Volumetric measurements in low-grade glioma: Are we there yet?

Along with clinical status, radiographic response assessment is one of the key determinants of a novel drug’s efficacy. However, there continues to be a significant discussion on the best way to measure radiographic response assessment. The current Response Assessment in Neuro-Oncology (RANO) criteria for both high-grade gliomas and low-grade gliomas recommend bidirectional (2D) measurements from a single slice that maximizes the area measured. However, this approach is plagued with several challenges. For example, gliomas can be quite variable in size and shape and a single 2D measurement fails to capture the complexity of these tumors. Growth is not necessarily evenly distributed over the tumor and one focal area may start to grow but measuring the slice demonstrating growth could fail to reach the threshold for progression despite clearly showing focal progression. In addition, there is significant inter- and intra-observer variability in how tumors are measured, particularly in low-grade gliomas which are measured on T2/FLAIR sequences and often have vague or imprecise boundaries.¹

Using MRIs from patients enrolled on the phase 1 trial of ivosidenib in IDH-mutant gliomas, Ellingson et al evaluated the benefit of using volumetric measurements vs standard 2D measurements on FLAIR images (all patients had nonenhancing disease) to assess tumor response.² In line with prior reports, they found higher reader discordant rates with 2D measurements vs volumetric measurements and the radiographic response assessment (CR, PR, SD, or PD) was significantly different at each time point between 2D and volumetric measurements.³ Volumetric measurements were also more stable over time with less variability, resulting in longer progression-free survival based on volumetrics—the variability in 2D measurements was hypothesized to result in progression being called earlier.

While this study confirms what many people intuitively believe—that volumetric measurements more reliably captured low-grade glioma complexity, there remain some challenges to implementation of volumetric measurements. Volumetric measurements are more time-consuming and typically require additional software for tumor segmentation. In the context of a clinical trial, a central lab would ideally provide tumor volumes in real time. As mentioned by the authors, some academic centers can provide volumetric measurements but this tends to come at additional cost and may or may not be provided in real time. With the increasing number of deep learning-based glioma segmentation algorithms, the ability to generate accurate volumetric segmentations in real time will hopefully increase, thus increasing the likelihood that this information can be provided to the physician at the time of the patient visit.⁴ Even outside the context of a clinical trial, having tumor volumetric measurements longitudinally over the patient’s entire history as part of the radiology report would provide a more accurate assessment of growth in these slow-growing tumors.

As these technical issues are resolved, using volumetrics will become a real possibility, hopefully leading to improved patient care and improved drug development although this still needs to be determined. Certainly decreasing the variability in radiographic response assessment will improve our confidence in tumor measurements given their importance in the care of patients with brain tumor.