Perfusion, diffusion, and anatomical MRI characteristics of pathologically confirmed malignant transformation in IDH-mutant gliomas

Nicholas S Cho; Viên Lam Le; Ashley Teraishi; Vicki Liu; Francesco Sanvito; Donatello Telesca; Masanori Nakajo; Chencai Wang; Sonoko Oshima; Blaine S C Eldred; Jingwen Yao; Phioanh L Nghiemphu; Noriko Salamon; Timothy F Cloughesy; Albert Lai; Benjamin M Ellingson

doi:10.1093/neuonc/noaf171

. 2025 Aug 28;27(12):3306–3317. doi: 10.1093/neuonc/noaf171

Perfusion, diffusion, and anatomical MRI characteristics of pathologically confirmed malignant transformation in IDH-mutant gliomas

Nicholas S Cho ^1,^2,^3,^4,^#, Viên Lam Le ^5,^6,^7,^#, Ashley Teraishi ^8,^9,^#, Vicki Liu ^10,¹¹, Francesco Sanvito ^12,¹³, Donatello Telesca ¹⁴, Masanori Nakajo ^15,¹⁶, Chencai Wang ^17,¹⁸, Sonoko Oshima ^19,²⁰, Blaine S C Eldred ^21,²², Jingwen Yao ^23,²⁴, Phioanh L Nghiemphu ^25,²⁶, Noriko Salamon ²⁷, Timothy F Cloughesy ^28,²⁹, Albert Lai ^30,³¹, Benjamin M Ellingson ^32,^33,^34,^35,^36,^✉

¹ Medical Scientist Training Program, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, CA, USA

² Department of Bioengineering, Henry Samueli School of Engineering and Applied Science, University of California, Los Angeles, Los Angeles, CA, USA

³ Department of Radiological Sciences, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, CA, USA

⁴ UCLA Brain Tumor Imaging Laboratory (BTIL), Center for Computer Vision and Imaging Biomarkers, University of California, Los Angeles, Los Angeles, CA, USA

⁵ Department of Bioengineering, Henry Samueli School of Engineering and Applied Science, University of California, Los Angeles, Los Angeles, CA, USA

⁶ Department of Radiological Sciences, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, CA, USA

⁷ UCLA Brain Tumor Imaging Laboratory (BTIL), Center for Computer Vision and Imaging Biomarkers, University of California, Los Angeles, Los Angeles, CA, USA

⁸ Department of Radiological Sciences, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, CA, USA

⁹ UCLA Brain Tumor Imaging Laboratory (BTIL), Center for Computer Vision and Imaging Biomarkers, University of California, Los Angeles, Los Angeles, CA, USA

¹⁰ Department of Neurology, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, CA, USA

¹¹ UCLA Neuro-Oncology Program, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, CA, USA

¹² Department of Radiological Sciences, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, CA, USA

¹³ UCLA Brain Tumor Imaging Laboratory (BTIL), Center for Computer Vision and Imaging Biomarkers, University of California, Los Angeles, Los Angeles, CA, USA

¹⁴ Department of Biostatistics, Fielding School of Public Health, University of California Los Angeles, Los Angeles, CA, USA

¹⁵ Department of Radiological Sciences, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, CA, USA

¹⁶ UCLA Brain Tumor Imaging Laboratory (BTIL), Center for Computer Vision and Imaging Biomarkers, University of California, Los Angeles, Los Angeles, CA, USA

¹⁷ Department of Radiological Sciences, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, CA, USA

¹⁸ UCLA Brain Tumor Imaging Laboratory (BTIL), Center for Computer Vision and Imaging Biomarkers, University of California, Los Angeles, Los Angeles, CA, USA

¹⁹ Department of Radiological Sciences, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, CA, USA

²⁰ UCLA Brain Tumor Imaging Laboratory (BTIL), Center for Computer Vision and Imaging Biomarkers, University of California, Los Angeles, Los Angeles, CA, USA

²¹ Department of Neurology, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, CA, USA

²² UCLA Neuro-Oncology Program, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, CA, USA

²³ Department of Radiological Sciences, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, CA, USA

²⁴ UCLA Brain Tumor Imaging Laboratory (BTIL), Center for Computer Vision and Imaging Biomarkers, University of California, Los Angeles, Los Angeles, CA, USA

²⁵ Department of Neurology, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, CA, USA

²⁶ UCLA Neuro-Oncology Program, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, CA, USA

²⁷ Department of Radiological Sciences, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, CA, USA

²⁸ Department of Neurology, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, CA, USA

²⁹ UCLA Neuro-Oncology Program, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, CA, USA

³⁰ Department of Neurology, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, CA, USA

³¹ UCLA Neuro-Oncology Program, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, CA, USA

³² Department of Psychiatry and Biobehavioral Sciences, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, CA, USA

³³ Department of Neurosurgery, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, CA, USA

³⁴ Department of Bioengineering, Henry Samueli School of Engineering and Applied Science, University of California, Los Angeles, Los Angeles, CA, USA

³⁵ Department of Radiological Sciences, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, CA, USA

³⁶ UCLA Brain Tumor Imaging Laboratory (BTIL), Center for Computer Vision and Imaging Biomarkers, University of California, Los Angeles, Los Angeles, CA, USA

Nicholas S Cho, Viên Lam Le and Ashley Teraishi contributed equally to this work.

^✉

Corresponding Author: Benjamin M. Ellingson, PhD, UCLA Brain Tumor Imaging Laboratory (BTIL), Departments of Radiological Sciences, Psychiatry, and Neurosurgery, David Geffen School of Medicine, University of California, Los Angeles, 924 Westwood Blvd., Suite 615, Los Angeles, CA 90024, USA (bellingson@mednet.ucla.edu)

PMCID: PMC12916721 PMID: 40888039

Abstract

Background

This study explored MRI characteristics at the time of tumor progression to study pathologically confirmed MT in IDHm 1p/19q-intact astrocytomas (IDHm-A) and IDHm 1p/19q-co-deleted oligodendrogliomas (IDHm-O).

Methods

N = 64 patients with initial pathological grade 2 IDH-mutant glioma diagnosis who underwent repeated tissue sampling and were classified as pathologically confirmed MT (n = 35) or non-MT (n = 29) with available presurgical anatomical (n = 64), diffusion-weighted (n = 61), and dynamic susceptibility contrast perfusion MRI (n = 53) were retrospectively studied. Measurable contrast enhancement (> 1000 mm³), tumor volume, tumor growth rate, sphericity, median apparent diffusion coefficient (ADC), and normalized relative cerebral blood volume (nrCBV) were compared between MT vs non-MT IDHm-A and IDHm-O.

Results

81% of contrast-enhancing IDHm-A and 100% of contrast-enhancing IDHm-O demonstrated MT, while 41% of IDHm-A and 62% IDHm-O exhibited both nonenhancing tumor progression and MT. Tumor volumes were significantly larger in patients with MT compared to non-MT groups for IDHm-A (P = .02) and IDHm-O (P = .04). T2/FLAIR tumor volume growth rate was significantly higher (P = .003), nrCBV was significantly higher (P = .002), and ADC trended lower (P = .06) in MT vs non-MT IDHm-A. There were no significant differences in growth rate, ADC, nrCBV, or sphericity when comparing MT vs non-MT IDHm-O (P > .05).

Conclusions

Many MT IDHm gliomas remain nonenhancing. Growth rate, diffusion, and perfusion MRI show differences between MT and non-MT in IDHm-A but not IDHm-O, which may reflect the different tumor biology of these IDHm molecular subtypes and their need for separate imaging biomarkers. Tumor volumes can help determine MT for both IDHm-A and IDHm-O.

Keywords: Glioma, IDH-mutant astrocytoma, IDH-mutant oligodendroglioma, malignant transformation, MRI

Key Points.

Many transformed tumors remain nonenhancing (41% IDHm-A, 62% IDHm-O).
Large tumor volume identifies MT in both IDHm-A and IDHm-O.
Growth rate, shape, diffusion, and perfusion detect MT in IDHm-A only.

Importance of the Study.

Low-grade isocitrate dehydrogenase (IDH)-mutant gliomas often later undergo malignant transformation (MT) to a higher grade, resulting in more aggressive disease. MT is usually defined noninvasively by the emergence of contrast-enhancement in post-contrast T1-weighted MRI, but previous studies have shown that contrast enhancement can be present in gliomas that have not transformed and conversely may not be present in some gliomas that have undergone MT. The purpose of this study was to (1) evaluate contrast-enhancement as an imaging biomarker for pathologically confirmed MT and (2) to characterize additional, quantitative imaging biomarkers for MT in IDH-mutant astrocytomas (IDHm-A) and IDH-mutant oligodendrogliomas (IDHm-O) using pathologically confirmed MT as ground truth. The present results demonstrate that tumor volume can differentiate MT across IDHm gliomas. In addition, shape, growth rate, diffusion, and perfusion can differentiate MT in IDHm-A but not IDHm-O, supporting the need for molecular subtype-specific MT imaging biomarkers for IDHm gliomas.

Isocitrate dehydrogenase (IDH)-mutant gliomas often initially present as low-grade (WHO grade 2) gliomas but are expected to eventually undergo malignant transformation (MT) into higher-grade (WHO grades 3 and 4) gliomas. After MT, these higher-grade gliomas require urgent treatment as they are more aggressive and have a significantly worse prognosis.¹ MT is often noninvasively identified using the emergence of contrast-enhancement in post-contrast T1-weighted MRI, which typically reflects the most active tumor region.² However, multiple studies have shown that contrast-enhancement can be present in low-grade gliomas and, conversely, contrast-enhancement can be absent in high-grade gliomas.^3,4 As a result, there is a need to study other imaging biomarkers to detect MT using pathologically confirmed MT as ground truth instead of the emergence of contrast-enhancement.

Prior imaging studies observed that apparent diffusion coefficient (ADC) values from diffusion MRI and normalized relative cerebral blood volume (nrCBV) from dynamic susceptibility contrast (DSC) perfusion MRI may help identify MT.^5,6 However, these studies on imaging biomarkers of low-grade glioma MT were performed when the field was based on pathologically defined glioma subtypes^2,6 and not current molecularly defined glioma subtypes based on IDH mutation and 1p/19q-codeletion status.⁷ As a result, there is also a need to study MT within IDH-mutant 1p/19q-intact astrocytomas and IDH-mutant 1p/19q-codeleted oligodendrogliomas as distinct entities.

The present study explored volumetric, shape, diffusion, and perfusion MRI characteristics to study pathologically confirmed MT in molecularly defined IDH-mutant gliomas. We hypothesized that contrast-enhancement would be present more often in gliomas with MT than in gliomas with non-MT. We also hypothesized that gliomas with MT would be associated with larger tumor volumes, higher tumor growth rates, lower sphericity, lower ADC, and higher nrCBV than gliomas with non-MT. We also hypothesized that, given the different tumor genetics of IDH-mutant glioma subtypes, different imaging biomarkers may be useful for identifying MT in IDHm-A and IDHm-O. In additional analyses, we also examined group differences in IDHm-A by studying only grade 4 MT IDHm-A compared to grade 2 non-MT IDHm-A and by combining grades 2 and 3 IDHm-A into a single group.

Materials and Methods

Patient Cohort

This retrospective study was performed in compliance with the Health Insurance Portability and Accountability Act and was approved by the UCLA Institutional Review Board. All patients provided written informed consent. All tumors were grouped based on the molecular subtypes of the 2021 World Health Organization Classification of Central Nervous System Tumors,⁷ and tumor grading was based on the 2016 criteria due to lack of key genetic factors in the updated criteria (eg, CDKN2A/B status). A total of n = 64 patients with the following inclusion criteria were studied: (1) received biopsy/surgery to confirm initial presentation as grade 2 IDH-mutant glioma, (2) underwent a repeat biopsy/surgery later in disease course with repeated tumor grading, (3) available 1p/19q-codeletion status, and (4) available T1-weighted pre-contrast, T1-weighted post-contrast, T2-weighted, and T2-weighted FLAIR MRI within 1 month prior to repeat biopsy/surgery (Table 1). Of these n = 64 patients, n = 37 patients were diagnosed with IDHm-A, and n = 27 patients were diagnosed with IDHm-O (Figure 1). Patients were categorized as having “malignant transformation” (“MT”) if the tumor grade increased to grade 3 or 4 at the repeat biopsy/surgery or as “nonmalignant transformation” (“non-MT”) if the tumor grade diagnosis remained at grade 2. More detailed patient information, including treatments before repeat biopsy/surgery, are presented in Supplementary Tables 1 and 2. In brief, n = 31 patients received no additional treatment before repeat biopsy/surgery, and the various treatments in the n = 33 patients that received treatment included radiation therapy (n = 8), radiation therapy/temozolomide (n = 9), temozolomide (n = 19), AG-881 IDH inhibitor therapy (n = 6), AG-120 IDH inhibitor therapy (n = 5), procarbazine/lomustine/vincristine (n = 3), lomustine (n = 2), gamma delta immunotherapy (n = 1), gamma knife (n = 1), procarbazine (n = 1), everolimus (n = 1), imiquimod (n = 1), and unknown chemotherapy (n = 1).

Table 1.

Patient Demographics and Information

	IDHm-A (1p/19q-intact)	IDHm-O (1p/19q-codeleted)
	Total patients (n = 37)	Total patients (n = 27)
Age in years (average/range)	38 (24–63)	46 (25–72)
Sex (male/female)	23/14	16/11
MT/Non-MT	22/15	13/14
Months between surgeries (average/range)	49.3 (3–205)	110.8 (1–331)

Open in a new tab

IDHm-A = isocitrate dehydrogenase mutant astrocytoma, IDHm-O = isocitrate dehydrogenase mutant oligodendroglioma, MT = malignant transformation.

Flowchart showing the breakdown of the patient cohort for cross-sectional and longitudinal analyses, followed by a schematic showing the scan and biopsy/surgery acquisition timeline for longitudinal analyses. — Overview of study design. (A) Flowchart for patient analyses. Out of n = 64 total patients, n = 37 were diagnosed with IDHm-A and n = 27 were diagnosed with IDHm-O. From the n = 37 IDHm-A patients, diffusion MRI was available for n = 34 and perfusion MRI was available for n = 28 patients. From the n = 27 IDHm-O patients, diffusion MRI was available for n = 27 and perfusion MRI was available for n = 25 patients. For longitudinal growth rate analyses, n = 15 IDHm-A and n = 18 IDHm-O were excluded for having fewer than 3 imaging timepoints, multiple recurrences, or multiple therapies. (B) General schematic of image acquisition and biopsy acquisition timeline. All patients received a pathological grade 2 diagnosis at initial surgery and at least one MRI before undergoing a second surgery to determine MT or non-MT status. IDHm-A = isocitrate dehydrogenase mutant astrocytoma, IDHm-O = isocitrate dehydrogenase mutant oligodendroglioma, MRI = magnetic resonance imaging, MT = malignant transformation.

Image Acquisition and Preprocessing

The MRI scans were obtained from 1.5T (Siemens Avanto, Symphony; GE Medical Systems Signa HDxt, Signa Excite) or 3T scanners (Siemens Prisma, Skyra, Trio, Vida; GE Medical Systems Discovery 750, Signa HDxt) using standard-of-care imaging protocols for anatomic scans. Diffusion-weighted imaging (DWI) (n = 61) and DSC perfusion MRI (n = 53) were also available for most patients (Figure 1). ADC maps were generated from the DWI scans. DSC perfusion MRI was motion-corrected using the FSL function, MCFLIRT (https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/MCFLIRT),⁸ then bidirectionally leakage-corrected using a leakage-correction algorithm.^9,10 Normalized rCBV (nrCBV) maps were generated using the bidirectional, leakage-corrected perfusion MRI signal^9,10 and then normalized using an automated method of dividing by the median brain rCBV value as done in a prior study.¹¹ All images were co-registered to the post-contrast T1-weighted MRI scan using linear registration (tkregister2, Freesurfer https://surfer.nmr.mgh.harvard.edu/fswiki/TkRegister¹² and FLIRT, FSL https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/FLIRT¹³).

Tumor Imaging Analysis

Three-dimensional volume of interest segmentations of the total T2/FLAIR hyperintense tumor, excluding necrotic regions, were generated using the deep learning software NS-HGlio artificial intelligence device (Neosoma Inc., Groton, MA, https://neosomainc.com). A radiologist with 16 years of experience (M.N.) inspected all tumor segmentations, and any segmentation errors were corrected using a semi-automated thresholding method using Analysis of Functional NeuroImages (AFNI, https://afni.nimh.nih.gov/) software.¹⁴ Tumors were considered “contrast-enhancing” if the contrast-enhancing tumor volume was > 1000 mm³, analogous to the definition of measurable disease (10 mm × 10 mm × 10 mm) in current Response Assessment in Neuro-Oncology 2.0 criteria¹⁵ and “nonenhancing” if otherwise. T2/FLAIR hyperintense tumor volume, ADC, nrCBV, and morphologic sphericity values were obtained for each patient using the Python package pyradiomics (Pyradiomics, https://radiomics.github.io/pyradiomics.html). To computationally perform hot-spot analyses, 10th-percentile ADC and 90th-percentile nrCBV values were also obtained as done in prior studies.^5,16,17 T2/FLAIR tumor volume growth rates were calculated for each patient using all timepoints leading up to the repeat surgery or biopsy that confirmed MT or non-MT among patients with 3+ imaging timepoints available. Patients who underwent multiple therapy regimens or multiple recurrences during the longitudinal study period were excluded, leaving n = 31 patients for T2/FLAIR tumor volume growth rate analysis (Figure 1). Growth rate values of millimeters cubed per day were calculated by fitting a linear regression curve within the T2/FLAIR tumor volume values for these timepoints. Similarly, median ADC rate of change and median nrCBV rate of change were calculated for patients with DWI (n = 29 patients) and DSC perfusion MRI (n = 21 patients) using the same criteria previously described for T2/FLAIR tumor volume growth rate analysis. Median ADC rate of change values and median nrCBV rate of change values were calculated by fitting a linear regression curve within the median ADC and median nrCBV values, respectively.

Statistical Analysis

All statistical analyses, aside from multiple logistic regression, were performed using GraphPad Prism Software. All analyses were performed separately within IDHm-A and IDHm-O. All statistical comparisons were conducted between different groups containing different individual patients (eg, patients with MT vs patients without MT). Fisher’s exact tests were performed to assess any relationships between the frequencies of contrast-enhancing and nonenhancing tumors between MT and non-MT categories. For group analyses, the D’Agostino and Pearson test was performed to assess the normality of the data, and parametric t-tests or nonparametric Mann–Whitney tests were performed to assess tumor volume, tumor growth rate, ADC, nrCBV, and sphericity differences between MT vs non-MT gliomas. Receiver operating characteristic (ROC) curves were calculated to evaluate the performance of tumor sphericity, growth rate, tumor volume, median ADC and its rate of change, and median nrCBV and its rate of change for classifying MT vs non-MT. These analyses were also repeated in only grade 4 MT vs grade 2 non-MT IDHm-A and in only grade 4 MT vs grades 2 and 3 IDHm-A. The significance level was set to P < .05. All bar graphs display the mean and 95% confidence interval.

Multiple logistic regression was performed to evaluate the combined classification performance of the cross-sectional imaging biomarkers from individual ROC results for classifying MT vs non-MT gliomas. A logistic regression model and subsequent ROC curves were created using Python’s scikit-learn (scikit-learn, https://scikit-learn.org/stable/) and scipy (SciPy, https://scipy.org) libraries. The results were plotted using GraphPad Prism Software. The DeLong test was used to compare ROC curves from multiple logistic regression and CE status alone for classifying MT.¹⁸

Results

Contrast-Enhancement, Volumetric, and Shape Characteristics

Two representative cases that underwent pathologically confirmed, contrast-enhancing MT, and nonenhancing non-MT are shown in Figure 2A and B, respectively. Patient A is a 38-year-old female patient diagnosed with a grade 4 IDHm-A, and patient B is a 71-year-old female patient diagnosed with a grade 2 IDHm-O. There was a significantly higher proportion of contrast-enhancing cases in the MT group compared to the non-MT group for both IDHm-A (P = .04, Figure 3A) and IDHm-O (P = .02, Figure 3A). A total of 81% of contrast-enhancing IDHm-A gliomas and 100% of contrast-enhancing IDHm-O gliomas underwent MT (Figure 3A). However, of all the gliomas that underwent MT, 41% of the IDHm-A and 62% of the IDHm-O were nonenhancing (Figure 3A). Of all the non-MT gliomas, 20% of the IDHm-A were contrast-enhancing, but none of the IDHm-O were contrast-enhancing (Figure 3A). The contrast-enhancement status at initial diagnosis and at final scan for the MT vs non-MT groups is summarized in Supplementary Tables 3 and 4. Within the IDHm-A MT group with contrast-enhancement at recurrence, n = 7 were newly contrast-enhancing and n = 6 were previously contrast-enhancing (Supplementary Table 3). Within the IDHm-O MT group with contrast-enhancement at recurrence, all n = 5 were newly contrast-enhancing (Supplementary Table 4).

Representative cases of two patients, one MT and the other non-MT showing T2-weighted FLAIR MRI, post-contrast T1-weighted MRI, ADC, and nrCBV at three timepoints to illustrate imaging changes in both patients. — Representative cases. Representative typical MT and non-MT cases. (A) Patient A is a 38-year-old female patient diagnosed with a contrast-enhancing grade 4 IDH-mutant astrocytoma (IDHm-A). (B) Patient B is a 71-year-old female patient diagnosed with a nonenhancing grade 2 IDH-mutant oligodendroglioma (IDHm-O). MT = malignant transformation, IDHm-A = isocitrate dehydrogenase mutant astrocytoma, IDHm-O = isocitrate dehydrogenase mutant oligodendroglioma, ADC = apparent diffusion coefficient, nrCBV = normalized relative cerebral blood volume, Sx = surgery.

Graphs on the anatomical MRI characteristics of malignantly transformed versus non-transformed IDH-mutant gliomas. Subfigures are labelled from a to f, illustrating statistical analyses. — Group differences and diagnostic performance of anatomical imaging features in malignant transformation. (A) For both IDHm-A (P < .05) and IDHm-O (P < .05), there was a significantly larger proportion of contrast-enhancing cases in the MT compared to non-MT groups, but there was also a considerable number of nonenhancing cases in the MT groups. (B) T2/FLAIR tumor volumes were significantly larger in the MT vs non-MT group for both IDHm-A (P < .05) and IDHm-O (P < .05). (C) T2/FLAIR tumor volume growth rates were significantly faster in the MT vs non-MT group for IDHm-A (P < .01) but not IDHm-O (P > .05). (D) Nonenhancing tumor sphericity was significantly lower in the MT vs non-MT group for IDHm-A (P < .05) but not IDHm-O (P > .05). (E) ROC curve analyses in IDHm-A showed an AUC = 0.696 for CE status (P < .05), AUC = 0.727 for T2/FLAIR tumor volume (P < .05), AUC = 0.863 for total growth rate (P < .01), and AUC = 0.718 for NET sphericity (P < .05). (F) ROC curve analyses in IDHm-O showed an AUC = 0.692 for CE status (P > .05), AUC = 0.731 for T2/FLAIR tumor volume (P < .05), AUC = 0.722 for total growth rate (P > .05), and AUC = 0.659 for NET sphericity (P > .05). IDHm-A = isocitrate dehydrogenase mutant astrocytoma, IDHm-O = isocitrate dehydrogenase mutant oligodendroglioma, MT = malignant transformation, NET = nonenhancing tumor, ROC = receiver operating characteristic, AUC = area under the curve, CE = contrast enhancement. * denotes P < .05, ** denotes P < .01

Total T2/FLAIR tumor volumes were significantly larger in the MT group than in the non-MT group for both IDHm-A (P = .02, mean (95% CI) MT vs non-MT: 75,916 (47,387–104,446) mm³ vs 35,031 (12,498–57,564) mm³, Figure 3B) and IDHm-O (P = .04, mean (95% CI) MT vs non-MT: 68,728 (37,677–99,779) mm³ vs 36,844 (21,963–51,725) mm³, Figure 3B). Total T2/FLAIR tumor growth rates were significantly larger in the MT group than in the non-MT group for only IDHm-A (P = .003, mean (95% CI) MT vs non-MT: 85.08 (23.63–146.5) mm³/day vs 19.67 (−5.543–44.88) mm³/day, Figure 3C) but not IDHm-O (P > .05, Figure 3C). The percent change in tumor volume over time is shown in Supplementary Figure 1 for both the IDHm-A (Supplementary Figure 1A) and IDHm-O (Supplementary Figure 1B) groups. Lastly, while total T2/FLAIR tumor sphericity was not significantly different between MT and non-MT IDHm-A (P > .05), the sphericity of specifically the nonenhancing tumor was significantly lower in MT than non-MT IDHm-A (P = .03, mean (95% CI) MT vs non-MT: 0.3708 (0.3411–0.4004) vs 0.4336 (0.3742–0.4929), Figure 3D). Similar group differences in proportion of enhancing cases, volume, and sphericity were observed when assessing only grade 4 IDHm-A MT vs non-MT groups (Supplementary Figure 2) and when assessing grade 4 IDHm-A MT vs grades 2/3 IDHm-A groups (Supplementary Figure 3).

ROC curve analyses of these anatomical MRI biomarkers for classifying MT vs non-MT in IDHm-A and IDHm-O are shown in Figure 3E and F. Contrast-enhancement status showed an AUC = 0.696 for IDHm-A (P < .05, Figure 3E) and an AUC = 0.692 for IDHm-O (P > .05, Figure 3F), total T2/FLAIR tumor volume showed an AUC = 0.727 for IDHm-A (P = .02, Figure 3E) and an AUC = 0.731 for IDHm-O (P = .04, Figure 3F), T2/FLAIR tumor volume growth rate showed an AUC = 0.863 for IDHm-A (P = .005, Figure 3E) and an AUC = 0.722 for IDHm-O (P > .05, Figure 3F), and nonenhancing tumor sphericity had an AUC = 0.718 for IDHm-A (P = .02, Figure 3E) and an AUC = 0.659 for IDHm-O (P > .05, Figure 3F).

Diffusion and Perfusion Characteristics

When assessing ADC and nrCBV, there were MT vs non-MT differences for IDHm-A but not for IDHm-O. Specifically, median ADC in the MT group trended toward being significantly lower than the non-MT group for IDHm-A (P = .06, mean (95% CI) MT vs non-MT: 1.267 (1.185–1.348) μm²/ms vs 1.363 (1.314–1.412) μm²/ms, Figure 4A). The 10th-percentile ADC was significantly lower in the MT group for IDHm-A (P = .03, Supplementary Figure 4), but there was no significant difference between the MT vs non-MT groups for IDHm-O (P > .05, Supplementary Figure 4). The 90th-percentile nrCBV was not significantly different between MT and non-MT groups for either IDHm-A or IDHm-O (P > .05, Supplementary Figure 4). Median nrCBV was significantly higher in MT than non-MT IDHm-A (P = .002, mean (95% CI) MT vs non-MT: 0.7859 (0.6267–0.9451) vs 0.5240 (0.4658–0.5822), Figure 4B). Neither T2/FLAIR tumor median ADC nor median nrCBV was significantly different between MT and non-MT IDHm-O (P > .05, Figure 4A and B). Similarly, there was significantly lower ADC and increased nrCBV when assessing only grade 4 IDHm-A MT vs grade 2 IDHm-A (Supplementary Figure 2) and when assessing grade 4 IDHm-A MT vs grades 2/3 IDHm-A groups (Supplementary Figure 3).

Graphs on the diffusion and perfusion MRI characteristics of malignantly transformed versus non-transformed IDH-mutant gliomas. Subfigures are labelled from a to f, illustrating statistical analyses. — Diffusion and perfusion differences in malignant transformation. (A and B) Within IDHm-A only, the median apparent diffusion coefficient (ADC) trended lower, and the normalized relative cerebral blood volume (nrCBV) was significantly higher in the T2/FLAIR tumor (P < 0.01). Neither of these imaging biomarkers were significantly different between MT vs non-MT IDHm-O. (C and D) The rate of change of median ADC and rate of change of median nrCBV was not significantly different between MT vs non-MT in IDHm-A or IDHm-O. (E) ROC curve analyses in IDHm-A showed an AUC = 0.686 for median ADC (P > .05), AUC = 0.839 for median nrCBV (P < .05), AUC = 0.635 for ADC rate of change (P > .05), and AUC = 0.537 for nrCBV rate of change (P > .05). (F) ROC curve analyses in IDHm-O showed an AUC = 0.621 for median ADC (P > .05), AUC = 0.513 for median nrCBV (P > .05), AUC = 0.833 for ADC rate of change (P > .05), and AUC = 0.889 for nrCBV rate of change (P > .05). IDHm-A = isocitrate dehydrogenase mutant astrocytoma, IDHm-O = isocitrate dehydrogenase mutant oligodendroglioma, MT = malignant transformation, ADC = apparent diffusion coefficient, nrCBV = normalized relative cerebral blood volume, ROC = receiver operating characteristic, AUC = area under the curve. ** denotes P < .01

No group differences in the median ADC rate of change or median nrCBV rate of change analyses were observed between the MT and non-MT cohorts for either IDHm-A or IDHm-O (P > .05, Figure 4C and D). ROC curve analyses using median ADC to classify MT vs non-MT showed an AUC = 0.686 for IDHm-A (P > .05, Figure 4E) and an AUC = 0.621 for IDHm-O (P > .05, Figure 4F), while ROC curve analyses using median nrCBV showed an AUC = 0.839 for IDHm-A (P = .003, Figure 4E) and an AUC = 0.513 for IDHm-O (P > .05, Figure 4F). In addition, ROC curve analyses using median ADC rate of change to classify MT vs non-MT showed an AUC = 0.635 for IDHm-A (P > .05, Figure 4E) and an AUC = 0.833 for IDHm-O (P > .05, Figure 4F) while ROC curve analyses using median nrCBV rate of change showed an AUC = 0.537 for IDHm-A(P > .05, Figure 4E) and an AUC = 0.889 for IDHm-O (P > .05, Figure 4F).

Multiple logistic regression ROC analyses using the cross-sectional characteristics of presence of contrast-enhancement, T2/FLAIR tumor volume, nonenhancing tumor sphericity, median ADC, and median nrCBV showed a higher AUC in the IDHm-A group (AUC = 0.881) compared to using the presence of contrast-enhancement alone (AUC = 0.696), though this difference was not statistically significant (P = .088, Figure 5A and B). Similar results were observed in the IDHm-O group, where the multiple logistic regression model showed an AUC of 0.826 compared to the presence of contrast-enhancement alone (AUC = 0.692), but the difference did not reach statistical significance (P = .239, Figure 5E and F). When T2/FLAIR tumor growth rate was added to the multiple logistic regression model for the IDHm-A group, the model achieved an AUC of 0.898 compared to the presence of contrast-enhancement alone (AUC = 0.696), but the difference remained statistically nonsignificant (P = .070, Figure 5C and D). This analysis could not be performed for the IDHm-O group due to insufficient data points.

Graphs showing the comparison of contrast-enhancement versus multiple logistic regression for classification of malignant transformation. Graphs are labelled from a to f, showing receiver operating characteristic curves and predicted probability distributions from the multiple logistic regression model. — Multiple logistic regression for predicting malignant transformation. (A and B) When performing multiple logistic regression using CE status, T2/FLAIR tumor volume, NET sphericity, median ADC, and median nrCBV to assess classification between MT vs non-MT, ROC curve analyses in the IDHm-A group showed an AUC = 0.881 (P = .009) (A and B) while the IDHm-O group showed an AUC = 0.826 (P = .013) (E and F). When including the T2/FLAIR tumor growth rate in the multiple logistic regression to classify MT, ROC curve analyses in the IDHm-A group showed an AUC = 0.898 (P = .025) (C and D), but this could not be performed in the IDHm-O group due to a lack of available growth rate data. Multiple logistic regression classified MT better than CE status alone, which showed AUC = 0.696 for IDHm-A (A–D) and AUC = 0.692 IDHm-O (E and F), though these differences were statistically nonsignificant (P > .05). NET = nonenhancing tumor, ADC = apparent diffusion coefficient, nrCBV = normalized relative cerebral blood volume, CE = contrast enhancement, MT = malignant transformation, ROC = receiver operating characteristic, AUC = area under the curve, IDHm-A = isocitrate dehydrogenase mutant astrocytoma, IDHm-O = isocitrate dehydrogenase mutant oligodendroglioma

Discussion

The primary findings of this study are that: (1) beyond contrast-enhancement, large tumor volume is useful for identifying MT in both IDHm-A and IDHm-O, and (2) shape, growth rate, diffusion, and perfusion characteristics are useful for identifying MT only in IDHm-A, but not in IDHm-O. This study adds to the literature by utilizing the contemporary, molecularly defined definition of glioma subtypes to study a unique cohort of patients with pathologically confirmed MT via repeat biopsy.

The current study’s observation that contrast-enhancement is a useful, but not perfectly indicative, biomarker for MT of grade 2 IDH-mutant gliomas to grade 3/4 IDH-mutant gliomas is consistent with previous research on gliomas.^3,4 Nearly half of the IDHm-A and IDHm-O cases with MT were nonenhancing, and some IDHm-A cases with non-MT were contrast-enhancing. These findings highlight the importance of using pathological confirmation of repeated biopsy as the definitive ground truth of updated MT tumor grading, rather than relying solely on the presence of contrast-enhancement. In addition, the study classified the presence of contrast-enhancement by utilizing a volumetric threshold of > 1000 mm³, adapted from the RANO 2.0 criteria of “measurable disease” for clinical trials,¹⁵ which may be useful for quantitative imaging studies for standardized assessment of contrast-enhancement.

This study also adds to the growing literature on the usefulness of diffusion and perfusion MRI for assessing glioma group differences.^5,6,19 MT in IDHm-A was associated with lower ADC values in T2/FLAIR hyperintense tumor regions, which aligns with prior studies on gliomas evaluating MT⁵ and tumor grade differences.²⁰ This study also utilized the 10th-percentile ADC and 90th-percentile nrCBV metrics as done in prior studies on glioma,^5,16,17 suggesting that percentile-based metrics analogous to hot-spot approaches may be valuable for identifying subtle tumor differences. Tumor angiogenesis also occurs with tumor growth,²¹ which is in line with the finding of higher nrCBV in the IDHm-A MT cohort compared to non-MT. Notably, diffusion and perfusion features were not useful for identifying MT in IDHm-O. IDHm-Os are known to have lower diffusion and higher perfusion characteristics than IDHm-A,¹⁹ which may explain why these imaging modalities may not be sufficiently sensitive to detect tumor changes associated with MT in IDHm-O. However, tumor volume and contrast-enhancement were able to identify IDHm-O MT, suggesting that they may serve as useful biomarkers for MT in IDHm-O.

This study also explored other advanced imaging biomarkers for characterizing MT. There was lower nonenhancing tumor sphericity in the MT cohort compared to non-MT for IDHm-A, which may reflect the infiltrative tumor behavior associated with more aggressive tumors.²² The shape feature of sphericity has begun to be utilized to study IDH-wild-type glioblastoma^23,24 and brain metastases,¹¹ and the present study demonstrates that sphericity may also have utility for studying IDHm-A MT. Furthermore, faster tumor growth rates were found to be useful for identifying MT in IDHm-A, but not IDHm-O. This finding could be due to IDHm-A being generally more aggressive and consequently having a worse prognosis compared to IDHm-O.²⁵ Furthermore, rates of change of tumor volumetric-based median ADC and median nrCBV were not useful for identifying MT in IDHm-A or IDHm-O. Future studies may consider “hot spot” image analysis approaches as done clinically and utilizing standardized number and timing of scans for longitudinal analyses, both of which were unable to be performed in this study, to further investigate longitudinal rates of changes in ADC and nrCBV.

It is also important to note the ongoing debate in the field regarding the distinction between grade 2 and grade 3 IDHm-A.^26,27 While the term “malignant transformation” in IDHm-A has been defined in previous literature as the transformation from grade 2 to 3 or from grade 2 to 4,^28,29 the present study included additional analyses combining grades 2/3 as one entity compared to grade 4 IDHm-A, as well as analyses limited to grade 2 vs grade 4 IDHm-A. Interestingly, the results showed the same overall directionality, with imaging features associated with grades 3 or 4 IDHm-A MT also being associated with cases of MT strictly to grade 4.

This study has additional limitations that should be addressed. The sample size was limited, and the study population was relatively heterogeneous in treatment status, which precluded the possibility of any subgroup analyses based on treatments, and reason for undergoing repeat biopsy/surgery. Because this study relied on requiring a second pathology analysis via biopsy/surgery, there is a bias for including more symptomatic patients or patients with evolving MRI findings, and likely more MT patients, than what would be representative of the broader IDH-mutant glioma patient population. For example, many patients may not undergo repeat biopsy/surgery due to adequate disease control with medical or radiation therapy, or also due to inoperable disease or poor surgical candidacy, and such patients are not represented in this study cohort. Nevertheless, the present findings remain valuable to our field because this study further advances our knowledge on imaging characteristics of IDHm glioma MT by studying a unique cohort with pathological confirmation. Further analyses on expanded or independent validation cohorts would be useful for supporting our findings. Another major limitation of this study was the variable time intervals between surgeries, including between the IDHm-A and IDHm-O groups, due to the retrospective nature of the present study and inclusion criteria for this study cohort to investigate imaging biomarkers of pathologically confirmed MT. This variability in time is also why longitudinal analyses of imaging biomarkers leading up to MT were limited to just 3 + timepoint-based tumor growth rate calculations. Future studies with less variable time intervals between scans and surgeries would be valuable. In addition, multiple comparisons corrections were not performed in this study. Lastly, the availability of more advanced metabolic MR techniques, such as CEST MRI³⁰ or MR spectroscopy,³¹ in future studies could provide insights into the metabolic shifts associated with MT³² beyond tumor cellularity and perfusivity alterations.

Conclusions

The presence of measurable contrast-enhancement does not perfectly indicate MT in IDHm gliomas. Tumor volume is useful for identifying MT in both IDHm-A and IDHm-O. Shape, growth rate, diffusion, and perfusion characteristics were only useful for identifying MT in IDHm-A but not in IDHm-O, reflecting the need for molecular subtype-specific imaging biomarkers. Further studies using advanced MR techniques to characterize MT biomarkers specific to IDHm-A and IDHm-O would be useful.

Supplementary material

Supplementary material is available online at Neuro-Oncology (https://academic.oup.com/neuro-oncology).

noaf171_Supplementary_Tables_1-4_Figures_1-4

noaf171_supplementary_tables_1-4_figures_1-4.docx^{(676.1KB, docx)}

Prior Presentation

This manuscript is adapted from a chapter in a doctoral dissertation. A portion of the results were presented at the 2024 Society for Neuro-Oncology Annual Meeting.

Contributor Information

Nicholas S Cho, Medical Scientist Training Program, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, CA, USA; Department of Bioengineering, Henry Samueli School of Engineering and Applied Science, University of California, Los Angeles, Los Angeles, CA, USA; Department of Radiological Sciences, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, CA, USA; UCLA Brain Tumor Imaging Laboratory (BTIL), Center for Computer Vision and Imaging Biomarkers, University of California, Los Angeles, Los Angeles, CA, USA.

Viên Lam Le, Department of Bioengineering, Henry Samueli School of Engineering and Applied Science, University of California, Los Angeles, Los Angeles, CA, USA; Department of Radiological Sciences, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, CA, USA; UCLA Brain Tumor Imaging Laboratory (BTIL), Center for Computer Vision and Imaging Biomarkers, University of California, Los Angeles, Los Angeles, CA, USA.

Ashley Teraishi, Department of Radiological Sciences, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, CA, USA; UCLA Brain Tumor Imaging Laboratory (BTIL), Center for Computer Vision and Imaging Biomarkers, University of California, Los Angeles, Los Angeles, CA, USA.

Vicki Liu, Department of Neurology, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, CA, USA; UCLA Neuro-Oncology Program, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, CA, USA.

Francesco Sanvito, Department of Radiological Sciences, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, CA, USA; UCLA Brain Tumor Imaging Laboratory (BTIL), Center for Computer Vision and Imaging Biomarkers, University of California, Los Angeles, Los Angeles, CA, USA.

Donatello Telesca, Department of Biostatistics, Fielding School of Public Health, University of California Los Angeles, Los Angeles, CA, USA.

Masanori Nakajo, Department of Radiological Sciences, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, CA, USA; UCLA Brain Tumor Imaging Laboratory (BTIL), Center for Computer Vision and Imaging Biomarkers, University of California, Los Angeles, Los Angeles, CA, USA.

Chencai Wang, Department of Radiological Sciences, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, CA, USA; UCLA Brain Tumor Imaging Laboratory (BTIL), Center for Computer Vision and Imaging Biomarkers, University of California, Los Angeles, Los Angeles, CA, USA.

Sonoko Oshima, Department of Radiological Sciences, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, CA, USA; UCLA Brain Tumor Imaging Laboratory (BTIL), Center for Computer Vision and Imaging Biomarkers, University of California, Los Angeles, Los Angeles, CA, USA.

Blaine S C Eldred, Department of Neurology, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, CA, USA; UCLA Neuro-Oncology Program, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, CA, USA.

Jingwen Yao, Department of Radiological Sciences, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, CA, USA; UCLA Brain Tumor Imaging Laboratory (BTIL), Center for Computer Vision and Imaging Biomarkers, University of California, Los Angeles, Los Angeles, CA, USA.

Phioanh L Nghiemphu, Department of Neurology, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, CA, USA; UCLA Neuro-Oncology Program, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, CA, USA.

Noriko Salamon, Department of Radiological Sciences, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, CA, USA.

Timothy F Cloughesy, Department of Neurology, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, CA, USA; UCLA Neuro-Oncology Program, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, CA, USA.

Albert Lai, Department of Neurology, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, CA, USA; UCLA Neuro-Oncology Program, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, CA, USA.

Benjamin M Ellingson, Department of Psychiatry and Biobehavioral Sciences, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, CA, USA; Department of Neurosurgery, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, CA, USA; Department of Bioengineering, Henry Samueli School of Engineering and Applied Science, University of California, Los Angeles, Los Angeles, CA, USA; Department of Radiological Sciences, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, CA, USA; UCLA Brain Tumor Imaging Laboratory (BTIL), Center for Computer Vision and Imaging Biomarkers, University of California, Los Angeles, Los Angeles, CA, USA.

Conflict of interest statement. B.M.E. is on the advisory board and is a paid consultant for Alpheus Medical, Carthera, Chimerix, Ellipses Pharma, Erasca, Global Coalition for Adaptive Research (GCAR), Imaging Endpoints, Medicenna, Voiant, Medscape, Monteris, Neosoma, Nerviano Medical Sciences, Nuvation Bio, Orbus Therapeutics, Sagimet Biosciences, Sapience Therapeutics, Servier Pharmaceuticals, Siemens, SonALAsense, Sumitomo Dianippon Pharma Oncology, Telix, the Sontag Foundation, the National Brain Tumor Society, and Third Rock Ventures. T.F.C. is cofounder, major stock holder, consultant and board member of Katmai Pharmaceuticals, holds stock for Erasca, member of the board and paid consultant for the 501c3 Global Coalition for Adaptive Research, holds stock in Chimerix and receives milestone payments and possible future royalties, member of the scientific advisory board for Break Through Cancer, member of the scientific advisory board for Cure Brain Cancer Foundation, has provided paid consulting services to Blue Rock, Vida Ventures, Lista Therapeutics, Stemline, Novartis, Roche, Sonalasense, Sagimet, Clinical Care Options, Ideology Health, Servier, Jubilant, Immvira, Gan & Lee, BrainStorm, Katmai, Sapience, Inovio, Vigeo Therapeutics, DNATrix, Tyme, SDP, Kintara, Bayer, Merck, Boehinger Ingelheim, VBL, Amgen, Kiyatec, Odonate Therapeutics QED, Medefield, Pascal Biosciences, Bayer, Tocagen, Karyopharm, GW Pharma, Abbvie, VBI, Deciphera, VBL, Agios, Genocea, Celgene, Puma, Lilly, BMS, Cortice, Novocure, Novogen, Boston Biomedical, Sunovion, Insys, Pfizer, Notable labs, Medqia, Trizel, Medscape, and has contracts with UCLA for the Brain Tumor Program with Roche, VBI, Merck, Novartis, BMS, AstraZeneca, Servier. The Regents of the University of California (T.F.C. employer) has licensed intellectual property co-invented by T.F.C. to Katmai Pharmaceuticals. P.L.N. has received grants/contracts from ERASCA, Millenium, Children’s Tumor Foundation, Dept of Defense, GCAR, Springsworks, and BMS and has received payment/honoraria from Alexion.

Funding

This work was supported by the National Institutes of Health, The National Institute of General Medical Sciences (T32GM008042 to N.S.C., T32GM152342 to N.S.C.), the National Institutes of Health National Cancer Institute (F30CA284809 to N.S.C., R01CA270027 to B.M.E. and T.F.C., R01CA279984 to B.M.E., P50CA211015 to B.M.E. and T.F.C.), and the Department of Defense Congressionally Directed Medical Research Programs (CA220732 to B.M.E. and T.F.C.).

Author contributions

Study design: N.S.C., V.L.L., A.T., B.M.E. Data collection: all authors. Statistical analysis: N.S.C., V.L.L., A.T., D.T., and B.M.E. Manuscript preparation: all authors.

Data availability

The datasets for the results presented in this study can be made available from the corresponding author (B.M.E.) upon reasonable request.

References

1. Murphy ES, Leyrer CM, Parsons M, et al. Risk factors for malignant transformation of low-grade glioma. Int J Radiat Oncol Biol Phys. 2018;100(4):965–971. [DOI] [PubMed] [Google Scholar]
2. Rees J, Watt H, Jäger HR, et al. Volumes and growth rates of untreated adult low-grade gliomas indicate risk of early malignant transformation. Eur J Radiol. 2009;72(1):54–64. [DOI] [PubMed] [Google Scholar]
3. Pierallini A, Bonamini M, Bozzao A, et al. Supratentorial diffuse astrocytic tumours: proposal of an MRI classification. Eur Radiol. 1997;7(3):395–399. [DOI] [PubMed] [Google Scholar]
4. Scott JN, Brasher PMA, Sevick RJ, Rewcastle NB, Forsyth PA.. How often are nonenhancing supratentorial gliomas malignant? A population study. Neurology. 2002;59(6):947–949. [DOI] [PubMed] [Google Scholar]
5. Chen IE, Swinburne N, Tsankova NM, et al. Sequential apparent diffusion coefficient for assessment of tumor progression in patients with low-grade glioma. AJNR Am J Neuroradiol. 2018;39(6):1039–1046. [DOI] [PMC free article] [PubMed] [Google Scholar]
6. Danchaivijitr N, Waldman AD, Tozer DJ, et al. Low-grade gliomas: do changes in rCBV measurements at longitudinal perfusion-weighted MR imaging predict malignant transformation? Radiology. 2008;247(1):170–178. [DOI] [PubMed] [Google Scholar]
7. Louis DN, Perry A, Wesseling P, et al. The 2021 WHO Classification of Tumors of the Central Nervous System: a summary. Neuro Oncol. 2021;23(8):1231–1251. [DOI] [PMC free article] [PubMed] [Google Scholar]
8. Jenkinson M, Beckmann CF, Behrens TE, Woolrich MW, Smith SM.. FSL. Neuroimage. 2012;62(2):782–790. [DOI] [PubMed] [Google Scholar]
9. Leu K, Boxerman JL, Cloughesy TF, et al. Improved leakage correction for single-echo dynamic susceptibility contrast perfusion MRI estimates of relative cerebral blood volume in high-grade gliomas by accounting for bidirectional contrast agent exchange. AJNR Am J Neuroradiol. 2016;37(8):1440–1446. [DOI] [PMC free article] [PubMed] [Google Scholar]
10. Leu K, Boxerman JL, Lai A, et al. Bidirectional Contrast agent leakage correction of dynamic susceptibility contrast (DSC)-MRI improves cerebral blood volume estimation and survival prediction in recurrent glioblastoma treated with bevacizumab. J Magn Reson Imaging. 2016;44(5):1229–1237. [DOI] [PubMed] [Google Scholar]
11. Sanvito F, Telesca D, Cho NS, et al. Small pretreatment lesion size and high sphericity as favorable prognostic factors after laser interstitial thermal therapy in brain metastases. J Neurosurg. 2024;140(2):338–349. [DOI] [PMC free article] [PubMed] [Google Scholar]
12. Fischl B. FreeSurfer. Neuroimage. 2012;62(2):774–781. [DOI] [PMC free article] [PubMed] [Google Scholar]
13. Smith SM, Jenkinson M, Woolrich MW, et al. Advances in functional and structural MR image analysis and implementation as FSL. Neuroimage. 2004;23(Suppl 1):S208–S219. [DOI] [PubMed] [Google Scholar]
14. Ellingson BM, Kim HJ, Woodworth DC, et al. Recurrent glioblastoma treated with bevacizumab: contrast-enhanced T1-weighted subtraction maps improve tumor delineation and aid prediction of survival in a multicenter clinical trial. Radiology. 2014;271(1):200–210. [DOI] [PMC free article] [PubMed] [Google Scholar]
15. Wen PY, van den Bent M, Youssef G, et al. RANO 2.0: update to the response assessment in neuro-oncology criteria for high- and low-grade gliomas in adults. J Clin Oncol. 2023;41(33):5187–5199. [DOI] [PMC free article] [PubMed] [Google Scholar]
16. Hempel JM, Schittenhelm J, Klose U, et al. In vivo molecular profiling of human glioma: cross-sectional observational study using dynamic susceptibility contrast magnetic resonance perfusion imaging. Clin Neuroradiol. 2019;29(3):479–491. [DOI] [PubMed] [Google Scholar]
17. Kickingereder P, Sahm F, Radbruch A, et al. IDH mutation status is associated with a distinct hypoxia/angiogenesis transcriptome signature which is non-invasively predictable with rCBV imaging in human glioma. Sci Rep. 2015;5(1):16238. [DOI] [PMC free article] [PubMed] [Google Scholar]
18. DeLong ER, DeLong DM, Clarke-Pearson DL.. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics. 1988;44(3):837–845. [PubMed] [Google Scholar]
19. Leu K, Ott GA, Lai A, et al. Perfusion and diffusion MRI signatures in histologic and genetic subtypes of WHO grade II–III diffuse gliomas. J Neurooncol. 2017;134(1):177–188. [DOI] [PMC free article] [PubMed] [Google Scholar]
20. Arevalo-Perez J, Peck KK, Young RJ, et al. Dynamic contrast-enhanced perfusion MRI and diffusion-weighted imaging in grading of gliomas. J Neuroimaging. 2015;25(5):792–798. [DOI] [PMC free article] [PubMed] [Google Scholar]
21. Jensen RL. Brain tumor hypoxia: tumorigenesis, angiogenesis, imaging, pseudoprogression, and as a therapeutic target. J Neurooncol. 2009;92(3):317–335. [DOI] [PubMed] [Google Scholar]
22. Barajas RF, Hodgson JG, Chang JS, et al. Glioblastoma multiforme regional genetic and cellular expression patterns: influence on anatomic and physiologic MR imaging. Radiology. 2010;254(2):564–576. [DOI] [PMC free article] [PubMed] [Google Scholar]
23. Sanghani P, Ti AB, Kam King NK, Ren H.. Evaluation of tumor shape features for overall survival prognosis in glioblastoma multiforme patients. Surg Oncol. 2019;29:178–183. [DOI] [PubMed] [Google Scholar]
24. Shidoh S, Savjani RR, Cho NS, et al. Relapse patterns and radiation dose exposure in IDH wild-type glioblastoma at first radiographic recurrence following chemoradiation. J Neurooncol. 2022;160(1):115–125. [DOI] [PMC free article] [PubMed] [Google Scholar]
25. Pekmezci M, Rice T, Molinaro AM, et al. Adult infiltrating gliomas with WHO 2016 integrated diagnosis: additional prognostic roles of ATRX and TERT. Acta Neuropathol. 2017;133(6):1001–1016. [DOI] [PMC free article] [PubMed] [Google Scholar]
26. van den Bent MJ. Interobserver variation of the histopathological diagnosis in clinical trials on glioma: a clinician’s perspective. Acta Neuropathol. 2010;120(3):297–304. [DOI] [PMC free article] [PubMed] [Google Scholar]
27. van den Bent MJ, French PJ, Brat D, et al. The biological significance of tumor grade, age, enhancement, and extent of resection in IDH-mutant gliomas: how should they inform treatment decisions in the era of IDH inhibitors? Neuro-Oncology. 2024;26(10):1805–1822. [DOI] [PMC free article] [PubMed] [Google Scholar]
28. Jiang H, Zhu Q, Wang X, et al. Characterization and clinical implications of different malignant transformation patterns in diffuse low-grade gliomas. Cancer Sci. 2023;114(9):3708–3718. [DOI] [PMC free article] [PubMed] [Google Scholar]
29. Satar Z, Hotton G, Samandouras G.. Systematic review—time to malignant transformation in low-grade gliomas: predicting a catastrophic event with clinical, neuroimaging, and molecular markers. Neurooncol Adv. 2021;3(1):vdab101. [DOI] [PMC free article] [PubMed] [Google Scholar]
30. Cho NS, Hagiwara A, Yao J, et al. Amine-weighted chemical exchange saturation transfer magnetic resonance imaging in brain tumors. NMR Biomed. 2023;36(6):e4785. [DOI] [PMC free article] [PubMed] [Google Scholar]
31. Avalos LN, Luks TL, Gleason T, et al. Longitudinal MR spectroscopy to detect progression in patients with lower-grade glioma in the surveillance phase. Neurooncol Adv. 2022;4(1):vdac175. [DOI] [PMC free article] [PubMed] [Google Scholar]
32. Ruiz-Rodado V, Malta TM, Seki T, et al. Metabolic reprogramming associated with aggressiveness occurs in the G-CIMP-high molecular subtypes of IDH1mut lower grade gliomas. Neuro Oncol. 2020;22(4):480–492. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

noaf171_Supplementary_Tables_1-4_Figures_1-4

noaf171_supplementary_tables_1-4_figures_1-4.docx^{(676.1KB, docx)}

Data Availability Statement

The datasets for the results presented in this study can be made available from the corresponding author (B.M.E.) upon reasonable request.

[CIT0001] 1. Murphy ES, Leyrer CM, Parsons M, et al. Risk factors for malignant transformation of low-grade glioma. Int J Radiat Oncol Biol Phys. 2018;100(4):965–971. [DOI] [PubMed] [Google Scholar]

[CIT0002] 2. Rees J, Watt H, Jäger HR, et al. Volumes and growth rates of untreated adult low-grade gliomas indicate risk of early malignant transformation. Eur J Radiol. 2009;72(1):54–64. [DOI] [PubMed] [Google Scholar]

[CIT0003] 3. Pierallini A, Bonamini M, Bozzao A, et al. Supratentorial diffuse astrocytic tumours: proposal of an MRI classification. Eur Radiol. 1997;7(3):395–399. [DOI] [PubMed] [Google Scholar]

[CIT0004] 4. Scott JN, Brasher PMA, Sevick RJ, Rewcastle NB, Forsyth PA.. How often are nonenhancing supratentorial gliomas malignant? A population study. Neurology. 2002;59(6):947–949. [DOI] [PubMed] [Google Scholar]

[CIT0005] 5. Chen IE, Swinburne N, Tsankova NM, et al. Sequential apparent diffusion coefficient for assessment of tumor progression in patients with low-grade glioma. AJNR Am J Neuroradiol. 2018;39(6):1039–1046. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0006] 6. Danchaivijitr N, Waldman AD, Tozer DJ, et al. Low-grade gliomas: do changes in rCBV measurements at longitudinal perfusion-weighted MR imaging predict malignant transformation? Radiology. 2008;247(1):170–178. [DOI] [PubMed] [Google Scholar]

[CIT0007] 7. Louis DN, Perry A, Wesseling P, et al. The 2021 WHO Classification of Tumors of the Central Nervous System: a summary. Neuro Oncol. 2021;23(8):1231–1251. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0008] 8. Jenkinson M, Beckmann CF, Behrens TE, Woolrich MW, Smith SM.. FSL. Neuroimage. 2012;62(2):782–790. [DOI] [PubMed] [Google Scholar]

[CIT0009] 9. Leu K, Boxerman JL, Cloughesy TF, et al. Improved leakage correction for single-echo dynamic susceptibility contrast perfusion MRI estimates of relative cerebral blood volume in high-grade gliomas by accounting for bidirectional contrast agent exchange. AJNR Am J Neuroradiol. 2016;37(8):1440–1446. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0010] 10. Leu K, Boxerman JL, Lai A, et al. Bidirectional Contrast agent leakage correction of dynamic susceptibility contrast (DSC)-MRI improves cerebral blood volume estimation and survival prediction in recurrent glioblastoma treated with bevacizumab. J Magn Reson Imaging. 2016;44(5):1229–1237. [DOI] [PubMed] [Google Scholar]

[CIT0011] 11. Sanvito F, Telesca D, Cho NS, et al. Small pretreatment lesion size and high sphericity as favorable prognostic factors after laser interstitial thermal therapy in brain metastases. J Neurosurg. 2024;140(2):338–349. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0012] 12. Fischl B. FreeSurfer. Neuroimage. 2012;62(2):774–781. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0013] 13. Smith SM, Jenkinson M, Woolrich MW, et al. Advances in functional and structural MR image analysis and implementation as FSL. Neuroimage. 2004;23(Suppl 1):S208–S219. [DOI] [PubMed] [Google Scholar]

[CIT0014] 14. Ellingson BM, Kim HJ, Woodworth DC, et al. Recurrent glioblastoma treated with bevacizumab: contrast-enhanced T1-weighted subtraction maps improve tumor delineation and aid prediction of survival in a multicenter clinical trial. Radiology. 2014;271(1):200–210. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0015] 15. Wen PY, van den Bent M, Youssef G, et al. RANO 2.0: update to the response assessment in neuro-oncology criteria for high- and low-grade gliomas in adults. J Clin Oncol. 2023;41(33):5187–5199. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0016] 16. Hempel JM, Schittenhelm J, Klose U, et al. In vivo molecular profiling of human glioma: cross-sectional observational study using dynamic susceptibility contrast magnetic resonance perfusion imaging. Clin Neuroradiol. 2019;29(3):479–491. [DOI] [PubMed] [Google Scholar]

[CIT0017] 17. Kickingereder P, Sahm F, Radbruch A, et al. IDH mutation status is associated with a distinct hypoxia/angiogenesis transcriptome signature which is non-invasively predictable with rCBV imaging in human glioma. Sci Rep. 2015;5(1):16238. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0018] 18. DeLong ER, DeLong DM, Clarke-Pearson DL.. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics. 1988;44(3):837–845. [PubMed] [Google Scholar]

[CIT0019] 19. Leu K, Ott GA, Lai A, et al. Perfusion and diffusion MRI signatures in histologic and genetic subtypes of WHO grade II–III diffuse gliomas. J Neurooncol. 2017;134(1):177–188. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0020] 20. Arevalo-Perez J, Peck KK, Young RJ, et al. Dynamic contrast-enhanced perfusion MRI and diffusion-weighted imaging in grading of gliomas. J Neuroimaging. 2015;25(5):792–798. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0021] 21. Jensen RL. Brain tumor hypoxia: tumorigenesis, angiogenesis, imaging, pseudoprogression, and as a therapeutic target. J Neurooncol. 2009;92(3):317–335. [DOI] [PubMed] [Google Scholar]

[CIT0022] 22. Barajas RF, Hodgson JG, Chang JS, et al. Glioblastoma multiforme regional genetic and cellular expression patterns: influence on anatomic and physiologic MR imaging. Radiology. 2010;254(2):564–576. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0023] 23. Sanghani P, Ti AB, Kam King NK, Ren H.. Evaluation of tumor shape features for overall survival prognosis in glioblastoma multiforme patients. Surg Oncol. 2019;29:178–183. [DOI] [PubMed] [Google Scholar]

[CIT0024] 24. Shidoh S, Savjani RR, Cho NS, et al. Relapse patterns and radiation dose exposure in IDH wild-type glioblastoma at first radiographic recurrence following chemoradiation. J Neurooncol. 2022;160(1):115–125. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0025] 25. Pekmezci M, Rice T, Molinaro AM, et al. Adult infiltrating gliomas with WHO 2016 integrated diagnosis: additional prognostic roles of ATRX and TERT. Acta Neuropathol. 2017;133(6):1001–1016. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0026] 26. van den Bent MJ. Interobserver variation of the histopathological diagnosis in clinical trials on glioma: a clinician’s perspective. Acta Neuropathol. 2010;120(3):297–304. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0027] 27. van den Bent MJ, French PJ, Brat D, et al. The biological significance of tumor grade, age, enhancement, and extent of resection in IDH-mutant gliomas: how should they inform treatment decisions in the era of IDH inhibitors? Neuro-Oncology. 2024;26(10):1805–1822. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0028] 28. Jiang H, Zhu Q, Wang X, et al. Characterization and clinical implications of different malignant transformation patterns in diffuse low-grade gliomas. Cancer Sci. 2023;114(9):3708–3718. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0029] 29. Satar Z, Hotton G, Samandouras G.. Systematic review—time to malignant transformation in low-grade gliomas: predicting a catastrophic event with clinical, neuroimaging, and molecular markers. Neurooncol Adv. 2021;3(1):vdab101. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0030] 30. Cho NS, Hagiwara A, Yao J, et al. Amine-weighted chemical exchange saturation transfer magnetic resonance imaging in brain tumors. NMR Biomed. 2023;36(6):e4785. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0031] 31. Avalos LN, Luks TL, Gleason T, et al. Longitudinal MR spectroscopy to detect progression in patients with lower-grade glioma in the surveillance phase. Neurooncol Adv. 2022;4(1):vdac175. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0032] 32. Ruiz-Rodado V, Malta TM, Seki T, et al. Metabolic reprogramming associated with aggressiveness occurs in the G-CIMP-high molecular subtypes of IDH1mut lower grade gliomas. Neuro Oncol. 2020;22(4):480–492. [DOI] [PMC free article] [PubMed] [Google Scholar]

PERMALINK

Perfusion, diffusion, and anatomical MRI characteristics of pathologically confirmed malignant transformation in IDH-mutant gliomas

Nicholas S Cho

Viên Lam Le

Ashley Teraishi

Vicki Liu

Francesco Sanvito

Donatello Telesca

Masanori Nakajo

Chencai Wang

Sonoko Oshima

Blaine S C Eldred

Jingwen Yao

Phioanh L Nghiemphu

Noriko Salamon

Timothy F Cloughesy

Albert Lai

Benjamin M Ellingson

Abstract

Background

Methods

Results

Conclusions

Key Points.

Importance of the Study.

Materials and Methods

Patient Cohort

Table 1.

Figure 1.

Image Acquisition and Preprocessing

Tumor Imaging Analysis

Statistical Analysis

Results

Contrast-Enhancement, Volumetric, and Shape Characteristics

Figure 2.

Figure 3.

Diffusion and Perfusion Characteristics

Figure 4.

Figure 5.

Discussion

Conclusions

Supplementary material

Contributor Information

Funding

Author contributions

Data availability

References

Associated Data

Supplementary Materials

Data Availability Statement

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases