “tumor marker” and “targetable” characteristics in the management of HCC. The Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses revealed that SPINK1 was engaged in immunity-related pathways, including T-cell activation. Thirdly, in the transcriptomic analyses of the 368 HCC specimens from The Cancer Genome Atlas (TCGA) cohort, the high abundance of SPINK1 was positively correlated with the high levels of activated tumor-infiltrating CD4⁺ and CD8⁺ T lymphocytes and dendritic and natural killer cells, while there were also positive correlations between SPINK1 and immune checkpoints, including PD-1, LAG-3, TIM-3, TIGIT, HAVCR2, and CTLA-4. The ESTIMATE algorithm calculated positive correlations between SPINK1 and the immune and ESTIMATE scores, suggesting a close correlation between SPINK1 and the immunogenic microenvironment within HCC tissues, which may possibly help in predicting the response of patients to ICB therapy.

Conclusions

SPINK1 could be a potential biomarker for the early detection, targeted therapy, and prediction of ICB treatment response in the management of HCC.

Lactate dehydrogenase as promising marker for prognosis of brain metastasis

BACKGROUND

Lactate dehydrogenase (LDH) is a biomarker for cancer. However, the relationship between serum LDH levels and the survival of patients with brain metastasis has been fully revealed. We aimed to evaluate the serum LDH levels and assess its prognostic value in patients with BM.

METHODS

The serum LDH levels were collected from 2507 patients with BM. Patients were categorized into four groups according to the quartile of serum LDH levels. The association between serum LDH levels and overall survival (OS) was evaluated using Cox regression models and Kaplan-Meier curves. Three predictive models were used to evaluate patients.

RESULTS

The Kaplan-Meier curve for survival by the serum LDH group demonstrates clear separation between four groups (P < 0.001). The participants in the lower group had longer OS than those in the higher group. After adjusting in multivariate Cox regression models remained significant for patients in the Q4 compared with patients in the Q1 (Q4:Q1 OR 1.58, 95% CI 1.38-1.80). Furthermore, the GPA-LDH model generates a pooled area under the curve of 0.630 (95% CI 0.600, 0.660).

CONCLUSIONS

Serum LDH levels and OS in patients with brain metastasis is an inverse association. Moreover, Serum LDH levels can improve the prognosis of the GPA model.

Analysis of Pyroptosis-Related Immune Signatures and Identification of Pyroptosis-Related LncRNA Prognostic Signature in Clear Cell Renal Cell Carcinoma

Clear cell renal cell carcinoma (ccRCC) is a common urinary system malignant tumor with a high incidence and recurrence rate. Pyroptosis is a kind of programmed cell death caused by inflammasomes. More and more evidence had confirmed that pyroptosis plays a very significant part in cancer, and it is controversial whether pyroptosis promotes or inhibits tumors. Consistently, its potential role in ccRCC treatment efficacy and prognosis remains unclear. In this study, we systematically investigated the role of pyroptosis in the ccRCC samples from The Cancer Genome Atlas (TCGA) database. Based on the differentially expressed pyroptosis-related genes (DEPRGs), we identified three pyroptosis subtypes with different clinical outcomes, immune signatures, and responses to immunotherapy. Gene set variation analysis (GSVA), Gene Ontology (GO) analysis, and Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis revealed that pyroptosis activation meant infiltration of more immune cells that is conducive to tumor progression. To further investigate the immunomodulatory effect of pyroptosis in ccRCC, we constructed a pyroptosis-score based on the common differential prognostic genes of the three pyroptosis subtypes. It was found that patients with high pyroptosis-score were in an unfavorable immune environment and the prognosis was worse. Gene set enrichment analysis suggested that immune-related biological processes were activated in the high pyroptosis-score group. Then, the least absolute shrinkage and selection operator (LASSO) Cox regression was implemented for constructing a prognostic model of eight pyroptosis-related long noncoding RNAs (PRlncRNAs) in the TCGA dataset, and the outcomes revealed that, compared with the low-risk group, the model-based high-risk group was intently associated with poor overall survival (OS). We further explored the relationship between high- and low-risk groups with tumor microenvironment (TME), immune infiltration, and drug therapy. Finally, we constructed and confirmed a robust and reliable PRlncRNA pairs prediction model of ccRCC, identified PRlncRNA, and verified it by experiments. Our findings suggested the potential role of pyroptosis in ccRCC, offering new insights into the prognosis of ccRCC and guiding effectual targeted therapy and immunotherapy.

A Nomogram Model to Predict Early Recurrence of Patients With Intrahepatic Cholangiocarcinoma for Adjuvant Chemotherapy Guidance: A Multi-Institutional Analysis

Background

The influence of different postoperative recurrence times on the efficacy of adjuvant chemotherapy (ACT) for intrahepatic cholangiocarcinoma (ICC) remains unclear. This study aimed to investigate the independent risk factors and establish a nomogram prediction model of early recurrence (recurrence within 1 year) to screen patients with ICC for ACT.

Methods

Data from 310 ICC patients who underwent radical resection between 2010 and 2018 at eight Chinese tertiary hospitals were used to analyze the risk factors and establish a nomogram model to predict early recurrence. External validation was conducted on 134 patients at the other two Chinese tertiary hospitals. Overall survival (OS) and relapse-free survival (RFS) were estimated by the Kaplan–Meier method. Multivariate analysis was conducted to identify independent risk factors for prognosis. A logistic regression model was used to screen independent risk variables for early recurrence. A nomogram model was established based on the above independent risk variables to predict early recurrence.

Results

ACT was a prognostic factor and an independent affecting factor for OS and RFS of patients with ICC after radical resection (p < 0.01). The median OS of ICC patients with non-ACT and ACT was 14.0 and 15.0 months, and the median RFS was 6.0 and 8.0 months for the early recurrence group, respectively (p > 0.05). While the median OS of ICC patients with non-ACT and ACT was 41.0 and 84.0 months, the median RFS was 20.0 and 45.0 months for the late recurrence group, respectively (p < 0.01). CA19-9, tumor size, major vascular invasion, microvascular invasion, and N stage were the independent risk factors of early recurrence for ICC patients after radical resection. The C-index of the nomogram was 0.777 (95% CI: 0.713~0.841) and 0.716 (95%CI: 0.604~0.828) in the training and testing sets, respectively.

Conclusion

The nomogram model established based on the independent risk variables of early recurrence for curatively resected ICC patients has a good prediction ability and can be used to screen patients who benefited from ACT.

Prospective external validation of biomarkers to predict acute graft-versus host disease severity

Biomarker panels at GVHD onset, independently from clinical parameters, were associated with survival and nonrelapse mortality.

Using 3 different biostatistical tools, biomarkers only slightly improved prediction over clinical parameters.

Acute graft-versus-host disease (GVHD) is still the major contributor to comorbidities and mortality after allogeneic hematopoietic stem cell transplantation. The use of plasmatic biomarkers to predict early outcomes has been advocated in the past decade. The purpose of this prospective noninterventional study was to test the ability of panels including 7 biomarkers (Elafin, HGF, IL2RA, IL8, REG3, ST2, and TNFRI), to predict day 28 (D28) complete response to steroid, D180 overall survival, and D180 nonrelapse mortality (NRM). Using previous algorithms developed by the Ann Arbor/MAGIC consortium, 204 patients with acute GVHD were prospectively included and biomarkers were measured at GVHD onset for all of them. Initial GVHD grade and bilirubin level were significantly associated with all those outcomes. After adjustment on clinical variables, biomarkers were associated with survival and NRM. In addition to clinical variables, biomarkers slightly improved the prediction of overall survival and NRM (concordance and net reclassification indexes). The potential benefit of adding biomarkers panel to clinical parameters was also investigated by decision curve analyses. The benefit of adding biomarkers to clinical parameters was however marginal for the D28 nonresponse and mortality endpoints.

Comprehensive analyses indicated the association between m6A related long non-coding RNAs and various pathways in glioma

BACKGROUND

Glioma is one of the most malignant brain tumors and diseases. N6-methyladenosine modification (m6A) is the most abundant and prevalent internal chemical modification of mRNA and long non-coding RNAs (lncRNAs) in eukaryotes. Nevertheless, the correlated pathways and clinical utilization of m6A-related lncRNAs have not been fully evaluated in glioma.

METHODS

Public RNA-sequencing and clinical annotation data were retrieved from TCGA, CGGA and GEO database. Differential expression analysis and univariate Cox regression analysis were performed to identify the m6A-related and differentially expressed lncRNAs with prognostic function (m6A-DELPF). The consensus clustering was performed to identify the expression pattern of m6A-DELPF. LASSO Cox regression analysis was performed to construct the lncRNA-based signature. The CIBERSORT and ESTIMATE algorithms were performed to analyze immune infiltration and tumor microenvironment, respectively. Immunotherapy sensitivity analysis was performed using data from TCIA. The small molecule drugs prediction analysis was performed using The Connectivity Map (CMap) database and STITCH database. A competing endogenous RNAs (ceRNA) network was constructed based on miRcode, miRDB, miRTarBase, TargetScan database.

RESULTS

Two clusters (cluster1 and cluster2) were identified after unsupervised cluster analysis based on m6A-DELPF. Additionally, a 15-gene prognostic signature namely m6A-DELPFS was constructed. Analyses of epithelial-mesenchymal-transition score, tumor microenvironment, immune infiltration, clinical characterization analysis, and putative drug prediction were performed to confirm the clinical utility and efficacy of m6A-DELPFS. The potential mechanisms including tumor immune microenvironment of m6A-DELPF influence the initiation and progression of glioma. A clinically accessible nomogram was also constructed based on the m6A-DELPF and other survival-relevant clinical parameters. Two miRNAs and 114 mRNAs were identified as the downstream of seven m6A-related lncRNAs in a ceRNA network.

CONCLUSION

Our present research confirmed the clinical value of m6A related lncRNAs and their high correlation with tumor immunity, tumor microenvironment, tumor mutation burden and drug sensitivity in glioma.

Utility of a Model for Predicting the Risk of Sentinel Lymph Node Metastasis in Patients With Cutaneous Melanoma

Importance

A neural network-based model (i31-GEP-SLNB) that uses clinicopathologic factors (thickness, mitoses, ulceration, patient age) plus molecular analysis (31-gene expression profiling) has become commercially available to guide selection for sentinel lymph node (SLN) biopsy in cutaneous melanoma, but its clinical utility is not well characterized.

Objective

To determine if use of the i31-GEP-SLNB model is associated with clinical benefit when used to select patients for SLN biopsy.

Design, Setting, and Participants

This decision-analytic study used data derived from a published external validation study of the i31-GEP-SLNB prediction model. Participants included patients with primary cutaneous melanoma.

Main Outcomes and Measures

The primary outcome was the net benefit associated with using the i31-GEP-SLNB model for SLN biopsy selection compared with other selection strategies (SLN biopsy for all patients and SLN biopsy for no patients) at a 5% risk threshold. Analyses were stratified by American Joint Committee on Cancer (AJCC) T category. The reduction in the number of avoidable SLN biopsies and relative utility were also calculated.

Results

Compared with other SLN biopsy selection strategies, use of the i31-GEP-SLNB model had greater net benefit for patients with T1b (+0.012), T2a (+0.002), and T2b melanoma (+0.002) but not for those with high-risk T1a (-0.003) disease. The improvement in relative utility was +22% in patients with T1b, +1% in T2a, and +2% in T2b melanoma. Compared with SLN biopsy for all patients, use of the model would equate to a 23% decrease in SLN biopsies among patients with T1b disease without an SLN metastasis with no increase in the number of patients with an SLN metastasis left untreated; among patients with T2a and T2b melanoma, the net decrease in avoidable biopsies compared with SLN biopsy for all was 3% and 4%, respectively.

Conclusions and Relevance

The findings of this decision-analytic study suggest that i31-GEP SLNB has significant potential for risk-stratifying patients with T1b melanoma if using a 5% risk threshold; its role among patients with T1a and T2 melanoma or using other risk thresholds requires further study. A prospective validation study confirming the added clinical benefit and cost-effectiveness of i31-GEP-SLNB compared with free clinicopathologic-based prediction models is needed in patients with T1b melanoma.

Integrative analysis of immune-related multi-omics profiles identifies distinct prognosis and tumor microenvironment patterns in osteosarcoma

Osteosarcoma (OS) is the most common primary malignancy of bone. Epigenetic regulation plays a pivotal role in cancer development in various aspects, including immune response. In this study, we studied the potential association of alterations in the DNA methylation and transcription of immune-related genes with changes in the tumor microenvironment (TME) and tumor prognosis of OS. We obtained multi-omics data for OS patients from the Therapeutically Applicable Research to Generate Effective Treatments (TARGET) and Gene Expression Omnibus (GEO) databases. By referring to curated immune signatures and using a consensus clustering method, we categorized patients based on immune-related DNA methylation patterns (IMPs), and evaluated prognosis and TME characteristics of the resulting patient subgroups. Subsequently, we used a machine-learning approach to construct an IMP-associated prognostic risk model incorporating the expression of a six-gene signature (MYC, COL13A1, UHRF2, MT1A, ACTB, and GBP1), which was then validated in an independent patient cohort. Furthermore, we evaluated TME patterns, transcriptional variation in biological pathways, somatic copy number alteration, anticancer drug sensitivity, and potential responsiveness to immune checkpoint inhibitor therapy with regard to our IMP-associated signature scoring model. By integrative IMP and transcriptomic analysis, we uncovered distinct prognosis and TME patterns in OS. Finally, we constructed a classifying model, which may aid in prognosis prediction and provide a potential rationale for targeted- and immune checkpoint inhibitor therapy in OS.

Collagen signature as a novel biomarker to predict axillary lymph node metastasis in breast cancer using multiphoton microscopy

Accurate identification of axillary lymph node (ALN) status is crucial for tumor staging procedure and decision making. This retrospective study of 898 participants from two institutions was conducted. The aim of this study is to evaluate the diagnostic performance of clinical parameters combined with collagen signatures (tumor-associated collagen signatures [TACS] and the TACS corresponding microscopic features [TCMF]) in predicting the probability of ALN metastasis in patients with breast cancer. These findings suggest that TACS and TCMF in the breast tumor microenvironment are both novel and independent biomarkers for the estimation of ALN metastasis. The nomogram based on independent clinical parameters combined with TACS and TCMF yields good diagnostic performance in predicting ALN status.

Establishment and Application of a Novel Difficulty Scoring System for da Vinci Robotic Pancreatoduodenectomy

Background

Robotic pancreatoduodenectomy (RPD) technology is developing rapidly, but there is still a lack of a specific and objective difficulty evaluation system in the field of application and training of RPD surgery.

Methods

The clinical data of patients who underwent RPD in our hospital from November 2014 to October 2020 were analyzed retrospectively. Univariate and multivariate logistic regression analyses were used to determine the predictors of operation difficulty and convert into a scoring system.

Results

A total of 72 patients were enrolled in the group. According to the operation time (25%), intraoperative blood loss (25%), conversion to laparotomy, and major complications, the difficulty of operation was divided into low difficulty (0–2 points) and high difficulty (3–4 points). The multivariate logistic regression model included the thickness of mesenteric tissue (P1) (P = 0.035), the thickness of the abdominal wall (B1) (P = 0.017), and the preoperative albumin (P = 0.032), and the nomogram was established. AUC = 0.773 (0.645–0.901).

Conclusions

The RPD difficulty evaluation system based on the specific anatomical relationship between da Vinci’s laparoscopic robotic arm and tissues/organs in the operation area can be used as a predictive tool to evaluate the surgical difficulty of patients before operation and guide clinical practice.

Decisions are not all equal-Introducing a utility metric based on case-wise raters' perceptions

Background and Objective Evaluation of AI-based decision support systems (AI-DSS) is of critical importance in practical applications, nonetheless common evaluation metrics fail to properly consider relevant and contextual information. In this article we discuss a novel utility metric, the weighted Utility (wU), for the evaluation of AI-DSS, which is based on the raters' perceptions of their annotation hesitation and of the relevance of the training cases. Methods We discuss the relationship between the proposed metric and other previous proposals; and we describe the application of the proposed metric for both model evaluation and optimization, through three realistic case studies. Results We show that our metric generalizes the well-known Net Benefit, as well as other common error-based and utility-based metrics. Through the empirical studies, we show that our metric can provide a more flexible tool for the evaluation of AI models. We also show that, compared to other optimization metrics, model optimization based on the wU can provide significantly better performance (AUC 0.862 vs 0.895, p-value <0.05), especially on cases judged to be more complex by the human annotators (AUC 0.85 vs 0.92, p-value <0.05). Conclusions We make the point for having utility as a primary concern in the evaluation and optimization of machine learning models in critical domains, like the medical one; and for the importance of a human-centred approach to assess the potential impact of AI models on human decision making also on the basis of further information that can be collected during the ground-truthing process.