European Radiology: imaging-based risk stratification enables prognostic assessment at the molecular level of diffuse glioma tissue

Studies have shown that the molecular status of isocitrate dehydrogenase (IDH) mutation and 1p/19q-codeletion of chromosome 1p and 19q chromosome arm is important for
the frontal prognostic assessment of gliomas.
NOS (unspecified) diagnosis includes gliomas
that have not been genetically tested or are not sufficient to place them in a specific molecular diagnostic category.
First mentioned in the 2016 revision of the World Health Organization (WHO) classification, which integrates histopathology and molecular diagnosis
of gliomas.
Because NOS refers to tumors that cannot be classified into a specific molecular subtype, prognosis is difficult to predict and clinical management is challenging
for clinicians.

At present, a correlation between specific imaging phenotypes and molecular subtypes of gliomas has been clinically found.

Previous studies have primarily evaluated the molecular subtypes of gliomas based on imaging phenotypes, including enhancement patterns, the presence of necrosis, and T2/fluid attenuation inversion recovery (FLAIR)
mismatches.
However, in NOS, the correlation between imaging and genomics is difficult to assess
because there is no reference standard endpoint for molecular analysis.

Recently, a study published in the journal European Radiology explored the value of imaging-based risk stratification in prognosis prediction of diffuse gliomas (NOS), providing imaging support
for the rapid, non-invasive and accurate clinical assessment of risk stratification and prognosis assessment of gliomas.

This review retrospectively included data
from 220 patients classified as diffuse gliomas between January 2011 and December 2020.
Two neuroimaging scientists analyzed preoperative CT and MRI images to assign gliomas to three image-based risk types
, taking into account well-known imaging phenotypes such as T2/FLAIR mismatches 。 According to the 2021 World Health Organization classification, these three risk types include: (1) low risk, expected oligodendroglioma, isocitrate dehydrogenase (IDH) mutation, 1p/19q code; (2) Medium risk, expected astrocytoma, IDH mutation; and (3) high risk, expected glioblastoma, IDH wild type
.
Progression-free survival (PFS) and overall survival (OS) were estimated
for each risk type.
Time-dependent ROC analysis of 10-fold cross-validation and 100-fold bootstrapping was used to compare the performance
of imaging-based survival models with molecular-based historical survival models published in 2015.

According to the three imaging-based risk types, PFS and OS were predicted (logarithmic test, P<0.
001).
<b20> Image-based survival models showed high prognostic value, with areas under the curve (AUC) of 1-year PFS and OS of 0.
772 and 0.
650, respectively, similar to the molecular-based survival models reported in previous studies (AUC=0.
74 for PFS and 0.
87 for OS).

Image-based survival models achieved high long-term performance
in both 3-year PFS (AUC = 0.
806) and 5-year OS (AUC = 0.
812).

The graph is based on the time-dependent ROC curve
of the survival model (PFS) of imaging.
Shown is the time-dependent ROC curve
of progression-free survival based on an imaging survival model in this study.
ROC curves of 1, 2, 3 and 5 years after surgery are provided

In this study, imaging-based risk stratification achieves a prognostic prediction
of diffuse gliomas (NOS) at the tissue molecular level.
Image-based survival models have shown good predictive performance in both progression-free survival and overall survival, comparable to molecular-based models in previous studies
.
The results of this study suggest that imaging-based risk stratification can assist in the clinical prediction of the clinical prognosis of diffuse glioma (NOS), which will help clinically guide patient referral and treatment
when molecular diagnosis is not available.

Original source:

Eun Bee Jang,Ho Sung Kim,Ji Eun Park,et al.
Diffuse glioma, not otherwise specified: imaging-based risk stratification achieves histomolecular-level prognostication.
DOI:10.
1007/s00330-022-08850-z