Skip to main content
SNO 28th Annual Meeting Main banner
Final Program


Program
Icon Legend
Presentation Icons
CME
CME
Non-CME
Non-CME
ISS CME
ISS CME
Named Award Winner
Named Award Winner
Award Winner
Award Winner
Poster Award Winner
Poster Award Winner
Back

Neuro-imaging

Session: Poster Session

NIMG-47 - Longitudinal trajectories from serial surveillance MRI imaging of progressive IDH mutant low-grade glioma patients

Friday, November 17, 2023
7:30 PM – 9:30 PM PT
Location: Exhibit Hall A/B
Non-CME

Background: Most patients with isocitrate dehydrogenase mutant (IDHm) low-grade glioma (LGG) undergo active MRI surveillance after initial surgery. Timely biopsy referral during surveillance is a complex trade-off between avoiding the harms of unnecessary surgery and early detection of transformation. Serial surveillance MRIs could be leveraged to develop dynamic imaging markers, yet prior research has primarily focused on the prognostication of pre-treatment images. We piloted the feasibility of an automated pipeline using patient-specific imaging trajectories.

Methods: We retrospectively identified 12 progressive IDHm LGG patients and collected serial surveillance MRIs. A previously developed segmentation model was fine-tuned on a subset of T2/FLAIR images (n=49) before automated tumor segmentation was performed on all MRIs. We extracted geometric and topological tumor features, including volume, cross product, and Euler Characteristic Transform, from each MRI. We used piecewise linear regression to quantify feature trajectories and identify change points in feature growth rates.

Results: Twelve patients received 14 biopsies/repeat resections (10 with transformation, 4 without) and 282 MRIs (median per patient: 17, range: 10-40). All biopsies/repeat resections were preceded by a change point in tumor volume trajectory, indicating an increase in tumor growth rate preceding clinician-determined MRI progression and biopsy referral. The mean increase in volume growth rate, pre to post change point, was 2.3cm3/month (SE: 0.9) in biopsies/repeat resections with transformation, compared to 1.1cm3/month (SE: 0.6) in biopsies/repeat resections without transformation. Among the ten patients with transformation, the median time from IDHm LGG diagnosis to transformation was 64.8 months (range: 38.6-141.1), and the median time between volume change point and transformation was 15.6 months (range: 4.6-28.2).

Discussion: This pilot study suggests that feature trajectories from serial MRIs of IDHm LGG patients may advance the detection and treatment of transformed lesions by 12-18 months. Larger studies can corroborate these findings and inform clinical decision-making.