Ben Babourina-Brooks1,2, Lesley MacPherson2, Laurence J Abernethy3, Theodoros N Arvanitis2,4, Simon Bailey5, Nigel P Davies1,2,6, Daniel Rodriguez Gutierrez 7,8,9, Tim Jaspan 7,10, Dipayan Mitra11, Paul S Morgan7,9,10, Barry Pizer 12, Richard G. Grundy 7, Dorothee P Auer7,8,10, and Andrew C Peet1,2
1University of Birmingham, Birmingham, United Kingdom, 2Birmingham children's hospital, Birmingham, United Kingdom, 3Department of Radiology, Alder Hey Children's NHS Foundation Trust, Liverpool, United Kingdom, 4Institute of Digital Healthcare, WMG, University of Warwick, Coventry, United Kingdom, 5Paediatric Oncology Department, Great North Children’s Hospital, Newcastle upon Tyne, United Kingdom, 6Department of Imaging and Medical Physics, University Hospitals Birmingham NHS Foundation Trust, Birmingham, United Kingdom, 7The Children‘s Brain Tumour Research Centre, University of Nottingham, Nottingham, United Kingdom, 8Radiological Sciences, Department of Clinical Neuroscience, University of Nottingham, Nottingham, United Kingdom, 9Medical Physics, Nottingham University Hospital, Queen’s Medical Centre, Nottingham, United Kingdom, 10Neuroradiology, Nottingham University Hospital, Queen’s Medical Centre, Nottingham, United Kingdom, 11Neuroradiology Department, Newcastle upon Tyne Hospitals, Newcastle upon Tyne, United Kingdom, 12Department of Paediatric Oncology, Alder Hey Children's NHS Foundation Trust, Liverpool, United Kingdom
Synopsis
Brain tumours have a high
mortality rate and are the most common solid tumour of childhood. The non-invasive
imaging technique, MRS, measures tumour metabolites which can provide additional
prognostic information to aid in clinical management. MRS metabolites Glycine,
Scyllo-Inositol, NAA and Lipids have been associated with prognosis for
pediatric brain tumour patients in a single centre 1.5T study. This study aimed
to validate these MRS survival markers in a 3T multicentre setting. In this preliminary study Lipids were
validated as a survival marker across childhood brain tumours.
PURPOSE:
Brain
tumours have a high mortality rate and are the most common solid tumour of
childhood [1]. Identification of high risk patients may allow for better
treatment stratification and more effective disease management, increasing
survival and reducing treatment related damage. Magnetic Resonance Spectroscopy
(MRS) provides a non-invasive measure of brain tumour metabolism[2],
which has been previously used to identify diagnostic and prognostic metabolite
markers in adult and paediatric brain tumours[3].
Metabolites such as high lipids, scyllo-inositol (SI), and glycine (Gly) have
been shown to reflect poor survival, whilst N-Acetylaspartic acid (NAA) is a
marker of good outcome in paediatric brain tumours[4, 5, 6].
However, studies have focussed on 1.5T spectroscopy and are based on single
centre studies. A study at higher field strengths would increase spectral
resolution and SNR therefore distinguish more metabolites. To date, metabolites
measured by MRS have not been investigated as a potential survival marker in a
multi-centre study for paediatric brain tumours. The aim of this preliminary study
were to investigate if the established 1.5T single centre survival markers hold
true in a 3T multi-centre setting. METHODS:
Ninety one
children with newly diagnosed brain tumours were examined with MRS on a 3T
system prior to treatment from four hospitals between 2012-2016, with follow up
until September 2017. Multicentre ethical and parental approval consent was
obtained. Clinical data was obtained including gender, age at diagnosis, and
dates of death. Tumour diagnosis was confirmed by a multidisciplinary team using
histopathology, clinical and radiological information. Most patients were treated
with surgical resection, followed by fractionated radiotherapy and chemotherapy
with the protocol based on diagnosis and clinical risk factors.
MR
imaging and MRS were acquired pre-treatment using single voxel spectroscopy
PRESS from different scanner systems (Phillips Acheiva, Siemens Verio) using
the following parameters TR =2000ms and TE=30-46ms. The MRS voxel was placed
within the solid component of the primary tumour. Unprocessed data was pushed
from the scanner to a local database, anonymised and uploaded, by individual
centres, to the CCLG database for retrospective analysis. Unprocessed MRS
signals were analyzed using TARQUIN[7] software (version 4.3.6). Spectra were excluded
based on displaced voxel location, unstable spectral baseline, artefacts present,
signal to noise ratio (SNR) < 4,
overall metabolite linewidths exceeded 0.15ppm and water linewidth
>10Hz.
Survival
analysis was performed on individual metabolites using univariate Cox
regression and Kaplan-Meier analysis based on median metabolite concentration cut-off
values. For significant metabolite markers of prognosis and the 1.5T
established markers of survival (NAA, Gly, Lipids and SI) the cut off was
optimised iteratively based on p value. Statistical
tests were performed in the R software package using the survival library with
P<0.05 chosen for significance.
RESULTS and DISCUSSION:
The final cohort
consisted of 71 patients after quality control, with a median follow up of
1182±583 days. All 6 patient deaths were a direct result of their disease. The
mean age of the patient cohort was 7.45±4.5 years and 67% were male, neither
were found to influence survival.
Lipids at 0.9ppm and
total lipids and macromolecules (TLM) at 1.3ppm, were found to be biomarkers of
poor prognosis (p<0.01), Figure 1, Table 1. In both cases the optimised cut
off identified all the patient deaths in the cohort, providing high hazard
ratios (Table 1). Given the limited events in this cohort this result indicates
that lipids are a strong marker of poor prognosis. NAA, SI and Gly, have been
previously reported as biomarkers [4,6]. However, in this preliminary study
they were not found to be significant survival markers, which could be due to
the limited number of events in this cohort.
Mean
spectral profiles were created for the patients that survived and died during
the follow up period (Figure 2). There is a clear difference in the peak height
of lipids at 1.3ppm. Patients that died have a lower NAA peak height even
though it was not found to be a significant survival marker. Lipids are well
documented as being associated with poor prognosis in both paediatric and adult
brain tumour patients being associated with apoptosis, necrosis and cytoplasmic
lipid droplets [8-11]. This is the first
study in paediatric brain tumours to validate lipids as survival markers using
MRS in a multicentre setting. Further work includes building up this cohort to
assess if with more numbers and events the other metabolites of survival can be
validated.
Conclusion:
Lipids have been found to be a predictor
of survival for childhood brain tumours in a multicentre setting, adding
confidence to using MRS for predicting prognosis in different centres.
Acknowledgements
We would like to thank Dr Paul Davies for
advice on the statistical methodology and analysis. References
[1] Bleyer W et al. 8
drugs in 1 day chemotherapy for brain tumors: a new approach and rationale for
preradiation chemotherapy. Med Pediat Oncol: 1983 11: 213
[2] Peet AC, Arvanitis TN, Auer DP, Davies
NP, Hargrave D, and F. Howe, The value of
magnetic resonance spectroscopy in tumour imaging. Arch Dis Child, 2008. 93: p. 725-727.
[3] Verma A, Kumar I, Verma N, Aggarwal P,
and R. Ojha, Magnetic resonance
spectroscopy — Revisiting the biochemical and molecular milieu of brain
tumours. BBA Clinical, 2016. 5:
p. 170-178.
[4] Wilson M, Cummins CL, and MacPherson
LM et al, Magnetic resonance spectroscopy
metabolite profiles predict survival in paediatric brain tumours. Eur. J.
Cancer, 2013. 49: p. 457-464.
[5] Wilson M, Gill SK, Macpherson L,
English M, Arvanitis TN, and A. Peet, Noninvasive
detection of glutamate predicts survival in paediatric medulloblastoma.
Clin Cancer Res, 2014. 20(17): p.
4532-4539.
[6] B.Babourina-Brooks,
S.Kohe, Gill SK, Wilson M, MacPherson L, N.P.Davies, A.C.Peet. Glycine a non-invasive marker of survival in
paediatric brain tumours. International
Society for Magnetic Resonance in Medicine (ISMRM), Honolulu,USA, 2017.
Abstract1102
[7] Wilson M. et al. A constrained
least-squares approach to the automated quantitation of in vivo ¹H magnetic
resonance spectroscopy data. Mag Reson Med 2011;65:1-12
[8] Murphy PS, Rowland IJ, Viviers L, Brada
M, Leach MO, and Dzik-Juras AS, Could assessment
of glioma methylene lipid resonance by in vivo (1)H-MRS be of clinical value?
Br J Radiol, 2003. 76(907): p.
459-463.
[9] Negendank W, Li CW, Padavic-Shaller K,
Murphy-Boesch J, and Brown TR, Phospholipid
metabolies in 1H-decoupled 31P MRS in vivo in human cancer: implications for
experimental models and clinical studies. Anticancer Res, 1996. 16(3B): p. 1539-1544.
[10] Kaminogo M, Ishimaru H, and Morikawa M
et al, Diagnostic potential of short echo
time MR spectroscopy of gliomas with single-voxel and point-resolved spatially
localised proton spectroscopy of brain. Neuroradiology, 2001. 43(5): p. 353-363.
[11] Opstad KS, Bell BA, Griffiths JR, and
Howe FA, An investigation of human brain
tumour lipids by high-resolution magic angle spinning 1H MRS and histological
analysis. NMR Biomed, 2008. 21(7):
p. 677-685.