Pre-Radiotherapy Synthetic MRI-Derived Quantitative Heterogeneity and Early Recurrence in Glioblastoma.

Abstract

Background: Timely identification of early recurrence (≤ 6 months) may improve prognosis of glioblastoma (GBM), but conventional MRI has shown limited accuracy in this setting.

Purpose: Risk-assessment models and a nomogram were constructed by integrating SyMRI metrics, clinical-pathological variables, and cMRI features, including contrast-enhanced T1-weighted imaging and fluid-attenuated inversion recovery (FLAIR) findings.

Study type: Retrospective observational study.

Population: Seventy-eight patients with GBM (median age, 59 years; 44 [56.4%] males).

Field strength/sequence: 3 T; pre- and post-contrast three-dimensional T1-weighted imaging, FLAIR, and SyMRI.

Assessment: Histogram-based quantitative metrics were extracted from SyMRI maps using subregions defined on fused FLAIR and contrast-enhanced T1-weighted images. All candidate clinical-pathological variables, cMRI features, and SyMRI-derived metrics were entered directly into least absolute shrinkage and selection operator (LASSO) regression for variable selection. Risk-assessment models and a nomogram were constructed.

Statistical tests: Multivariable logistic regression, multicollinearity assessment using variance inflation factors, receiver operating characteristic curve analysis with DeLong test, stratified 10-fold cross-validation, leave-one-out cross-validation, nested 5-fold cross-validation, calibration curves, and decision curve analysis. A two-sided p < 0.05 was considered statistically significant.

Results: Multivariate analysis identified reduced T2 entropy (< 2.113) in enhancement-corresponding regions (odds ratio [OR] = 0.08), thick linear or nodular residual cavity wall enhancement (OR = 5.28), corpus callosum involvement (OR = 5.08), and O6-methylguanine-DNA methyltransferase (MGMT) promoter methylation (OR = 0.19) as variables associated with early recurrence. The integrated model achieved the highest performance (AUC = 0.864). Nested cross-validation showed moderate internal validation performance, with an AUC of 0.724 (0.722-0.726).

Data conclusion: Histogram-based pre-radiotherapy SyMRI metrics, particularly T2 entropy, were associated with early GBM recurrence. The integrated model achieved the highest apparent performance; nested internal validation showed moderate performance. External validation in larger multicenter cohorts is required before clinical implementation.

Evidence level: 3.

Technical efficacy: Stage 2.