Tumor volume features predict survival outcomes for patients diagnosed with diffuse intrinsic pontine glioma

Abstract

Background: Diffuse intrinsic pontine glioma (DIPG) is a fatal childhood central nervous system tumor. Diagnosis and monitoring of tumor response to therapy is based on magnetic resonance imaging (MRI). MRI-based analyses of tumor volume and appearance may aid in the prediction of patient overall survival (OS).

Methods: Contrast-enhanced T1- and FLAIR/T2-weighted MR images were retrospectively collected from children with classical DIPG diagnosed by imaging (n = 43 patients). MRI features were evaluated at diagnosis (n = 43 patients) and post-radiation (n = 40 patients) to determine OS outcome predictors. Features included 3D tumor volume (T_wv), contrast-enhancing tumor core volume (T_c), T_c relative to T_wv (T_C/T_wv), and T_wv relative to whole brain volume. Support vector machine (SVM) learning was used to identify feature combinations that predicted OS outcome (defined as OS shorter or longer than 12 months from diagnosis).

Results: Features associated with poor OS outcome included the presence of contrast-enhancing tumor at diagnosis, >15% T_c/T_wv post-radiation therapy (RT), and >20% ∆Tc/T_wv post-RT. Consistently, SVM learning identified T_c/T_wv at diagnosis (prediction accuracy of 74%) and ∆T_c/T_wv at <2 months post-RT (accuracy = 75%) as primary features of poor survival.

Conclusions: This study demonstrates that tumor imaging features at diagnosis and within 4 months of RT can predict differential OS outcomes in DIPG. These findings provide a framework for incorporating tumor volume-based predictive analyses into the clinical setting, with the potential for treatment customization based on tumor risk characteristics and future applications of machine-learning-based analysis.

Keywords: MRI; SVM learning; diffuse intrinsic pontine glioma; survival outcomes; tumor volume.

Figure 1.

Diagnostic tumor MRI volume is predictive of survival outcomes (A) Representative axial MRI from one patient showing overlaying T_wv and T_c tumor volume segmentations. (B) T_wv relative to whole brain volume (WBV) between shorter (n = 22) versus longer (n = 21 with available WBV data) survivors. Mann–Whitney test, P = .622. (C) Left: Survival comparison of patients with contrast-enhancing tumor on T1 imaging (T_c+) (n = 23) versus those without contrast-enhancing tumor on T1 imaging (T_c−) (n = 20), Log-rank test, P = .010. Right: Survival comparison of patients with T_c/T_wv > 25% (n = 3) vs. those with T_c/T_wv < 25% (n = 40) at diagnosis, Log-rank test, P = .009. (D) Comparison of T_c/T_wv (%) between shorter (OS < 12 months, n = 22) versus longer survivors (OS ≥ 12 months, n = 21), Mann–Whitney test, P = .0255.

Figure 2.

T_c/T_wv following radiation therapy predicts survival outcomes. (A) Comparison of T_c/T_wv at initial diagnosis (‘Pre-RT’; n = 43) and following RT at <2 months (n = 37; Wilcoxon signed rank, corrected P = .0007) and 2–4 months post-RT (n = 37; Wilcoxon signed rank, corrected P = .0001). There was no significant difference in the T_c/T_wv ratios between the <2 months and 2–4 months MRI timepoints (P = .475) (B) Difference in T_c/T_wv at < 2 mo post-RT between shorter (n = 19) and longer (n = 18) survivors, Mann–Whitney test, P = .0242. (C) Survival comparison between patients with ≥15% T_c/T_wv (n = 19) and those with <15% T_c/T_wv (n = 18) at <2 months post-RT, Log-rank test, P = .03. (D) Difference in ∆T_c/T_wv from diagnosis to <2 mo post-RT between shorter (n = 19) and longer (n = 18) survivors, Mann–Whitney test, P = .0132. (E) Survival comparison at the <2 months post-RT timepoint between patients with a large ∆T_c/T_wv increase (n = 9; ∆ T_c/T_wv > 19%, top 25% of all patients) and patients within the bottom 25% (n = 9; ∆ T_c/T_wv between −1.6% and 0.15%), Log-rank test, P = .023.

Figure 3.

Whole tumor volume changes post-radiation provide limited insight into patient survival outcomes. (A) Change in T_wv from baseline to <2 months RT and baseline to 2–4 months post-RT (n = 37 patients for each timepoint). There was no difference in T_wv volume change between the 2 timepoints (Wilcoxon matched-pairs signed rank test, P = .598). (B) T_wv percent change at <2 months post-RT between shorter and longer survivors. Mann–Whitney test, P = .105. (C) Survival comparison between patients who experienced a decrease (n = 21) versus an increase (n = 16) in T_wv volume from diagnosis to <2 months post-RT, Log-rank test, P = .025. (D) OS outcome of patients who experienced a T_wv percent increase greater than 25% of the initial diagnosis volume (“non-responders”) (n = 10), decrease greater than 25% (‘responders’) (n = 13), or a T_wv percent change less than 25% (“stable”) (n = 14). Log-rank test, P = .5. (E) OS outcome of patients who experienced a T_wv percent increase greater than 50% of the initial diagnosis volume (extreme “non-responders”) (n = 9), decrease greater than 50% (extreme ‘responders’) (n = 4), or a T_wv percent change less than 50% (“stable”)(n = 24). Log-rank test, P = .2. (F) Summary of MRI volume features at each timepoint that are independently predictive of OS.