Date of Award

January 2024

Document Type

Open Access Thesis

Degree Name

Medical Doctor (MD)

Department

Medicine

First Advisor

Mariam Aboian

Abstract

Response assessment in neuro-oncology relies on radiographic assessment of tumor burden on magnetic resonance (MR) imaging. The most widely used criteria were developed by the Response Assessment in Neuro-Oncology (RANO) group. The RANO criteria rely on bidimensional (2D) measurements of tumor on MR images. The RANO criteria were originally developed to assess response in adult high-grade glioma. However, the heterogeneous appearance of pediatric low-grade gliomas make application of RANO criteria challenging. Volumetric assessment of pediatric gliomas may offer a more comprehensive method for characterizing response.

The goal of this thesis was to compare 2D and volumetric assessment methods in two pediatric glioma clinical trials from the Pacific Pediatric Neuro-Oncology Consortium (PNOC). The primary purpose of the thesis was to compare 2D and volumetric response to a clinical reference standard – neuroradiologist visual response assessment via the Brain Tumor Reporting and Data System (BT-RADS). A secondary aim was to determine optimal thresholds for categorizing volumetric response using BT-RADS as a reference standard. A third aim was to compare 2D and volumetric posttreatment trajectories in trial participants. Retrospective analyses of two pediatric glioma clinical trials (PNOC-001 and PNOC-002) were conducted. Changes in tumor 2D area, whole tumor volume, and solid tumor volume were compared to assess response. Follow-up images were assigned a response score on BT-RADS by two neuroradiologists. Empirical receiver operating characteristic (ROC) curves of changes in 2D area, whole, and solid tumor volume were constructed to classify partial response (PR) and progressive disease (PD) based on BT-RADS. In the PNOC-002 trial, a mathematical model was used to construct posttreatment trajectories of changes in 2D area and whole tumor volume in a subset of participants.

Empirical ROC curves to classify BT-RADS PD among the 65 follow-up images assessed in the PNOC-001 trial yielded an AUC of 0.78 (95% CI: 0.66-0.90) for 2D area percent change, 0.84 (95% CI: 0.74-0.94) for whole volume percent change, and 0.96 (95% CI: 0.92-1.00) for solid volume percent change. DeLong tests revealed that there was a significant increase in AUC of the solid volume ROC curve compared to both 2D area (p = 0.005) and whole volume (p = 0.006). The empirical ROC curves to classify BT-RADS PR yielded an AUC of 0.87 (95% CI: 0.77-0.96) for 2D area percent change, 0.84 (95% CI: 0.70-0.99) for whole volume percent change, and 0.97 (95% CI: 0.94-1.00) for solid volume percent change. DeLong tests revealed that there was a significant increase in AUC of the solid volume ROC curve compared to 2D area (p = 0.02) but not whole volume (p = 0.08). The thresholds for solid volume percent change that included an 80% sensitivity in their 95% confidence intervals for classifying BT-RADS PD ranged from 15-25% and 15-20% for classifying BT-RADS PR.

The empirical ROC curves for classification of BT-RADS PR in the 31 participants at the end of treatment or last available follow-up produced the following AUC values: 0.92 (95% CI: 0.80-1.00) for 2D area percent change, 0.99 (95% CI: 0.97-1.00) for whole volume percent change, and 0.99 (95% CI: 0.97-1.00) for solid volume percent change. DeLong test revealed no statistically significant difference in AUC between 2D area and either solid (p = 0.17) or whole volume (p = 0.17) ROC curves. The empirical ROC curves for classification of BT-RADS PR at the first time of BT-RADS PR detection produced the following AUC values: 0.84 (95% CI: 0.69-0.99) for 2D area percent change, 0.91 (95% CI: 0.80-1.00) for whole volume percent change, and 0.92 (95% CI: 0.82-1.00) for solid volume percent change. There was no statistically significant difference in AUC between the 2D area ROC curve and either solid (p = .34) or whole volume (p = .39) ROC curves based on DeLong tests. Based on mathematically modeled trajectories, there was no significant correlation in time to best response obtained from 2D area vs. whole volume posttreatment changes (? = 0.39, p = 0.054). Eight out of 25 participants (32%) had a difference of 90 days or more in transition time from partial response to stable disease between 2D area and whole volume trajectories. Moreover, of the 16 participants with tumor regrowth following stable disease, 50% had a difference of 90 days or more in transition time from stable disease to progressive disease between 2D area and whole volume trajectories.

Solid tumor volume better predicted neuroradiologist assessment of partial response and progressive disease according to BT-RADS criteria in the PNOC-001 trial but performed as well as 2D measurements in classifying partial response in the PNOC-002 trial. Although volumetrics was not consistently superior to 2D measurements in detecting response in our study, there were differences in individual participant 2D and volumetric posttreatment trajectories. Future research comparing volumetric to 2D assessment in prospective trials is required to understand the significance of these differences to clinical management.

Comments

This is an Open Access Thesis.

Download
Open Access

This Article is Open Access

Share

COinS