Lejla Alic1, Sanneke C Willekens1,2, Henk-Jan M.M. Mutsaerts2, Jan Petr3, Netteke A.Y.N. Schouten-van Meeteren4,5, Maarten M.H. Lequin6, and Evita E.C. Wiegers2
1Magnetic Detection & Imaging Group, University of Twente, Enschede, Netherlands, 2Department of Radiology, University Medical Center Utrecht, Utrecht, Netherlands, 3Helmholtz-Zentrum Dresden-Rossendorf, Institute for Radiopharmaceutical Cancer Research, Dresden, Germany, 4Radiology and Nuclear Medicine, Amsterdam Neuroscience, Amsterdam, Netherlands, 5Department of Neuro-oncology, Princess Máxima Centre, Utrecht, Netherlands, 6Radiology, University Medical Center Utrecht, Utrecht, Netherlands
Synopsis
ASL-MRI is reported as an option to
assess potentially heterogeneous physiological processes important for tumour
treatment. Therefore, we explored the heterogeneity in normalised CBF as an
imaging biomarker for assessment of treatment effect in pLGG. There is a
noticeable effect of chemotherapy observed as a change in texture of healthy
appearing brain tissue. A high difference in texture between treated and non-treated
patients for non-enhancing tumour part is observed, suggesting that texture,
based on co-occurrence matrices, is suitable as an imaging biomarker for
assessment of treatment effect in pLGG.
Introduction
Paediatric low-grade gliomas (pLGG)
account for 30-50% of all paediatric brain tumours1.
The presentation of pLGG varies between symptomless, motor impairment, visual
impairment, endocrine deficiencies, and hypothalamic failure. Consequently,
primary treatment for pLGG consist of surgical resection and/or chemotherapy depending
on these clinical symptoms and tumour location. Progression of residual disease
can require multiple phases of chemotherapy over the years2. Despite 10-year overall survival of 85-96%, management of pLGG
remains challenging3 and is considered a
chronic disorder in over 50% of the cases4. Due to the underlying condition and/or treatment, the patients
often suffer from functional, visual,
and neurological impairment5. Arterial spin labelling (ASL) perfusion MRI assesses cerebral perfusion and is
reported as an option to quantify underlying potentially heterogeneous
physiological processes important in therapeutic decision making6. This study explores the heterogeneity in ASL-MRI as an imaging
biomarker for assessment of treatment effect in pLGG.Methods
17 Paediatric patients with
confirmed pLGG (at the following brain locations: 5 posterior fossa, 1 lateral
ventricles, 8 chiasmatic, 1 suprasellar, 1 spinal cord, 1 temporal hemisphere)
referred to the Princess Máxima Centre for Paediatric Oncology were included in
this study: 5 untreated
patients (2f/3m, age=9.4±5.0year) and 12 patients (6f/6m, age=7.3±3.4year)
treated by chemotherapy. For untreated patients, MRI was acquired at baseline
and for treated patients at follow-up. MRI was acquired at 1.5T or 3T Philips MRI
system using a 32-channel receive head coil. Imaging protocol included a T1-weighted (T1w), T1w
after gadolinium injection (T1w-c), T2-weighted (T2w), FLAIR, and
pseudo-continuous ASL (PCASL). The PCASL-MRI was acquired as a 2D EPI sequence
with background suppression, label duration of 1800 ms, and an initial post-label delay
of 1800ms or 1525ms at 1.5T or 3T respectively.
Three-step processing framework of
the structural PCASL images is illustrated by flow chart in Figure 2.
Pre-processing of structural MRI involved registration to T1w images, skull-stripping
by SPM127, and segmentation of brain structural data of normal-appearing grey matter
(NAGM) and normal-appearing white matter (NAWM). Subsequently, co-registered
skull-stripped structural MR images were used to segment tumour by utilising
HD-GLIO8,9 that resulted in individual tumour components: contrast enhancing (CE)
and non-contrast enhancing (NCE), cyst, and necrosis.
The quality of image registration and automatic
tumour segmentation was assessed in a group of ten patients randomly selected
from our LGG cohort. Registration was evaluated quantitatively using manual
annotations of anatomical landmarks (e.g., AC and PC) and is assessed as the
root mean square error (RMSE) between individual observers. Automatic tumour
segmentation was compared with a manual delineation by a radiologist and quantified
by DICE coefficient9. PCASL-MRI was processed by ExploreASL10 into
quantitative CBF and normalised by average CBF in the HAGM, producing
normalised CBF (nCBF). Treatment effect was assessed by three first order
statistical (FOS) features (mean, median, standard deviation) and 22
co-occurrence texture features11-13 with
a matrix size of 26 averaged over four directions. The resulting features were
averaged over all slices containing the tumour and were presented separately for
CE, NCE, and NAWM. A total of 25 features per ROI was ranked independently using
a feature selection algorithm maximising the area between the empirical ROC
curve and the classifier slope. The Wilcoxon signed-rank test was used to
assess the differences between the treated and untreated patients.Results
The checkerboard view in Figure 2A
illustrates an example of registration results at the ventricles level and at
the tumour level. The averaged RMSE for the two anatomical landmarks is similar
for all three sequences (Figure 2B). All tumour masks produced by HD-GLIO were
manually post-processed. For two tumours, HD-GLIO produced a false negative
segmentation for both CE and NCE tumour masks. The remaining eight tumours had
an average DICE score of 0.76 for the CE tumour mask and 0.34 for the NCE
tumour masks. Figure 3 illustrates the final masks (CE, NCE, NAGM, and NAWM)
and nCBF-map.
Considering
differentiation between treated and untreated patients, none of the FOS
features demonstrate significance while a number of textual features were found
significant. The most potent texture feature (i.e. cluster shade) assessing the
differences between treated and untreated patients showed an average change of
109% for NCE tumour mask and 18 % change in NAWM.Discussion & Conclusions
Quantitative validation of the
registration gives a lower RMSE than the size of the AC and the PC suggesting a
good registration accuracy in these paediatric patients. There is a noticeable
effect of chemotherapy observed as a change in texture of NAWM. This is in line
with previous studies showing that chemotherapy is associated with damage of
normal-tissue in the brain14. The significantly higher difference in
texture between treated and untreated patients for NCE suggests that texture analysis,
based on co-occurrence matrices, are suitable to quantify the change in
heterogeneity and therefore could potentially serve as an imaging biomarker for
assessment of treatment effect in pLGG. These conclusions are also supported by previous findings
in paediatric patients after radiotherapy15. Acknowledgements
No acknowledgement found.References
- Ryall S, Tabori U, Hawkins C. Pediatric
low-grade glioma in the era of molecular diagnostics. Acta Neuropathol Commun.
2020;8(1):30.
- Lassaletta A, Scheinemann K, Zelcer SM,
Hukin J, Wilson BA, Jabado N, et al. Phase II Weekly Vinblastine for
Chemotherapy-Naive Children With Progressive Low-Grade Glioma: A Canadian
Pediatric Brain Tumor Consortium Study. J Clin Oncol. 2016;34(29):3537-43.
- de Blank P, Bandopadhayay P, Haas-Kogan D,
Fouladi M, Fangusaro J. Management of pediatric low-grade glioma. Curr Opin
Pediatr. 2019;31(1):21-7.
- Fangusaro J, Witt O, Hernaiz Driever P,
Bag AK, de Blank P, Kadom N, et al. Response assessment in paediatric low-grade
glioma: recommendations from the Response Assessment in Pediatric
Neuro-Oncology (RAPNO) working group. Lancet Oncol. 2020;21(6):e305-e16.
-
de Lande RSV, Maurice-Stam H, Marchal JP,
Vuurden DGV, Vandertop WP, Grootenhuis MA, et al. Adaptive behavior impaired in
children with low-grade glioma. Pediatr Blood Cancer. 2019;66(1):e27419.
- Carlsson J, Ren ZP, Wester K, Sundberg AL,
Heldin NE, Hesselager G, et al. Planning for intracavitary anti-EGFR
radionuclide therapy of gliomas. Literature review and data on EGFR expression.
J Neurooncol. 2006;77(1):33-45.
- Friston KJ. Experimental design and
Statistical Parametric Mapping Queen Square, London, UK WC1N 3BG University
College London [Available from: https://www.fil.ion.ucl.ac.uk/spm/doc/intro/intro.pdf.
- Kickingereder P, Isensee F, Tursunova I,
Petersen J, Neuberger U, Bonekamp D, et al. Automated quantitative tumour
response assessment of MRI in neuro-oncology with artificial neural networks: a
multicentre, retrospective study. Lancet Oncol. 2019;20(5):728-40.
- Taha AA, Hanbury A. Metrics for evaluating
3D medical image segmentation: analysis, selection, and tool. BMC Med Imaging.
2015;15:29.
- Mutsaerts H, Petr J, Groot P, Vandemaele P,
Ingala S, Robertson AD, et al. ExploreASL: An image processing pipeline for
multi-center ASL perfusion MRI studies. Neuroimage. 2020;219:117031.
- Haralick RM, Shanmugam K, Dinstein I.
Textural Features for Image Classification. Ieee T Syst Man Cyb.
1973;Smc3(6):610-21.
- Alic L, van Vliet M, van Dijke CF, Eggermont
AM, Veenland JF, Niessen WJ. Heterogeneity in DCE-MRI parametric maps: a
biomarker for treatment response? Phys Med Biol. 2011;56(6):1601-16.
- Alic L, Veenland JF, van Vliet M, van Dijke
CF, Eggermont AM, Niessen WJ. Quantification of heterogeneity in dynamic
contrast enhanced MRI data for tumor treatment assessment. 3rd IEEE International Symposium on
Biomedical Imaging: Nano to Macro; 6-9 April 2006; Arlington, VA, USA2006.
- Ahles TA, Root JC, Ryan EL. Cancer- and
cancer treatment-associated cognitive change: an update on the state of the
science. J Clin Oncol. 2012;30(30):3675-86.
- Li MD, Forkert ND, Kundu P, Ambler C, Lober
RM, Burns TC, et al. Brain Perfusion and Diffusion Abnormalities in Children
Treated for Posterior Fossa Brain Tumors. J Pediatr. 2017;185:173-80 e3.