Objective: Identification of acute ischemic stroke with large vessel occlusion (AIS-LVO) etiology is crucial for effective revascularization therapy. As collaterals are pivotal in maintaining cerebral perfusion in intracranial atherosclerotic stenosis (ICAS), we investigated whether multiphase CT angiography (mCTA) collateral score can be a diagnosis marker of ICAS-related LVO. Methods: We reviewed clinical and imaging data from 92 patients who presented with AIS-LVO and underwent mCTA (57 ICAS-related LVO and 35 embolic LVO). Logistic regression was used to identify ICAS-related LVO. The diagnostic accuracy of the mCTA collateral score for identifying ICAS-related LVO was determined using receiver operating characteristic (ROC) analysis. Results: Compared with patients with embolic LVO, those with ICAS-related LVO had a high median mCTA collateral score (4 vs. 3; P<0.0001). The multinomial logistic regression analysis revealed a significant increase in the mCTA collateral score (OR: 3.717, 95% CI: 2.009-6.876, P<0.0001) in patients with ICAS-related LVO. ROC analysis revealed that the optimal cutoff point of the mCTA collateral score to diagnosis the ICAS-related LVO was 3.5, the area under the curve (AUC) was 0.817 (95% CI: 0.736-0.899; P<0.0001), sensitivity was 80.7%, and specificity was 74.3%. Further analysis revealed that patients with a 4 to 5 mCTA collateral score exhibited a significantly higher median modified Rankin Scale (mRS) at discharge compared with those with a 0 to 3 score (P=0.0464). Conclusions: The mCTA collateral score may be associated with ICAS-related LVO and could be beneficial in identifying the etiology of AIS-LVO.
To address the lack of dynamic prognostic tools for the rare Giant Cell Glioblastoma (gcGBM), this study utilized conditional survival (CS) analysis and a population-based nomogram to provide evolving, personalized survival estimates beyond baseline diagnosis. We identified 490 patients with gcGBM from the Surveillance, Epidemiology, and End Results (SEER) database (2000-2022). The cohort was split into training (n = 343) and validation (n = 147) sets. Conditional survival probabilities-CS(t|s) = S(t + s)/S(s)-were calculated to assess the likelihood of surviving an additional t years given s years already survived. A Random Survival Forest with Recursive Feature Elimination (RSF-RFE) algorithm was employed to select the most parsimonious predictors. A CS-nomogram was constructed and evaluated using calibration curves, time-dependent AUC, and Decision Curve Analysis (DCA). An interactive web-based calculator was also developed. The baseline 5-year overall survival (OS) was 14%. However, CS analysis revealed that the prognosis improved significantly over time: for patients who survived 3 years, the probability of surviving to year 5 rose to 69%, and reached 83% for those surviving to year 4. The RSF-RFE identified six key predictors: age, sex, tumor site, surgery, radiotherapy, and chemotherapy. The resulting CS-nomogram demonstrated high discriminatory power and clinical net benefit in both cohorts. Risk stratification effectively distinguished high-risk and low-risk groups. Prognosis in gcGBM is a dynamic process. The risk of mortality is highest in the early years post-diagnosis but decreases substantially for long-term survivors. Our validated CS-nomogram and web-based tool provide clinicians with a practical means to offer personalized, time-adjusted survival estimates, facilitating better patient counseling and refined follow-up strategies.
Anaplastic ganglioglioma (AGG) is an extremely rare glioneuronal tumor with low incidence and heterogeneous survival patterns driven by a limited set of prognostic factors, and conditional survival has not been systematically evaluated in this entity. Patients histologically diagnosed with AGG between 2000 and 2022 were extracted from the SEER 17 registries. Age-adjusted incidence and annual percentage change (APC) were estimated by Joinpoint regression. Overall survival (OS) was analyzed with Kaplan–Meier curves and log-rank tests; 5-year conditional survival (CS) was computed at successive landmark times. Independent prognostic factors were identified by univariate and multivariate Cox regression. A CS-nomogram was constructed and evaluated by calibration plots, time-dependent ROC and C-index. Nomogram-derived risk scores were dichotomized using a cut-point selected on the basis of restricted cubic splines (RCS), and survival between strata was compared. A total of 143 AGG patients were identified, with 62 deaths during follow-up. Patients included pediatric (29.4%), AYA (38.5%), and adults ≥ 40 years (32.2%), and 64.3% were male; most tumors arose in cerebral lobes (73.4%). Age-adjusted incidence of AGG remained low throughout the study window, with crude annual rates of approximately 0.082 per 1,000,000 person-years. Five-year OS was 64.7%, while CS5 increased from 64.7% at diagnosis to 73.7% at 1 year, 82.5% at 2 years, and > 90% after 3 years. Multivariate analysis identified age ≥ 40, regional/distant stage, and multiple primary tumors as independent adverse factors. The CS nomogram showed good performance, with time-dependent AUCs > 0.75 and well-calibrated predictions, and risk stratification yielded significantly separated survival outcomes (P < 0.001). Long-term prognosis in AGG is highly dynamic: patients who survive the first three years acquire a substantially improved residual life expectancy. A CS-based nomogram, combined with RCS-guided risk stratification, translates this dynamic information into an individualized tool for counselling and risk-adapted surveillance.