The ketogenic diet for patients with brain tumours: Two parallel randomised trials

Abstract

The ketogenic diet (KD) has attracted considerable interest from the brain tumour community. There is evidence for synergy with radiotherapy in animal models, and improved seizure control in children with epilepsy due to structural abnormalities. However, there is little data on mechanism or effect in brain tumour patients, and different options for implementing the ketogenic diet.
In November 2016 we assembled multiple stakeholders, including the James Lind PSP in neuro-oncology and several charities to develop a trial outline. We addressed mechanistic, clinical and practical factors to develop a trial protocol.
Based on the meeting, we are planning two parallel open-label phase 2 randomised trials of the Modified Ketogenic Diet (MKD) in patients with high grade gliomas (HGG) receiving chemoradiotherapy, and in patients with low- grade gliomas (LGG). Patients will be enrolled following a histological diagnosis and informed consent. They will receive training on a Modified Ketogenic Diet (MKD) and dietetic support throughout the trial.
Patients will be aged 18+, performance score <=2, with adequate baseline blood parameters and will not have received previous treatment (except surgery). Patients will receive standard clinical care as usual. The primary endpoint will be overall survival (not powered for comparison) (HGG) and symptom levels (LGG), with secondary endpoints including the degree and length of ketosis, disease progression and quality of life measures, including seizures. Translational assessments will include blood ketone (BHB) levels, changes in specific miRNA levels in plasma, tumour sample analysis and MRI/ MRS-based assessment of ketosis and tumour response. There is an embedded psychological component to explore patient experience, explore reasons for participation and the psychological impact of the trial.
Given the nature of the trial, we are intending to open the trial in 3 - 5 centres to concentrate dietetic support.

Full text

Abstracts
i7
NEURO-ONCOLOGY • JANUARY 2018
CLINICAL FEATURES, MANAGEMENT AND OUTCOMES
OF PAEDIATRIC CENTRAL NERVOUS SYSTEM TUMOURS
DISSEMINATED AND NONDISSEMINATED AT PRESENTATION: A
FOURTEEN-YEAR EXPERIENCE
Mr MatthewKirkman, Mrs KimPhipps, Mr KristianAquilina; Great
Ormond Street Hospital
BACKGROUND: It is unusual for paediatric central nervous system (CNS)
tumours to be disseminated at presentation, and literature on the clinical
features, management and outcomes of this specic group is scarce. Surgi-
cal management decisions can be challenging, particularly in the presence
of hydrocephalus. We present our experience in managing paediatric CNS
tumours disseminated and nondisseminated at presentation over fourteen
years. METHODS: Retrospective review of prospectively-collected data on
children presenting to our tertiary centre between 2003 and 2016 inclusive.
RESULTS: We identied 53 patients with CNS tumour dissemination at
presentation (M:F=34:19, median age=3.8years, range=7 days-15.6 years)
and 308 without dissemination (M:F=161:147, median age=5.8 years,
range=1day-16.9years). Five tumour groups were studied: medulloblastoma
(disseminated n=29/nondisseminated n=74), other primitive neuroectodermal
tumour (n=8/n=17), atypical teratoid rhabdoid tumour (n=8/n=22), pilocytic
astrocytoma (n=6/n=138), and ependymoma (n=2/n=57). Median follow-up
in survivors was not signicantly different between those with (64.0months,
range=5.2–152.0months) and those without disease dissemination at pres-
entation (74.5 months, range=4.7–170.1 months; P>0.05). Dissemination
status at presentation signicantly impacted survival, risk of recurrence, rates
of complete resection of the operated lesion, chance of proceeding to pallia-
tive care as an early management strategy, surgical complication rate, and
risk of requiring CSF diversion (all Ps<0.05). Differences between the ve
tumour groups were evident. No factors to predict the need for permanent
CSF diversion following temporary external ventricular drainage were identi-
ed on multivariate analysis, and there was no clear superiority of either ven-
triculoperitoneal shunt or endoscopic third ventriculostomy as a permanent
CSF diversion procedure. CONCLUSIONS: Tumour type and dissemination
status at initial presentation signicantly affect outcomes across a range of
measures. The management of hydrocephalus in patients with CNS tumours
is challenging, and further prospective studies are required to identify the
optimal CSF diversion strategy or strategies in this population.
DISCONNECTION VS EXCISION? ATEN YEAR REVIEW OF
HYPOTHALAMIC HAMARTOMAS
Dr RibhavPasricha, Dr Saquib AzadSiddiqui, Prof AshishSuri,
Dr SachinBorkar, Prof Ashok KMahapatra; All India Institute of Medical
Sciences, New Delhi
OBJECTIVE: Hypothalamic hamartomas (HH) are rare tumors of child-
hood. They are usually associated with gelastic seizures. The optimal man-
agement of these tumors lacks a consensus. We present our experience with
hypothalamic hamartomas over a ten year period. METHODS: This study
presents a retrospective review of 16 patients with HH’s treated between
2002 and 2012 at the All India Institute of Medical Sciences (AIIMS), New
Delhi, India, a tertiary care neurosurgical centre. RESULTS: There were 16
patients with an age ranging from 1.5 years to 20 years and included 9
males and 7 females. The most common symptom was seizures (81%; gelas-
tic seizures-62.5%) followed by precocious puberty (56.2%). The median
tumor volume was 5.9 cc (range 1.3 to 108 cc). Fourteen patients underwent
surgery while two were managed conservatively. Three patients received sec-
ondary Gamma Knife therapy. Agood seizure outcome (Engel classIand II)
post surgery was seen in 8 (50%) patients. The median follow up period was
39months (range 1 to 114months). The odds ratio for a better seizure out-
come was 2.5 times more in disconnection than for excision. The symptoms
of precocious puberty had resolved in 7 of the 9 affected patients (78%).
Transient diabetes insipidus was seen in one (6.2%) patient postoperatively
while hyperphagia was noted in four (25%) patients. Hyperphagia persisted
in all four patients until last follow up. CONCLUSIONS: HHs present with
gelastic seizures or precocious puberty. Disconnection of the HH is more
effective and a safer procedure vis-a-vis excision for controlling seizures.
PROBING GLIOBLASTOMA INFILTRATION INTO HEALTHY
TISSUE BY MAGNETIC RESONANCE PERFUSION IMAGING: A
QUANTITATIVE MRI EVALUATION
Dr AntoineVallatos, Dr JoannaBirch,
Mr HaithamAl-Mubarak, Mrs LindsayGallagher,
Dr LesleyGilmour, Dr WilliamHolmes, Prof AnthonyChalmers; Glasgow
experimental MRI centre, Institute of Neuroscience and Psychology,
University of Glasgow
The ability of tumour cells to invade healthy brain tissue is a major obsta-
cle to successful treatment of Glioblastoma (GBM). This makes complete
removal of the tumour by surgery impossible, leading to high recurrence
rates, and reduces the accuracy of target volume delineation for radiother-
apy planning. Providing with a wide range of imaging modalities, Magnetic
Resonance Imaging (MRI) is an important tool for GBM diagnosis and
characterisation. However, conventional MRI techniques fail to detect
regions of low tumour cell density that may be responsible for subsequent
tumour recurrence, while novel techniques are often not reliable due to a lack
of robust evaluation protocols. Invading tumour cells often progress along
blood vessels and recent results indicate that even individual cells can disrupt
the normal function of the blood brain barrier (BBB), providing an oppor-
tunity to detect tumour invasion at its earliest stages. Our research focuses
on the development of perfusion imaging techniques, and their quantita-
tive assessment in detecting low tumour inltration regions on mouse GBM
models presenting highly invasive tumour margins. Ahigh SNR multiple
adiabatic boli Arterial Spin Labelling technique (mbASL) was implemented
and optimised for rodent brain perfusion imaging and its ability to probe
tumour invasion was evaluated by comparison with standard MRI tech-
niques (T2 and diffusion weighted imaging) and immunohistochemistry sec-
tions. To achieve a quantitative MRI evaluation, multiple histological slices
(HLA stain for human GBM) were cut in the MRI plane, registered and
stacked to account for MRI slice thickness. This approach, leading to the
production tumour density maps in the MRI plane, allows going beyond the
standard evaluation tests to consider a voxel-to-voxel comparison between
MRI and histology. Our results conrm the limitations of standard imaging
modalities in probing low tumour densities and demonstrate the existence of
a quantitative relation between glioblastoma invasion and brain perfusion.
REGULATION OF DNA DOUBLE STRAND BREAK REPAIR BY
EGF AND VEGF SIGNALLING REVEALS AKT TO BE A CRITICAL
THERAPEUTIC TARGET IN GLIOBLASTOMA
Dr NatividadGomez-Roman, Mr Ming YaoChong,
Mrs KatrinaStevenson, Dr LesleyGilmour, Prof AnthonyChalmers;
University of Glasgow
Glioblastoma (GBM) is currently incurable. Its radioresistance has been
attributed to a subpopulation of cells termed ‘GBM stem-like cells’ char-
acterised by multipotentiality and tumorigenicity. The discrepancy between
pre-clinical and clinical effects of molecular targeted agents on radiosensitiv-
ity indicates that 2D in vitro models of GBM do not recapitulate the clini-
cal scenario. In a 3D model developed in our laboratory, EGFR inhibitors
failed to enhance radiosensitivity recapitulating their lack of efcacy in the
clinic, contrasting with their radiosensitising activity in 2D cultures. Con-
versely, inhibition of VEGF signalling caused signicant radiosensitisation
of 3D cultures but had no effect in 2D conditions. The critical role of the
DNA damage response in mediating these effects is illustrated by the con-
sistent correlation between radiosensitivity, unrepaired double-strand breaks
(γH2AX foci), mitotic catastrophe and micronuclei in both 2D and 3D
models. Further investigation revealed unrepaired DSB to be associated with
delayed resolution of phosphorylated DNA-PKcs nuclear foci and reduced
formation of Rad51 foci. Hence in 2D conditions, EGFR signalling appeared
to promote efcient non-homologous end-joining (NHEJ) repair, while in
3D conditions this process was dependent on VEGF signalling. Motivated
by previous reports that radiation induced EGFR signalling promotes NHEJ
via Akt mediated phosphorylation of DNA-PKcs, we investigated the role
of Akt signalling in 2D and 3D systems. Specic inhibition of Akt using the
small molecule inhibitor MK-2206 or Akti knockdown increased radiation
sensitivity in both 2D and 3D models to a similar extent as EGFR or VEGF
inhibition respectively. In keeping with this, phosphorylation of Akt was
EGFR dependent in 2D GSC cultures but VEGF dependent in the 3D model.
MK-2206 induced radiosensitivity was correlated with increased unrepaired
DSBs and extended mouse survival in an U87MGLuc2 orthotopic model
compared to radiation only. Our data identify Akt as a promising therapeu-
tic target in combination with radiation for GBM.
THE KETOGENIC DIET FOR PATIENTS WITH BRAIN TUMOURS:
TWO PARALLEL RANDOMISED TRIALS
Dr MattWilliams, Prof HelenCross,
Mr Michael DJenkinson, Ms KirstyMartin, Mrs SusanWood,
Prof Adrienne C.Scheck, Dr NeloferSyed, Dr KevinO’Neill,
Ms KatieSheen, Ms CatherineZabilowicz, Dr KieranBreen,
Dr KathyOliver, Ms EmmaWilliams, Prof MikeJohnson,
Ms WendyFulcher; Imperial College Healthcare Trust
The ketogenic diet (KD) has attracted considerable interest from the brain
tumour community. There is evidence for synergy with radiotherapy in animal
models, and improved seizure control in children with epilepsy due to structural
abnormalities. However, there is little data on mechanism or effect in brain
tumour patients, and different options for implementing the ketogenicdiet.
In November 2016 we assembled multiple stakeholders, including the
James Lind PSP in neuro-oncology and several charities to develop a trial
outline. We addressed mechanistic, clinical and practical factors to develop
a trial protocol.
Based on the meeting, we are planning two parallel open-label phase 2
randomised trials of the Modied Ketogenic Diet (MKD) in patients with
high grade gliomas (HGG) receiving chemoradiotherapy, and in patients
Downloaded from https://academic.oup.com/neuro-oncology/article/20/suppl_1/i7/4831293 by guest on 02 November 2020

Abstracts
i8 NEURO-ONCOLOGY • JANUARY 2018
with low- grade gliomas (LGG). Patients will be enrolled following a his-
tological diagnosis and informed consent. They will receive training on a
Modied Ketogenic Diet (MKD) and dietetic support throughout thetrial.
Patients will be aged 18+, performance score <=2, with adequate base-
line blood parameters and will not have received previous treatment (except
surgery). Patients will receive standard clinical care as usual. The primary
endpoint will be overall survival (not powered for comparison) (HGG) and
symptom levels (LGG), with secondary endpoints including the degree and
length of ketosis, disease progression and quality of life measures, including
seizures. Translational assessments will include blood ketone (BHB) levels,
changes in specic miRNA levels in plasma, tumour sample analysis and
MRI/ MRS-based assessment of ketosis and tumour response. There is an
embedded psychological component to explore patient experience, explore
reasons for participation and the psychological impact of thetrial.
Given the nature of the trial, we are intending to open the trial in 3 - 5
centres to concentrate dietetic support.
FACTORS PREDICTING FAVOURABLE OUTCOME FOR RECOVERY
OF PERI-OPERATIVE NEUROLOGICAL DEFICITS IN CRANIO-
SPINAL INTRINSIC TUMOURS: ASINGLE CENTRE EXPERIENCE
Mr KishorChoudhari, Dr ShashankJoshi; Royal Hallamshire Hospitals,
Shefeld Teaching Hospitals, Shefeld
OBJECTIVES: To review demographic, pathological and surgical fac-
tors that minimise long-term neurological decits in cranio-spinal intrin-
sic tumours. DESIGN Retrospective analysis. SUBJECTS 33 patients with
cranio-spinal intrinsic tumours located in eloquent areas treated at a single
centre over a period of ve years. METHODS: From a cohort of 780 neuro-
oncology patients, 33 (24 cranial and9 spinal) patients harbouring brain
or spinal cord intrinsic tumours in eloquent areas were identied who had
signicant perioperative neurological decits but near-complete recovery
of the decits within three months of surgical intervention. In addition to
demographic data, technical factors related to surgery and tumour pathol-
ogy were analysed to determine which factors may inuence recovery of
decits. Surgical technical factors were studied for feasibility of gross mac-
roscopic resection. RESULTS: All patients who recovered from their de-
cits were less than 63years of age, all except three patients had pathology
ranging Gr I-III. In three patients, it wasGr IV. All except three had radical
resection of tumour as proven by contrastimaging in the immediate post-
operative period and at three months. Age, tumourgrade &amp; extent of
surgical resection appeared to be the factors inuencingrecovery of neuro-
logical decits. Younger age, lower grade of tumour and grosstotal excision
of the tumour at surgery appeared to be favourable factors forrecovery of
peri-operative decits. CONCLUSIONS: In a retrospectiveanalysis, younger
age, lower grade of tumour and radical surgical excisionappear to be factors
predicting favourable outcome for recovery ofperioperative neurological
decits in patients with cranio-spinal intrinsictumours.
IMPACT OF OPEN ACCESS CT (OACT) FOR HEADACHE
SUSPECTED OF BRAIN CANCER IN LOTHIAN
Dr KarolisZienius, Dr RobinGrant, Dr PaulBrennan; University of
Edinburgh, Western General Hospital, NHS Lothian
Guidance to General Practitioners on referral for direct-access brain imaging,
based on levels of risks for CNS malignancy (red/orange/yellow) are proposed.
However, the effect of such guidance upon referral behaviour is unknown.
We reviewed all Open-Access-Computed- Tomography (OACT) referrals
for possible CNS malignancy from Lothian-based GPs for 2010–2015. 3302
OACTs were performed. We report initial ndings on a total of 54 scans pos-
itive for brain tumour (rate 1.63%). 57% were females, mean age 62years.
57% patients had contrast-enhanced scan. There were 7 high-grade gliomas,
4 low-grade tumours, 16 metastases, 8 pituitary, 17 meningioma, 1 CNS
lymphoma, and 1 skull metastasis. Six patients whose initial report queried
a tumour had a normal follow-up imaging. All 6 patients had initial non-
contrast scanonly.
Headache was the commonest complaint for patients whose scan identi-
ed a brain tumour: 16 referrals for headache-alone, 12 for headache-plus-
neurological decit, 4 for headache-plus-cognitive decit and 22 for focal/
cognitive symptoms but no headache. Time to scan from a referral, reported
in median days, was: 19, 12.5, 18.5 and 17, respectively. It was shorter for
patients with a known history of cancer (median, 9)than without (18) (MW
u, p=0.06). Overall, pre-diagnostic symptomatic interval (median weeks)
was longest for patients with non-focal symptoms: 14.4, 7.3, 14.8 and 9.6,
respectively. Thirty-one (59.6%) referrals were for red-ag, 14 (26.9%) for
orange-ag, and 7 (13.5%) for yellow-ag symptoms.
We are now reviewing all OACTs and will report the frequencies of
normal and signicant abnormal other than brain tumour scans and will
perform correlation whether these ndings relate to a particular presenting
symptom. We will report the frequency of symptom groups belonging to a
suggested criteria of presumed risk of CNS malignancy (Kernick etal). We
will also calculate measures of diagnostic performance for each symptom
groups based on 3 levels of risk.
TUMOUR TREATING FIELDS (TTFIELDS) HAVE ANTI-
PROLIFERATIVE EFFECTS ON PAEDIATRIC BRAIN TUMOUR CELL
LINES AT CLINICALLY DELIVERABLE FIELD SETTINGS
Mr JoshuaBranter, Dr Mariade los Angeles Estevez-Cebrero,
Prof RichardGrundy, Dr SurajitBasu, Dr StuartSmith; The University of
Nottingham
INTRODUCTION: Approximately 400 children in the UK are diagnosed
with a CNS tumour each, which represents a quarter of childhood cancer
cases in the UK. The overall 5-year survival rate of peadiatric brain tumour
patients is roughly 75%, however survival still remains poor for some brain
tumours, particularly high grade gliomas. Pre-clinical studies have shown ef-
cacy of TTFields on high grade glioma cell lines, while TTFields have only
been approved for adult Glioblastoma multiforme (GBM) patients in combi-
nation with the Stupp treatment protocol. The lack of overlapping toxicities
associated with OptuneTM has made TTFields an attractive treatment strat-
egy for poor prognosis brain tumours. METHODS: The InovitroTM system is
the laboratory testing system used to develop the clinically approved Optune
system. InovitroTM was used to deliver TTFields over 72 hours at a range of
clinically relevant frequencies (100-400kHz) and intensities (1-3V/cm) to a
panel of paediatric GBM, Medulloblastoma and Ependymoma cell lines. The
effects of TTFields on cell viability and cell cycle was assessed using metabolic
viability tests (Presto blue), cell proliferation assays and cell cycle analysis
using ow cytometry. RESULTS: TTFields treatment reduces cell viability of
all paediatric brain tumour cell lines tested within the frequency titration. The
optimum frequencies were between 200 and 300kHz for all cell lines tested.
TTFields treatment had signicant effects on cell cycling relative to con-
trolcells. CONCLUSIONS: This preliminary study has conrmed the efcacy
of electric elds on paediatric brain tumour cell lines. The optimum frequen-
cies for the paediatric GBM cell lines did not uniformly fall into the 200kHz
range as previously reported for glioma cell lines. This report warrants further
study into the potential for TTFields therapy for paediatric patients.
AN OBSERVATIONAL STUDY OF NEUROCOGNITIVE FUNCTION
IN PATIENTS UNDERGOING STEREOTACTIC RADIOSURGERY AT
VELINDRE CANCER CENTRE
Dr JamesPowell, Dr Najmus Sahariqbal, Dr MaeveSmyth,
Dr AnneJohnson, Dr MichelleSmalley, Prof RichardWise,
Prof DerekJones, Dr OwenTilsley, Mr AndrewBryant,
Dr AnthonyMillin, Dr David GLewis, Dr JohnStaffurth; Velindre Cancer
Centre
Radiation treatment for brain metastases has traditionally involved whole
brain radiotherapy (WBRT). Technical developments now allow radiotherapy
to be targeted precisely to small areas of cancer within the brain using ste-
reotactic radiosurgery (SRS). Combining WBRT and SRS improves survival
compared with WBRT alone. WBRT has signicant side effects particularly
on short and long term memory and even with SRS treatment alone recent
studies have demonstrated that around half of patients suffer memory impair-
ment. Certain areas of the brain, such as the hippocampus, are considered
particularly sensitive to radiotherapy but the mechanism for memory impair-
ment is unknown. Hypotheses include changes in structure, cerebral blood
ow and brain connectivity. Equally, radiotherapy tolerance doses for the hip-
pocampus and other structures within the brain important for memory are
not well dened. We have obtained funding and ethical approval to conduct
a prospective observational study in collaboration with the Cardiff Univer-
sity Brain Research Imaging Centre (CUBRIC). In this study we aim to recruit
40 patients undergoing SRS treatment over 18months and we will correlate
data from neurocognitive function (NCF) tests with radiation dosimetry to
cognitive structures and with detailed MRI scans using latest MRI techniques.
Detailed NCF assessments, according to the RANO group recommendations,
will be conducted before and after SRS treatment. MRI brain scans using a
combination of functional structural, spectroscopic and microstructural MRI
scans will be performed before and after SRS. Radiotherapy doses received by
brain structures important for memory including the hippocampus, amygdala,
prefrontal cortex and limbic system will be calculated and correlated with NCF
and MRI assessments. We predict that by correlating this information we will
obtain important new insights into radiotherapy effects on memory and on the
radiotherapy tolerance doses for these crucial structures when delivering SRS.
ROLES FOR NEUROPSYCHOLOGY IN THE NEURO-ONCOLOGY MDT
Dr StuartAnderson; Brighton & Sussex University Hospitals NHS Trust
Cognitive impairment is detectable in the majority of patients presenting
with primary brain tumour and is a major cause of disability, often being
cited as the single greatest cause of burden to affected individuals and their
carers (Locke etal. 2008). In addition, psychological reactions to diagno-
sis and treatment can include acute distress, anxiety and depression. Col-
lectively, these symptoms may complicate treatment and inuence quality
of life (QOL). This presentation provides a rationale for the inclusion of
neuropsychologists in neuro-oncology multidisciplinary teams. It summa-
rises relevant outcomes of brain tumours and associated treatments and
Downloaded from https://academic.oup.com/neuro-oncology/article/20/suppl_1/i7/4831293 by guest on 02 November 2020