Recursive partitioning analysis for survival stratification and early imaging prediction of molecular biomarker in glioma patients.

BACKGROUND: Glioma is the most common primary brain tumor with high mortality and disability rates. Recent studies have highlighted the significant prognostic consequences of subtyping molecular pathological markers using tumor samples, such as IDH, 1p/19q, and TERT. However, the relative importance of individual markers or marker combinations in affecting patient survival remains unclear. Moreover, the high cost and reliance on postoperative tumor samples hinder the widespread use of these molecular markers in clinical practice, particularly during the preoperative period. We aim to identify the most prominent molecular biomarker combination that affects patient survival and develop a preoperative MRI-based predictive model and clinical scoring system for this combination. METHODS: A cohort dataset of 2,879 patients was compiled for survival risk stratification. In a subset of 238 patients, recursive partitioning analysis (RPA) was applied to create a survival subgroup framework based on molecular markers. We then collected MRI data and applied Visually Accessible Rembrandt Images (VASARI) features to construct predictive models and clinical scoring systems. RESULTS: The RPA delineated four survival groups primarily defined by the status of IDH and TERT mutations. Predictive models incorporating VASARI features and clinical data achieved AUC values of 0.85 for IDH and 0.82 for TERT mutations. Nomogram-based scoring systems were also formulated to facilitate clinical application. CONCLUSIONS: The combination of IDH-TERT mutation status alone can identify the most distinct survival differences in glioma patients. The predictive model based on preoperative MRI features, supported by clinical assessments, offers a reliable method for early molecular mutation prediction and constitutes a valuable scoring tool for clinicians in guiding treatment strategies.

Development and Validation of a Predictive Model for Acute Kidney Injury in Sepsis Patients Based on Recursive Partition Analysis.

BACKGROUND: Sepsis-associated acute kidney injury (SA-AKI) is a critical condition with significant clinical implications, yet there is a need for a predictive model that can reliably assess the risk of its development. This study is undertaken to bridge a gap in healthcare by creating a predictive model for SA-AKI with the goal of empowering healthcare providers with a tool that can revolutionize patient care and ultimately lead to improved outcomes. METHODS: A cohort of 615 patients afflicted with sepsis, who were admitted to the intensive care unit, underwent random stratification into 2 groups: a training set (n = 435) and a validation set (n = 180). Subsequently, a multivariate logistic regression model, imbued with nonzero coefficients via LASSO regression, was meticulously devised for the prognostication of SA-AKI. This model was thoughtfully rendered in the form of a nomogram. The salience of individual risk factors was assessed and ranked employing Shapley Additive Interpretation (SHAP). Recursive partition analysis was performed to stratify the risk of patients with sepsis. RESULTS: Among the panoply of clinical variables examined, hypertension, diabetes mellitus, C-reactive protein, procalcitonin (PCT), activated partial thromboplastin time, and platelet count emerged as robust and independent determinants of SA-AKI. The receiver operating characteristic curve analysis for SA-AKI risk discrimination in both the training set and validation set yielded an area under the curve estimates of 0.843 (95% CI: 0.805 to 0.882) and 0.834 (95% CI: 0.775 to 0.893), respectively. Notably, PCT exhibited the most conspicuous influence on the model's predictive capacity. Furthermore, statistically significant disparities were observed in the incidence of SA-AKI and the 28-day survival rate across high-risk, medium-risk, and low-risk cohorts (P < .05). CONCLUSION: The composite predictive model, amalgamating the quintet of SA-AKI predictors, holds significant promise in facilitating the identification of high-risk patient subsets.

The number of cycles of adjuvant chemotherapy in stage III and high-risk stage II rectal cancer: a nomogram and recursive partitioning analysis.

OBJECTIVE: The prognostic role of the number of cycles of adjuvant chemotherapy (ACT) after total mesorectal excision in stage III and high-risk stage II rectal cancer is unknown. As a result of this, our study was designed to assess the effect of the number of cycles of ACT on the prediction of cancer-specific survival. METHODS: Four hundred patients that were diagnosed as stage III and high-risk stage II rectal cancer from January 2012 to January 2018 and who had received total mesorectal excision were enrolled in this study. A nomogram incorporating the number of cycles of ACT was also developed in this study. For internal validation, the bootstrap method was used and the consistency index was used to evaluate the accuracy of the model. The patients were stratified into risk groups according to their tumor characteristics by recursive partitioning analysis. RESULTS: We found that the risk of death was decreased by 26% (HR = 0.74, 95% CI: 0.61-0.89, P = 0.0016) with each increasing ACT cycle. The N stage, positive lymph node ratio (PLNR), carcinoembryonic antigen, neutrophil-to-lymphocyte ratio, and the number of cycles of ACT were chosen and entered into the nomogram model. Recursive partitioning analysis-based risk stratification revealed a significant difference in the prognosis in rectal cancer patients with high-risk, intermediate-risk, and low-risk (3-year cancer-specific survival: 0.246 vs. 0.795 vs. 0.968, P < 0.0001). Seven or more cycles of ACT yielded better survival in patients with PLNR ≥ 0.28 but not in patients with PLNR < 0.28. CONCLUSION: In conclusion, the nomogram prognosis model based on the number of cycles of ACT predicted individual prognosis in rectal cancer patients who had undergone total mesorectal excision. These findings further showed that in patients with PLNR ≥ 0.28, no fewer than 7 cycles of ACT are needed to significantly reduce the patient's risk of death.

Risk factors for lymph node metastasis in rectal neuroendocrine tumors: A recursive partitioning analysis based on multicenter data.

BACKGROUND: The well-differentiated rectal neuroendocrine tumors (RNETs) can also have lymph node metastasis (LNM). Large multicenter data were reviewed to explore the risk factors for LNM in RNETs. Further, we developed a model to predict the risk of LNM in RNETs. METHODS: In total, 223 patients with RNETs from the Fujian Medical University Union Hospital, the First Affiliated Hospital of Fujian Medical University, and the First Affiliated Hospital of Xiamen University were retrospectively enrolled. Logistic regression analysis was performed to study the factors affecting LNM, and recursive partitioning analysis (RPA) was performed to stratify the risk of LNM. RESULTS: Among the 223 patients diagnosed with RNETs, the incidence of LNM was 10.8%. Univariate and multivariate regression analyses revealed that tumor size, World Health Organization (WHO) grade, and depth of tumor invasion were independent risk factors for LNM (p < 0.05). The area under the curve was 0.948 (95% confidence interval: 0.890-1.000). Furthermore, the incidence of LNM in patients divided into low- and high-risk groups according to RPA was 1.1% and 56.4%, respectively. CONCLUSION: Compared with tumor size, the depth of tumor invasion and WHO grade are more important factors in predicting LNM. Then, we developed a model based on RPA to predict the risk of LNM in RNETs and identify patients who are suitable for local resection.

Cumulative Intracranial Tumor Volume as a Prognostic Factor in Patients with Brain Metastases Undergoing Stereotactic Radiosurgery.

Approximately 25-35% of all cancer patients suffer from brain metastases (BM), and many of them-in particular, those with a limited number of intracranial tumors-are treated with stereotactic radiosurgery (SRS). Accurate prediction of survival remains a key clinical challenge in this population. Several prognostic scales have been developed to facilitate this prognostication, including the Recursive Partitioning Analysis (RPA) classification, the modified Recursive Partitioning Analysis (mRPA) subclassifications, the Basic Score for Brain Metastases (BS-BM), the Score Index for Radiosurgery (SIR), the Graded Prognostic Assessment (GPA), and the diagnosis-specific Graded Prognostic Assessment (dsGPA). However, none of these scales include consideration of the cumulative intracranial tumor volume (CITV), which is defined as the sum of all intracranial tumor volumes. Since there is mounting evidence that the CITV carries significant prognostic value in SRS-treated patients with BM, this variable should be considered during survival prognostication, along with other pertinent clinical, pathological, and molecular characteristics.

Integrating postradiotherapy plasma Epstein-Barr virus DNA and TNM stage for risk stratification of nasopharyngeal carcinoma to adjuvant therapy.

BACKGROUND: After curative radiotherapy (RT) or chemoradiation (CRT), there is no validated tool to accurately identify patients for adjuvant therapy in nasopharyngeal carcinoma (NPC). Post-RT circulating plasma Epstein-Barr virus (EBV) DNA can detect minimal residual disease and is associated with recurrence and survival independent of TNM (tumor-lymph node-metastasis) stage. We aimed to develop and validate a risk model for stratification of NPC patients after completion of RT/CRT to observation or adjuvant therapy. PATIENTS AND METHODS: The prospective multicenter 0502 EBV DNA screening cohort (Hong Kong NPC Study Group 0502 trial) enrolled from 2006 to 2015 (n = 745) was used for model development. For internal validation, we pooled independent patient cohorts from prospective clinical studies enrolled from 1997 to 2006 (n = 340). For external validation, we used retrospective cohort of NPC patients treated at Sun Yat-sen University Cancer Center from 2009 to 2012 (n = 837). Eligible patients had histologically confirmed NPC of Union for International Cancer Control (UICC) 7th Edition stage II-IVB who completed curative RT/CRT with or without neoadjuvant chemotherapy, had post-RT EBV DNA tested within 120 days after RT and received no adjuvant therapy. The primary end point was overall survival (OS). We used recursive-partitioning analysis (RPA) to classify patients into groups of low, intermediate, and high risk of death. RESULTS: Combining post-RT EBV DNA level (0, 1-49, 50-499, and ≥500 copies/ml) and TNM stage (II, III, IVAB), RPA model classified patients into low-, intermediate-, and high-risk groups with 5-year OS of 89.4%, 78.5% and 37.2%, respectively. The RPA low-risk group had comparable OS to TNM stage II (5-year OS 88.5%) but identified more patients (64.8% versus stage II 28.1%) that could potentially be spared adjuvant therapy toxicity. The RPA model (c-index 0.712) showed better risk discrimination than either the TNM stage (0.604) or post-RT EBV DNA alone (0.675) with improved calibration and consistence. These results were validated in both internal and external cohorts. CONCLUSION: Combining post-RT EBV DNA and TNM stage improved risk stratification in NPC.

Risk-Stratification of Children Presenting to Ambulatory Paediatrics with First-Onset Seizures: Should We Order an Urgent CT Brain?

INTRODUCTION: A computed tomography (CT) brain scan is an often-utilised emergency department imaging modality to detect emergent intra-cranial pathology in a child with a first seizure. Identifying children at low risk of having a clinically significant intra-cranial abnormality could prevent unnecessary radiation exposure and contrast/sedation-related risks. OBJECTIVES: To identify clinical variables which could predict clinically significant CT brain abnormalities and use recursive partitioning analysis to define a low-risk group of children in whom emergent CT brain can be deferred. METHODS: Retrospective cross-sectional review of 468 children who underwent emergent CT brain after presenting to a low- and middle-income paediatric emergency department following first seizure. RESULTS: In total 133/468 (28.4%) of CT brain scans had clinically significant abnormalities. Failure to return to neurological baseline and focal neurological deficit persisting >36 h had statistical significance in a multiple regression analysis. Recursive partitioning analysis, applied to a subgroup without suspected tuberculous meningitis (n = 414), classified 153 children aged between 6 months and 5 years, who had a normal neurological baseline, had returned to baseline post-seizure, and were not in status epilepticus, as non-clinically significant scans and 98% were correctly classified. CONCLUSION: Our study re-inforces the American Academy of Neurology recommendation that children with persistent post-ictal abnormal neurological status and/or post-ictal focal deficit be prioritised for emergent CT brain. Having excluded children with suspected tuberculous meningitis, the remaining subgroup aged 6 months to 5 years presenting with a non-status first seizure, normal neurological baseline and return to baseline post-seizure, are at very low risk of having a clinically significant CT brain abnormality.

The addition of pretreatment plasma Epstein-Barr virus DNA into the eighth edition of nasopharyngeal cancer TNM stage classification.

The eighth edition of the American Joint Committee on Cancer (AJCC)/Union for International Cancer Control (UICC) stage classification (TNM) for nasopharyngeal carcinoma (NPC) was launched. It remains unknown if incorporation of nonanatomic factors into the stage classification would better predict survival. We prospectively recruited 518 patients with nonmetastatic NPC treated with radical intensity-modulated radiation therapy ± chemotherapy based on the eighth edition TNM. Recursive partitioning analysis (RPA) incorporating pretreatment plasma Epstein-Barr virus (EBV) DNA derived new stage groups. Multivariable analyses to calculate adjusted hazard ratios (AHRs) derived another set of stage groups. Five-year progression-free survival (PFS), overall survival (OS) and cancer-specific survival (CSS) were: Stage I (PFS 100%, OS 90%, CSS 100%), II (PFS 88%, OS 84%, CSS 95%), III (PFS 84%, OS 84%, CSS 90%) and IVA (PFS 71%, OS 75%, CSS 80%) (p < 0.001, p = 0.066 and p = 0.002, respectively). RPA derived four new stages: RPA-I (T1-T4 N0-N2 & EBV DNA <500 copies per mL; PFS 94%, OS 89%, CSS 96%), RPA-II (T1-T4 N0-N2 & EBV DNA ≥500 copies per mL; PFS 80%, OS 83%, CSS 89%), RPA-III (T1-T2 N3; PFS 64%, OS 83%, CSS 83%) and RPA-IVA (T3-T4 N3; PFS 63%, OS 60% and CSS 68%) (all with p < 0.001). AHR using covariate adjustment also yielded a valid classification (I: T1-T2 N0-N2; II: T3-T4 N0-N2 or T1-T2 N3 and III: T3-T4 N3) (all with p < 0.001). However, RPA stages better predicted survival for PS and CSS after bootstrapping replications. Our RPA-based stage groups revealed better survival prediction compared to the eighth edition TNM and the AHR stage groups.

Proposed modifications and incorporation of plasma Epstein-Barr virus DNA improve the TNM staging system for Epstein-Barr virus-related nasopharyngeal carcinoma.

BACKGROUND: The prognosis of patients who have Epstein-Barr virus (EBV)-related nasopharyngeal carcinoma (NPC) in which the tumor tissues harbor EBV have a better prognosis than those without EBV-related NPC. Therefore, the eighth edition of the TNM staging system could be modified for EBV-related NPC by incorporating the measurement of plasma EBV DNA. METHODS: In total, 979 patients with NPC who received intensity-modulated radiotherapy (IMRT) were retrospectively reviewed. Recursive partitioning analysis was conducted based on tumor (T) classification, lymph node (N) classification, and EBV DNA measurement to derive objectively the proposed stage groupings. The validity of the proposed stage groupings was confirmed in a prospective cohort of 550 consecutive patients who also received with IMRT. RESULTS: The pretreatment plasma EBV DNA level was identified as a significant, negative prognostic factor for progression-free survival and overall survival in univariate analysis (all P < .001) and multivariate analysis (all P < .05). Recursive partitioning analysis of the primary cohort to incorporate EBV DNA generated the following proposed stage groupings: stage RI (T1N0), RIIA (T2-T3N0 or T1-T3N1, EBV DNA ≤2000 copies/mL), stage RIIB (T2-T3N0 or T1-T3N1, EBV DNA >2000 copies/mL; T1-T3N2, EBV DNA ≤2000 copies/mL), stage RIII (T1-T3N2, EBV DNA >2000 copies/mL; T4N0-N2), and stage RIVA (any T and N3). In the validation cohort, the 5-year progression-free survival rate was 100%, 87.9%, 76.7%, 68.7%, and 50.4% for proposed stage RI, RIIA, RIIB, RIII, and RIV NPC, respectively (P < .001). Compared with the eighth edition TNM stage groupings, the proposed stage groupings incorporating EBV DNA provided better hazard consistency, hazard discrimination, outcome prediction, and sample size balance. CONCLUSIONS: The proposed stage groupings have better prognostic performance than the eighth edition of the TNM staging system. EBV DNA titers should be included in the TNM staging system to assess patients who have EBV-related NPC.

Recursive partitioning analysis of patients with oligometastatic non-small cell lung cancer: a retrospective study.

BACKGROUND: Local consolidative treatment (LCT) is important for oligometastasis, defined as the restricted metastatic capacity of a tumor. This study aimed to determine the effects and prognostic heterogeneity of LCT in oligometastatic non-small cell lung cancer. METHODS: This retrospective study identified 436 eligible patients treated for oligometastatic disease at the Guangdong Provincial People's Hospital during 2009-2016. A Cox regression analysis was used to identify potential predictors of overall survival (OS). After splitting cases randomly into training and testing sets, risk stratification was performed using recursive partitioning analysis with a training dataset. The findings were confirmed using a validation dataset. The effects of LCT in different risk groups were evaluated using the Kaplan-Meier method. RESULTS: The T stage (p = 0.001), N stage (p = 0.008), number of metastatic sites (p = 0.031), and EGFR status (p = 0.043) were identified as significant predictors of OS. A recursive partitioning analysis was used to establish a prognostic risk model with the following four risk groups: Group I included never smokers with N0 disease (3-year OS: 55.6%, median survival time [MST]: 42.8 months), Group II included never smokers with N+ disease (3-year OS: 32.8%, MST: 26.5 months), Group III included smokers with T0-2 disease (3-year OS: 23.3%, MST: 19.4 months), and Group IV included smokers with T3/4 disease (3-year OS: 12.5%, MST: 11.1 months). Significant differences in OS according to LCT status were observed in all risk groups except Group IV (p = 0.45). CONCLUSIONS: Smokers with T3/4 oligometastatic non-small cell lung cancer may not benefit from LCT.

Risk factors of lymph node metastasis or lymphovascular invasion for early gastric cancer: a practical and effective predictive model based on international multicenter data.

BACKGROUND: Most lymph node metastasis (LNM) models for early gastric cancer (EGC) include lymphovascular invasion (LVI) as a predictor. However, LVI must be confirmed by postoperative pathology. In this study, we aimed to develop a model for predicting the risk of LNM/LVI in EGC using preoperative factors. METHODS: EGC patients who underwent radical gastrectomy at Fujian Medical University Union Hospital and Sun Yat-sen University Cancer Center (n = 1460) were selected as the training set. The risk factors of LNM/LVI were investigated. Data from the International study group on Minimally Invasive surgery for GASTRIc Cancer trial (n = 172) were selected as the validation set. RESULTS: In the training set, the incidence of LNM/LVI was 21.6%. The 5-year cancer-specific survival rates of patients with and without LNM/LVI were 92.4 and 95.0%, respectively, with significant difference (P = 0.030). Multivariable logistic regression analysis showed that the four independent risk factors for LNM/LVI were female, tumor larger than 20 mm, submucosal invasion and undifferentiated tumor histological type (all P <  0.05); the area under the curve (AUC) was 0.694 (95% confidence interval [CI]: 0.659-0.730). Patients were divided into low-risk, intermediate-risk, high-risk and extremely high-risk groups by recursive partitioning analysis; the incidences of LNM/LVI were 5.4, 12.6, 24.2 and 37.8%, respectively (P <  0.001). The AUC of the validation set was 0.796 (95%CI, 0.662-0.851) and the predictive performance of the LNM/LVI risk in the validation set was consistent with that in the training set. CONCLUSIONS: The risk of LNM/LVI in differentiated mucosal EGC is low, which indicated that endoscopic resection is a treatment option. The risk of LNM/LVI in undifferentiated mucosal EGC and submucosa EGC are high and gastrectomy with lymph node dissection is suggested.

A competing-risks nomogram and recursive partitioning analysis for cause-specific mortality in patients with esophageal neuroendocrine carcinoma.

The objective of this study is to estimate the probability of cause-specific mortality using a competing-risks nomogram and recursive partitioning analysis in a large population-based cohort of patients with esophageal neuroendocrine carcinoma. The surveillance, epidemiology and end results database was used to identify 162 patients diagnosed with esophageal neuroendocrine carcinoma from 1998 to 2014. We estimated a cumulative incidence function for cause-specific mortality. A nomogram was constructed by using a proportional subdistribution hazard model, validated using bootstrap cross-validation, and evaluated with a decision curve analysis to assess its clinical utility. Finally, we performed risk stratification using a recursive partitioning analysis to divide patients with esophageal neuroendocrine carcinoma into clinically useful prognostic groups. Tumor location, distant metastasis, surgery, radiotherapy, and chemotherapy were significantly associated with cause-specific mortality. The calibration plots demonstrated good concordance between the predicted and actual outcomes. The discrimination performance of a Fine-Gray model was evaluated by using the c-index, which was 0.723 for cause-specific mortality. The decision curve analysis ranged from 0.268 to 0.968 for the threshold probability at which the risk model provided net clinical benefits relative to hypothetical all-screening and no-screening scenarios. The risk groups stratified by a recursive partitioning analysis allowed significant distinction between cumulative incidence curves. We determined the probability of cause-specific mortality in patients with esophageal neuroendocrine carcinoma and developed a nomogram and recursive partitioning analysis stratification system based on a competing-risks model. The nomogram and recursive partitioning analysis appear to be suitable for risk stratification of cause-specific mortality in patients with esophageal neuroendocrine carcinoma and will help clinicians to identify patients at increased risk of cause-specific mortality to guide treatment and surveillance decisions.

Investigation of genetic variation and lifestyle determinants in vitamin D levels in Arab individuals.

BACKGROUND: Differences in the concentrations of circulating 25-hydroxyvitamin D [25(OH)D] are associated with a wide range of health outcomes; however, most studies on genetic variants that impact 25(OH)D levels have been conducted in European populations. Here we aimed to identify common genetic variants that affect vitamin D concentrations in individuals of self-reported Arab ethnicity. METHODS: The study included 1151 Arab subjects living in Kuwait. Common variants of single-nucleotide polymorphisms and genes previously associated with vitamin D levels, such as GC, PDE3B, CYP2R1, and NADSYN1, were genotyped. Raw vitamin D level data were corrected for age, body mass index, and sex and then normalized. Regression tree analyses were performed to identify the impact of genetic variants on vitamin D levels. RESULTS: Compared with other gene variants, the GC gene variants exhibited the greatest impact on vitamin D levels in our study population, of which rs2298850 had the lowest p value (0.003). Individuals homozygous for the derived allele C had lower vitamin D levels. Analyses of the interaction between the number of years for which the subjects had lived in Kuwait and genetic variation in the GC gene showed that those with the CC genotype of rs2298850 who had lived in Kuwait for < 51 years had a mean 25(OH)D level of 10 ng/ml, whereas those who were homozygous for the ancestral allele had a mean 25(OH)D level of 17 ng/ml. Furthermore, subjects who had lived in Kuwait for > 51 years had higher vitamin D levels (mean 28 ng/ml) regardless of the genotype of their GC gene. CONCLUSIONS: The GC gene may play a major role in determining vitamin D levels in Arab populations.

Recursive partitioning analysis is predictive of overall survival for patients undergoing spine stereotactic radiosurgery.

Spine stereotactic radiosurgery (SRS) offers excellent radiographic and pain control for patients with spine metastases. We created a prognostic index using recursive partitioning analysis (RPA) to allow better patient selection for spine SRS. Patients who underwent single-fraction spine SRS for spine metastases were included. Primary histologies were divided into favorable (breast/prostate), radioresistant (renal cell/sarcoma/melanoma) and other. Cox proportional hazards regression was done to identify factors associated with overall survival (OS). RPA was performed to identify factors to classify patients into distinct risk groups with respect to OS. A total of 444 patients were eligible. Median dose was 16 Gy (range 8-18) in 1 fraction and median follow-up was 11.7 months. At time of analysis, 103 (23.1%) patients were alive. Median OS was 12.9 months. RPA identified three distinct classes. Class 1 was defined as KPS > 70 with controlled systemic disease (n = 142); class 3 was defined as KPS ≤ 70 and age < 54 years or KPS ≤ 70 age ≥ 54 years and presence of visceral metastases (n = 95); all remaining patients comprise class 2 (n = 207). Median overall survival was 26.7 months for class 1, 13.4 months for class 2, and 4.5 months for class 3 (p < 0.01). Our analysis demonstrates that there is considerably variability in survival among patients undergoing spine SRS. We created an objective risk stratification via RPA for spine SRS. Given the safety and efficacy of spine SRS and good survival in class 1 and 2 patients, this RPA can help clinicians identify patients who may benefit from upfront spine SRS.

Prognostic factors for survival in adult patients with recurrent glioblastoma: a decision-tree-based model.

We assessed prognostic factors in relation to OS from progression in recurrent glioblastomas. Retrospective multicentric study enrolling 407 (training set) and 370 (external validation set) adult patients with a recurrent supratentorial glioblastoma treated by surgical resection and standard combined chemoradiotherapy as first-line treatment. Four complementary multivariate prognostic models were evaluated: Cox proportional hazards regression modeling, single-tree recursive partitioning, random survival forest, conditional random forest. Median overall survival from progression was 7.6 months (mean, 10.1; range, 0-86) and 8.0 months (mean, 8.5; range, 0-56) in the training and validation sets, respectively (p = 0.900). Using the Cox model in the training set, independent predictors of poorer overall survival from progression included increasing age at histopathological diagnosis (aHR, 1.47; 95% CI [1.03-2.08]; p = 0.032), RTOG-RPA V-VI classes (aHR, 1.38; 95% CI [1.11-1.73]; p = 0.004), decreasing KPS at progression (aHR, 3.46; 95% CI [2.10-5.72]; p < 0.001), while independent predictors of longer overall survival from progression included surgical resection (aHR, 0.57; 95% CI [0.44-0.73]; p < 0.001) and chemotherapy (aHR, 0.41; 95% CI [0.31-0.55]; p < 0.001). Single-tree recursive partitioning identified KPS at progression, surgical resection at progression, chemotherapy at progression, and RTOG-RPA class at histopathological diagnosis, as main survival predictors in the training set, yielding four risk categories highly predictive of overall survival from progression both in training (p < 0.0001) and validation (p < 0.0001) sets. Both random forest approaches identified KPS at progression as the most important survival predictor. Age, KPS at progression, RTOG-RPA classes, surgical resection at progression and chemotherapy at progression are prognostic for survival in recurrent glioblastomas and should inform the treatment decisions.

Relationship between tumor location, size, and WHO grade in meningioma.

OBJECTIVE Prior studies have investigated preoperative risk factors for meningioma; however, no association has been shown between meningioma tumor size and tumor grade. The objective of this study was to investigate the relationship between tumor size and grade in a large single-center study of patients undergoing meningioma resection. METHODS A retrospective chart review of patients undergoing meningioma resection at the University of California, San Francisco, between 1985 and 2015 was performed. Patients with incomplete information, spinal meningiomas, multiple meningiomas, or WHO grade III meningiomas were excluded. The largest tumor dimension was used as a surrogate for tumor size. Univariate and multivariate logistic regression models were used to investigate the relationship between tumor grade and tumor size. A recursive partitioning analysis was performed to identify groups at higher risk for atypical (WHO grade II) meningioma. RESULTS Of the 1113 patients identified, 905 (81%) had a WHO grade I tumor and in 208 (19%) the tumors were WHO grade II. The median largest tumor dimension was 3.6 cm (range 0.2-13 cm). Tumors were distributed as follows: skull base (n = 573, 51%), convexity/falx/parasagittal (n = 431, 39%), and other (n = 109, 10%). On univariate regression, larger tumor size (p < 0.001), convexity/falx/parasagittal location (p < 0.001), and male sex (p < 0.001) were significant predictors of WHO grade II pathology. After controlling for interactions, multivariate regression found male sex (OR 1.74, 95% CI 1.25-2.43), size 3-6 cm (OR 1.69, 95% CI 1.08-2.66), size > 6 cm (OR 3.01, 95% CI 1.53-5.94), and convexity/falx/parasagittal location (OR 1.83, 95% CI 1.19-2.82) to be significantly associated with WHO grade II. Recursive partitioning analysis identified male patients with tumors > 3 cm as a high-risk group (32%) for WHO grade II meningioma. CONCLUSIONS Larger tumor size is associated with a greater likelihood of a meningioma being WHO grade II, independent of tumor location and male sex, which are known risk factors.

Tuberculum sellae meningiomas: grading scale to assess surgical outcomes using the transcranial versus transsphenoidal approach.

OBJECTIVE Tuberculum sellae meningiomas (TSMs) are surgically challenging tumors that can severely impair vision. Debate exists regarding whether the transcranial (TC) or endoscopic transsphenoidal (TS) approach is best for resecting these tumors, and there are few large series comparing these approaches. METHODS A retrospective chart review was performed at 2 academic centers comparing TC and TS approaches with respect to vision, extent of resection, recurrence, and complications. The authors report surgical outcomes and propose a simple preoperative tumor grading scale that scores tumor size (1-2), optic canal invasion (0-2), and arterial encasement (0-2). The authors performed univariate, multivariate, and recursive partitioning analysis (RPA) to evaluate outcomes. RESULTS The TSMs were resected in 139 patients. The median follow-up was 29 months. Ninety-five (68%) cases were resected via a TC and 44 (32%) via a TS approach. Tumors treated via a TC approach had a higher tumor (p = 0.0007), artery (p < 0.0001), and total score (p = 0.0012) on the grading scale. Preoperative visual deficits were present in 87% of patients. Vision improved in 47%, stayed the same in 35%, declined in 10%, and was not recorded in 8%. The extent of resection was 65% gross-total resection, 23% near-total resection (95%-99% resection), and 12% subtotal resection (< 95%). A lower tumor score was significantly associated with better or stable vision postoperatively (p = 0.0052). The RPA confirmed low tumor score as the key predictor of postoperative visual improvement or stability. Multivariate analysis and RPA demonstrate that lower canal score (p < 0.0001) and TC approach (p = 0.0019) are associated with gross-total resection. Complications occurred in 20 (14%) patients, including CSF leak (5%) and infection (4%). There was no difference in overall complication rates between TC and TS approaches; however, the TS approach had more CSF leaks (OR 5.96, 95% CI 1.10-32.04). The observed recurrence rate was 10%, and there was no difference between the TC and TS approaches. CONCLUSIONS Tuberculum sellae meningiomas can be resected using either a TC or TS approach, with low morbidity and good visual outcomes in appropriately selected patients. The simple proposed grading scale provides a standard preoperative method to evaluate TSMs and can serve as a starting point for selection of the surgical approach. Higher scores were associated with worsened visual outcomes and subtotal resection, regardless of approach. The authors plan a multicenter review of this grading scale to further evaluate its utility.

Two-year experience of multi-disciplinary team (MDT) outcomes for brain metastases in a tertiary neuro-oncology centre.

Brain metastases (BMs) are the most common intracranial tumour in adults and form a significant proportion of the neuro-oncology workload. Their management has progressed significantly in the last few decades but a gold-standard evidence-based management strategy has not been defined to date and several guidelines based on available evidence exist to support clinical decision-making. This paper evaluates the decision-making process of the neuro-oncology multi-disciplinary team (MDT) in a tertiary neuro-oncology centre over a two-year period. A retrospective review of all patients with BM discussed in the MDT was conducted. Data on patient demographics, tumour characteristics and MDT decision were collected from the MDT database, clinical notes and imaging studies. Patients were stratified into the three recursive partitioning analysis (RPA) classes and according to the graded prognostic assessment (GPA) score. MDT decisions were analysed by RPA class and for GPA score as well as single versus multiple BM. There were 362 patients with BM, representing 22% of the total cases discussed at the MDT. Decision-making was largely consistent with available guidelines. A concrete treatment decision was reached in 77.5% of patients and 32.2% of these received neurosurgical input. More patients with solitary BM underwent surgery compared to multiple BM (p = 0.001), and more patients in RPA classes I and II had surgical resection compared to class III (p = 0.005 and 0.001, respectively). Surgical patients also had higher GPA scores compared to palliative patients (p = 0.005). A greater absolute number and proportion of patients in RPA class II vs. class I underwent neurosurgical intervention. These patients were stratified into class II because of their age but would otherwise have been placed into class I. Survival data were available for 195 patients (53.8%) at 1 year post MDT discussion. A pattern of declining survival was observed along RPA classes which was statistically significant (p = 0.0025). Median survival was 4.7 (0-41), 3.7 (0-23), and 2.5 (range 0-24) months for RPA class I, II and III respectively. A similar pattern that did not reach statistical significance was found between GPA scores (p = 0.101). Median survival was 3 (0-13), 4.6 (range 0-41), and 4.6 (0-35) months for GPA scores 0-1.0, 1.5-2.5 and 3-4.0, respectively. Patient selection was generally in accordance to RPA class and GPA scoring, with the exception of surgery offered to elderly patients: this can be explained by the increasing number of otherwise fit patients as population ages.

Recursive partitioning analysis of prognostic factors in post-transplant lymphoproliferative disorders (PTLD): a 120 case single institution series.

The post-transplant lymphoproliferative disorders (PTLD) comprise a heterogeneous group of lymphocytic and plasma cell proliferations occurring in recipients of tissue allografts in the setting of immunosuppression. We describe our experience of 120 patients with PTLD seen between 1990 and 2009, one of the largest series reported by a single institution. Post-transplant lymphoproliferative disorders characteristics were analysed with regard to paediatric and adult patients, and with regard to the decade of diagnosis, 1990-1999 (pre-rituximab era) versus 2000-2009 (the rituximab era). We present a new prognostic score using the recursive partitioning model, consisting of the Eastern Cooperative Oncology Group (ECOG) score (0-1 vs. 2-4), age [paediatrics (<16 years old), adults (16-60 years old) and elderly (>60 years old)] and CD20 status (positive vs negative); separating patients into 4 risk categories based on overall survival. Low-risk included paediatric patients with ECOG score of 0-1; intermediate-low-risk included adults aged 16-60 years with an ECOG score of 0-1; intermediate-high-risk included elderly patients with an ECOG score 0-1 or paediatric patients and adults aged 16-60 years with an ECOG score of 2-4 and CD20 positive; high-risk group included patients of any age with an ECOG score of 2-4 and CD20 negative, and elderly patients with an ECOG score of 2-4 with CD20-positive PTLD.

Survival of patients treated with radiation therapy for anaplastic astrocytoma.

BACKGROUND: Anaplastic astrocytoma (AA) represents 7% of primary brain tumors in adults. Patient-, tumor-, and treatment-related factors are thought to be predictive of survival. We retrospectively assessed the association of patient-, tumor-, and treatment-related factors with survival in AA treated with radiotherapy (RT) at our institution. PATIENTS AND METHODS: Medical records of patients with AA treated with RT between 1987 and 2007 were reviewed. Patient-, tumor-, and treatment-related variables were recorded and used to assign patients to a Radiation Therapy Oncology Group recursive partitioning analysis (RTOG RPA) classification. First use of chemotherapy was recorded. Log-rank tests and Cox regression models were used to assess for an association of patient-, tumor- and treatment-related factors with survival. RESULTS: One-hundred twenty-six patients were eligible for study. Median age, Karnofsky performance status, and duration of symptoms were 43 years, 90, and 8 weeks. Median radiation dose was 59.4 Gy; 61% of patients underwent tumor resection, and 17% and 41% of patients received temozolomide during and after RT. Median survival was 31 months, and 2-year survival was 58%. RTOG RPA class was associated with survival (p < 0.001), but use of temozolomide during or after RT was not (p > 0.05). CONCLUSIONS: In this retrospective study with inherent limitations, RTOG RPA classification was associated with survival. Further studies are necessary to confirm or refute this finding.