Prognostic factors and clinical nomogram predicting survival in high-grade glioma

Development and validation of a deep learning-based survival prediction model for pediatric glioma patients: A retrospective study using the SEER database and Chinese data

OBJECTIVE

Develop a time-dependent deep learning model to accurately predict the prognosis of pediatric glioma patients, which can assist clinicians in making precise treatment decisions and reducing patient risk.

STUDY DESIGN

The study involved pediatric glioma patients from the Surveillance, Epidemiology, and End Results (SEER) Registry (2000-2018) and Tangdu Hospital in China (2010-2018) within specific time frames. For training, we selected two neural network-based algorithms (DeepSurv, neural multi-task logistic regression [N-MTLR]) and one ensemble learning-based algorithm (random survival forest [RSF]). Additionally, a multivariable Cox proportional hazard (CoxPH) model was developed for comparison purposes. The SEER dataset was randomly divided into 80 % for training and 20 % for testing, while the Tangdu Hospital dataset served as an external validation cohort. Super-parameters were fine-tuned through 1000 repeated random searches and 5-fold cross-validation on the training cohort. Model performance was assessed using the concordance index (C-index), Brier score, and Integrated Brier Score (IBS). Furthermore, the accuracy of predicting survival at 1, 3, and 5 years was evaluated using receiver operating characteristic (ROC) curves, calibration curves, and the area under the ROC curves (AUC). The generalization ability of the model was assessed using the C-index of the Tangdu Hospital data, ROC curves for 1, 3, and 5 years, and AUC values. Lastly, decision curve analysis (DCA) curves for 1, 3, and 5-year time frames are provided to assess the net benefits across different models.

RESULTS

A total of 9532 patients with pediatric glioma were included in this study, comprising 9274 patients from the SEER database and 258 patients from Tangdu Hospital in China. The average age at diagnosis was 9.4 ± 6.2 years, and the average survival time was 96 ± 66 months. Through comprehensive performance comparison, the DeepSurv model demonstrated the highest effectiveness, with a C-index of 0.881 on the training cohort. Furthermore, it exhibited excellent accuracy in predicting the 1-year, 3-year, and 5-year survival rates (AUC: 0.903-0.939). Notably, the DeepSurv model also achieved remarkable performance and accuracy on the Chinese dataset (C-index: 0.782, AUC: 0.761-0.852). Comprehensive analysis of DeepSurv, N-MTLR, and RSF revealed that tumor stage, radiotherapy, histological type, tumor size, chemotherapy, age, and surgical method are all significant factors influencing the prognosis of pediatric glioma. Finally, an online version of the pediatric glioma survival predictor based on the DeepSurv model has been established and can be accessed through https://pediatricglioma-tangdu.streamlit.app.

CONCLUSIONS

The DeepSurv model exhibits exceptional efficacy in predicting the survival of pediatric glioma patients, demonstrating strong performance in discrimination, calibration, stability, and generalization. By utilizing the online version of the pediatric glioma survival predictor, which is based on the DeepSurv model, clinicians can accurately predict patient survival and offer personalized treatment options.

Practical Nomograms and Risk Stratification System for Predicting the Overall and Cancer-specific Survival in Patients with Anaplastic Astrocytoma

OBJECTIVE

Anaplastic astrocytoma (AA) is an uncommon primary brain tumor with highly variable clinical outcomes. Our study aimed to develop practical tools for clinical decision-making in a population-based cohort study.

METHODS

Data from 2997 patients diagnosed with AA between 2004 and 2015 were retrospectively extracted from the Surveillance, Epidemiology, and End Results database. The Least Absolute Shrinkage and Selection Operator and multivariate Cox regression analyses were applied to select factors and establish prognostic nomograms. The discriminatory ability of these nomogram models was evaluated using the concordance index and receiver operating characteristic curve. Risk stratifications were established based on the nomograms.

RESULTS

Selected 2997 AA patients were distributed into the training cohort (70%, 2097) and the validation cohort (30%, 900). Age, household income, tumor site, extension, surgery, radiotherapy, and chemotherapy were identified as independent prognostic factors for both overall survival (OS) and cancer-specific survival (CSS). In the training cohort, our nomograms for OS and CSS exhibited good predictive accuracy with concordance index values of 0.752 (95% CI: 0.741-0.764) and 0.753 (95% CI: 0.741-0.765), respectively. Calibration and decision curve analyses curves showed that the nomograms demonstrated considerable consistency and satisfactory clinical utilities. With the establishment of nomograms, we stratified AA patients into high- and low-risk groups, and constructed risk stratification systems for OS and CSS.

CONCLUSIONS

We constructed two predictive nomograms and risk classification systems to effectively predict the OS and CSS rates in AA patients. These models were internally validated with considerable accuracy and reliability and might be helpful in future clinical practices.

A Study on Prognosis of Diffuse Glioma Based on Clinical Factors and Magnetic Resonance Imaging Radiomics

OBJECTIVE

To construct an optimal prognostic model to assess the prognosis of patients with diffuse glioma.

METHODS

Preoperative magnetic resonance imaging and clinical data were retrospectively collected from 266 patients (training cohort: validation cohort=7:3) with pathologically confirmed diffuse gliomas. A radiomics prognostic model (R-model) based on the radiomics features was constructed. A prognostic model based on clinical factors (C-model) and a fusion model (F-model) was also constructed. Based on the optimal model of three models, the nomogram was constructed. Finally, a "Prognosis Calculator for Diffuse Glioma" was constructed based on the nomogram.

RESULTS

The c-index of the R-, C-, and F-models in the validation cohort was 0.742, 0.796, and 0.814, respectively. In the validation cohort, the 1-year area under the curve of the R-, C-, and F-models was 0.749, 0.806, and 0.836, respectively; the 3-year area under the curve was 0.896, 0.966, and 0.963, respectively. In the training cohort, validation cohort, all cohorts, and different grades of glioma cohorts, F-model (optimal model) could identify low- and high-risk groups well. The "Prognosis Calculator for Diffuse Glioma" was available at https://github.com/HDCurry/prognosis.

CONCLUSIONS

Among the three models, the F-model (radiomics combined with clinical factors) had optimal predictive efficacy and could more accurately assess the prognosis of diffuse glioma. The "Prognosis Calculator for Diffuse Glioma" constructed based on this model could assist clinicians in more easily and accurately assessing the prognosis of patients with diffuse glioma, thus enabling them to make more reasonable treatment strategies.

Prognostic factors and clinical nomogram for in-hospital mortality in traumatic brain injury

BACKGROUND

Traumatic brain injury (TBI) is a major cause of death and functional disability in the general population. The nomogram is a clinical prediction tool that has been researched for a wide range of medical conditions. The purpose of this study was to identify prognostic factors associated with in-hospital mortality. The secondary objective was to develop a clinical nomogram for TBI patients' in-hospital mortality based on prognostic factors.

METHODS

A retrospective cohort study was conducted to analyze 14,075 TBI patients who were admitted to a tertiary hospital in southern Thailand. The total dataset was divided into the training and validation datasets. Several clinical characteristics and imaging findings were analyzed for in-hospital mortality in both univariate and multivariable analyses using the training dataset. Based on binary logistic regression, the nomogram was developed and internally validated using the final predictive model. Therefore, the predictive performances of the nomogram were estimated by the validation dataset.

RESULTS

Prognostic factors associated with in-hospital mortality comprised age, hypotension, antiplatelet, Glasgow coma scale score, pupillary light reflex, basilar skull fracture, acute subdural hematoma, subarachnoid hemorrhage, midline shift, and basal cistern obliteration that were used for building nomogram. The predictive performance of the nomogram was estimated by the training dataset; the area under the receiver operating characteristic curve (AUC) was 0.981. In addition, the AUCs of bootstrapping and cross-validation methods were 0.980 and 0.981, respectively. For the temporal validation with an unseen dataset, the sensitivity, specificity, accuracy, and AUC of the nomogram were 0.90, 0.88, 0.88, and 0.89, respectively.

CONCLUSION

A nomogram developed from prognostic factors had excellent performance; thus, the tool had the potential to serve as a screening tool for prognostication in TBI patients. Furthermore, future research should involve geographic validation to examine the predictive performances of the clinical prediction tool.

E3 ubiquitin-ligase RNF138 may regulate p53 protein expression to regulate the self-renewal and tumorigenicity of glioma stem cells

BACKGROUND

Glioblastoma multiforme (GBM), the most malignant tumor of the central nervous system, is characterized by poor survival and high recurrence. Glioma stem cells (GSCs) are key to treating GBM and are regulated by various signaling pathways. Ubiquitination, a post-translational modification, plays an important regulatory role in many biological processes. Ring finger protein 138 (RNF138) is an E3 ubiquitin-protein ligase that is highly expressed in several tumors; however, its role in GBM is unclear. This study investigated whether RNF138 regulates the self-renewal ability of glioma stem GSCs to treat GBM.

MATERIALS AND METHODS

The expression of RNF138 in glioma tissues and its correlation with GSCs were analyzed using bioinformatics. Short hairpin ribonucleic acid (RNA) was designed to downregulate the expression of RNF138 in GSCs, and immunofluorescence, secondary pellet formation, and western blotting were used to detect changes in GSC markers and self-renewal ability. The effects of RNF138 on p53 protein expression were determined by immunofluorescence and western blotting. The effects of RNF138 on the self-renewal and tumorigenic abilities of GSCs were evaluated in vivo.

RESULTS

RNF138 expression was higher in glioma tissues than in normal brain tissues, and was highly expressed in GSCs. RNF138 downregulation significantly decreased the expression of the GSC markers cluster of differentiation 133 (CD133) and nestin. Mechanistically, RNF138 may interfere with the self-renewal ability of GSCs by regulating the expression of p53. RNF138 downregulation in vivo prolonged survival time and regulated the expression of p53 protein in tumor-bearing mice.

CONCLUSION

RNF138 may regulate the expression of p53 protein through ubiquitination, thereby affecting the self-renewal and tumorigenic ability of GSCs. This study provides a scientific basis for the treatment of glioblastoma by targeting RNF138 to inhibit GSCs.

Forkhead box transcription factors (FOXOs and FOXM1) in glioma: from molecular mechanisms to therapeutics

Glioma is the most aggressive and malignant type of primary brain tumor, comprises the majority of central nervous system deaths, and is categorized into different subgroups according to its histological characteristics, including astrocytomas, oligodendrogliomas, glioblastoma multiforme (GBM), and mixed tumors. The forkhead box (FOX) transcription factors comprise a collection of proteins that play various roles in numerous complex molecular cascades and have been discovered to be differentially expressed in distinct glioma subtypes. FOXM1 and FOXOs have been recognized as crucial transcription factors in tumor cells, including glioma cells. Accumulating data indicates that FOXM1 acts as an oncogene in various types of cancers, and a significant part of studies has investigated its function in glioma. Although recent studies considered FOXO subgroups as tumor suppressors, there are pieces of evidence that they may have an oncogenic role. This review will discuss the subtle functions of FOXOs and FOXM1 in gliomas, dissecting their regulatory network with other proteins, microRNAs and their role in glioma progression, including stem cell differentiation and therapy resistance/sensitivity, alongside highlighting recent pharmacological progress for modulating their expression.

Nomogram Model for Predicting the Prognosis of High-Grade Glioma in Adults Receiving Standard Treatment: A Retrospective Cohort Study

OBJECTIVES

To identify the critical factors associated with the progression-free survival (PFS) and overall survival (OS) of high-grade glioma (HGG) in adults who have received standard treatment and establish a novel graphical nomogram and an online dynamic nomogram.

PATIENTS AND METHODS

This is a retrospective study of adult HGG patients receiving standard treatment (surgery, postoperative radiotherapy, and temozolomide (TMZ) chemotherapy) at Huashan Hospital, Fudan University between January 2017 and December 2019. We used uni- and multi-variable COX models to identify the significant prognostic factors for PFS and OS. Based on the significant predictors, graphical and online nomograms were established.

RESULTS

A total of 246 patients were enrolled in the study based on the inclusion criteria. The average PFS and OS were 22.99 ± 11.43 and 30.51 ± 13.73 months, respectively. According to the multi-variable COX model, age, extent of resection (EOR), and IDH mutation were associated with PFS and OS, while edema index (EI) was relevant to PFS. In addition, patients with IDH and TERT promoter co-mutations had longer PFSs and OSs, and no apparent survival benefit was found in the long-cycle TMZ adjuvant chemotherapy compared with the standard Stupp protocol. Based on these critical factors, a graphical nomogram and online nomogram were developed for predicting PFS and OS, respectively. The calibration curve showed favorable consistency between the predicted and actual survival rates. C-index and time-dependent AUC showed good discrimination abilities.

CONCLUSIONS

We identified the significant predictors for the PFS and OS of HGG adults receiving standard treatment and established user-friendly nomogram models to assist neurosurgeons in optimizing clinical management and treatment strategies.

Comparison of nomogram with random survival forest for prediction of survival in patients with spindle cell carcinoma

Purpose

Spindle cell carcinoma (SpCC) is a relatively rare tumor with an unfavorable prognosis. This study aimed to develop and validate a prediction model for the individual survival of patients with SpCC using Cox regression and the random survival forest (RSF) model.

Methods

Patients diagnosed with SpCC between 2004 and 2016 were selected from the Surveillance, Epidemiology, and End Results (SEER) database, and randomly divided into training and validating cohorts. Cox regression and RSF were used to identify prognostic predictors and build prediction models. A nomogram based on Cox regression was constructed to predict the 1-, 3-, and 5-year survival of patients with SpCC. Internal validation was conducted using the bootstrapping method. We evaluated the discrimination accuracy and calibration of the model using Harrell's C-index and calibration plot, respectively.

Results

Two hundred and fifty patients diagnosed with SpCC with required information were enrolled in this study. Multivariate Cox regression and RSF identified age, primary site, grade, SEER stage, tumor size, and treatment as significant prognostic predictors of SpCC. The bootstrapped and validated C-indices were 0.812 and 0.783 for nomogram, and 0.790 and 0.768 for RSF, respectively. Calibration plot of the nomogram showed an agreement between the prediction and actual observation.

Conclusions

The nomogram developed in this study is a promising tool with a simplified presentation that can easily be used and interpreted by clinicians for evaluating the survival of each patient with SpCC; its performance was comparable to that of RSF. Application of such models are needed to help oncologists identify the high-risk patients and improve clinical decision making of SpCC treatment.

Development and internal validation of a nomogram to predict massive blood transfusions in neurosurgical operations

Objectives

A massive blood transfusion (MBT) is an unexpected event that may impact mortality. Neurosurgical operations are a major operation involving the vital structures and risk to bleeding. The aims of the present research were (1) to develop a nomogram to predict MBT and (2) to estimate the association between MBT and mortality in neurosurgical operations.

Material and Method

We conducted a retrospective cohort study including 3660 patients who had undergone neurosurgical operations. Univariate and multivariate logistic regression analyses were used to test the association between clinical factors, pre-operative hematological laboratories, and MBT. A nomogram was developed based on the independent predictors.

Results

The predictive model comprised five predictors as follows: Age group, traumatic brain injury, craniectomy operation, pre-operative hematocrit, and pre-operative international normalized ratio and the good calibration were observed in the predictive model. The concordance statistic index was 0.703. Therefore, the optimism-corrected c-index values of cross-validation and bootstrapping were 0.703 and 0.703, respectively.

Conclusion

MBT is an unexpectedly fatal event that should be considered for appropriate preparation blood components. Further, this nomogram can be implemented for allocation in limited-resource situations in the future.