Available via license: CC BY-NC-ND 4.0
Content may be subject to copyright.
Original Study
[
18
F]fluorothymidine and [
18
F]fluorodeoxyglucose PET Imaging
Demonstrates Uptake and Differentiates Growth in
Neurofibromatosis 2 Related Vestibular Schwannoma
yJose M. Anton-Rodriguez, zDaniel Lewis, §Ibrahim Djoukhadar, jjDavid Russell,
yPeter Julyan, zDavid Coope, zAndrew T. King, zSimon K. L. Lloyd,
{D. Gareth Evans, #Alan Jackson, and #Julian C. Matthews
Division of Informatics, Imaging and Data Sciences, MAHSC, University of Manchester;
y
Christie Medical Physics and Engineering,
The Christie NHS Foundation Trust, Manchester;
z
Manchester Skull Base Unit, Manchester Centre for Clinical Neurosciences,
Manchester Academic Health Science Centre, Salford Royal NHS Foundation Trust; §Department of Neuroradiology, Manchester
Centre for Clinical Neurosciences, Salford Royal NHS Foundation Trust, Manchester Academic Health Science Centre;
jj
Department
of Radiology, Manchester University NHS Foundation Trust;
{
Manchester Centre for Genomic Medicine, St Mary’s Hospital,
Manchester University Hospitals National Health Service Foundation Trust and Manchester Academic Health Science Centre,
Manchester, UK; and
#
Division of Neuroscience and Experimental Psychology, School of Biological Sciences, Faculty of Biology,
Medicine and Health, Manchester Academic Health Science Centre (MAHSC), University of Manchester
Objective: To investigate whether [
18
F]fluorothymidine
(FLT) and/or [
18
F]fluorodeoxyglucose (FDG) positron emis-
sion tomography (PET) can differentiate growth in neurofi-
bromatosis 2 (NF2) related vestibular schwannomas (VS)
and to evaluate the importance of PET scanner spatial
resolution on measured tumor uptake.
Methods: Six NF2 patients with 11 VS (4 rapidly growing,
7 indolent), were scanned with FLT and FDG using a high-
resolution research tomograph (HRRT, Siemens) and a
Siemens Biograph TrueV PET-CT, with and without resolu-
tion modeling image reconstruction. Mean, maximum, and
peak standardised uptake values (SUV) for each tumor were
derived and the intertumor correlation between FDG and
FLT uptake was compared. The ability of FDG and FLT
SUV values to discriminate between rapidly growing and
slow growing (indolent) tumors was assessed using receiver
operator characteristic (ROC) analysis.
Results: Tumor uptake was seen with both tracers, using
both scanners, with and without resolution modeling. FDG
and FLT uptake was correlated (R
2
¼0.67–0.86, p<0.01)
and rapidly growing tumors displayed significantly higher
uptake (SUV
mean
and SUV
peak
) of both tracers ( p<0.05,
one tailed ttest). All of the PET analyses performed
demonstrated better discriminatory power (AUC
ROC
range ¼0.71–0.86) than tumor size alone (AUC
ROC
¼0.61).
The use of standard resolution scanner with standard
reconstruction did not result in a notable deterioration of
discrimination accuracy.
Conclusion: NF2 related VS demonstrate uptake of both
FLT and FDG, which is significantly increased in rapidly
growing tumors. A short static FDG PET scan with standard
clinical resolution and reconstruction can provide relevant
information on tumor growth to aid clinical decision
making. Key Words: [
18
F]fluorodeoxyglucose (FDG)—
[
18
F]fluorothymidine (FLT)—Neurofibromatosis 2—PET—
Vestibular schwannoma.
Otol Neurotol 40:826–835, 2019.
Address correspondence and reprint requests to Jose M. Anton-Rodriguez,
Ph.D., Wolfson Molecular Imaging Centre, 27 Palatine Road, Withington,
Manchester, M20 3LJ, UK; E-mail: jose.anton@manchester.ac.uk
This work is supported by Cancer Research UK and the Engineering
and Physical Sciences Research Council Cancer Imaging Centre in
Cambridge and Manchester (C8742/A18097).
J.M.A.-R. was undertaking a PhD supported by The Christie NHS
Foundation Trust and The University of Manchester. D.G.E. is an NIHR
Senior Investigator supported by the Biomedical Research Centre, Man-
chester (NIHR programme IS-BRC-1215-20007).
The authors disclose no conflicts of interest.
Supplemental digital content is available in the text.
This is an open access article distributed under the terms of the
Creative Commons Attribution-Non Commercial-No Derivatives
License 4.0 (CCBY-NC-ND), where it is permissible to download
and share the work provided it is properly cited. The work cannot be
changed in any way or used commercially without permission from the
journal.
DOI: 10.1097/MAO.0000000000002272
Copyright ß2019 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of Otology & Neurotology, Inc.
Neurofibromatosis 2 (NF2) is a dominantly inherited
tumor predisposition syndrome affecting approximately
1 in 33,000 live births (1). Thehallmark of this condition is
the development of bilateral vestibular schwannomas (VS)
(2,3) and once diagnosed patients typically undergo annual
magnetic resonance imaging (MRI) screening to evaluate
the size and growth with a cohort of tumors displaying
relatively rapid growth (3,4).The cornerstone of modern
NF2 management is conservation of hearing function and
quality of life (3,5). While surgery plays a role in the
management of rapidly growing tumors, the decision to
operate depends on multiple factors including hearing
deterioration rate, tumor growth rate, and tumor size
(3). Surgery carries significant risks such as facial nerve
injury (3), but early surgery, before tumors become too
large, reduces the complication risk, and improves the
outcome of adjunctive hearing preservation techniques
such as auditory brainstem implantation (3,6– 9). There
is therefore a clinical need to identify rapidly growing
VS early but with current MRI screening regimens
there is a danger of missing significant growth due to
the time-interval between scans. Furthermore, the accu-
racy and interobserver reproducibility of tumor measure-
ments varies considerably depending on the measurement
method used (10,11) and there is considerable debate
within the literature as to what constitutes significant
growth within lesions (12). An imaging biomarker that
allows earlier identification of rapidly growing tumors
would therefore be of clinical utility, particularly to
patients harboring tumors that are approaching the thresh-
old size for increased surgical risk. In these cases detecting
further growth through serial MRI may mean that the
optimum window for management has been missed
whereas predicting growth offers the best opportunity to
maximize surgical outcomes, patient quality of life, and
avoid resulting costly treatment (13).
The positron emission tomography (PET) tracers fluo-
rine-18 labeled deoxy-2-D-glucose (FDG) and 3’-deoxy-
3’-fluorothymidine (FLT) have been increasingly used in
oncology as imaging biomarkers of cellular metabolism
and cellular proliferation respectively. Tumor cells pref-
erentially accumulate FDG due to increased expression
of glucose membrane transporters and the enzyme hexo-
kinase, alongside a tendency to favor the more inefficient
anaerobic pathway resulting in greater metabolic demand
(Warburg effect) (14–18). FLT is transported into cells
by the same nucleoside transporters as thymidine, and
undergoes intracellular phosphorylation through the
enzyme thymidine kinase 1. Elevation of thymidine
kinase 1 occurs in rapidly dividing cells and consequently
FLT uptake is a marker of cellular proliferation rate (19).
Whereas FDG use within the central nervous system has
been limited due to constitutively high uptake within the
normal brain (20,21), brain uptake of FLT is normally
limited by the blood–brain barrier (22,23), but has been
demonstrated in regions of blood–brain barrier disrup-
tion such as within intrinsic glioma (24,25).
PET imaging in VS can be challenging and previous
FDG PET studies in non-NF2 patients with sporadic
tumors have shown inconclusive results due to low uptake
within the tumor compared with the adjacent cerebellum
(20,21). Similarly inconclusive results have been reported
when using other PET tracers relevant to central nervous
system tumors such as [
11
C]methionine (20). FLT or FDG-
PET has not, however, been previously described in NF2
patients, and there is growing evidence that sporadic and
NF2 related tumors are biologically different both at the
macroscopic level (26), but also with regard to their
cellular proliferation indices (27,28).
The rationale of this pilot study was to investigate
whether PET with FLT and/or FDG in combination with
MR could be used in the future to assist in refining clinical
decision making in NF2 related VS. The objective of this
study was to therefore first assess if VS in a cohort of NF2
patients have measurable FLT and/or FDG uptake, and
second to determine if rapidly growing tumors displayed
differences in the uptake of these PET tracers compared
with more indolent tumors. Given the comparatively
small size and technically challenging location of VS in
the context of PET imaging, a novel study design was
adopted by which patients were scanned using both a
conventional PET-CT and a high-resolution research
tomograph (HRRT), which has the highest spatial resolu-
tion for human brain PET (29). Through such an approach
the effect of scanner spatial resolution and reconstruction
methods on tracer uptake could also be assessed.
METHODS
Patients
Patients were recruited via the nationally commissioned,
specialized NF2 multidisciplinary team meeting in Manches-
ter, UK. Adult patients (aged between 18 and 70 yr of age) with
a confirmed diagnosis of NF2 and at least one vestibular
schwannoma (VS) were recruited. Exclusion criteria included:
female patients pregnant or intending to become pregnant;
patients who had undergone previous radiotherapy or antian-
giogenic treatment; and patients with contra-indications to
MRI. All patients gave informed written consent. The study
was approved by an independent research ethics committee
(REC 13/NW/0260) and by the United Kingdom Administra-
tion of Radioactive Substances Advisory Committee (ARSAC
RPC 595/3586/30119).
All patients had undergone previous routine clinical assess-
ment including MRI at 6 to 12 month intervals and the median
length of follow-up across all patients was 1.52 years
(range ¼0.60–7.30 yr). The study MRI scan was reviewed
in addition to the results of previous clinical MR imaging by
the multidisciplinary team and tumors were classified as either
rapidly growing or indolent. This classification reflected clini-
cal decision making in these patients with tumors being classi-
fied as indolent undergoing further radiological surveillance
and rapidly growing tumors being considered for either surgical
resection or treatment with the antiangiogenic agent bevacuzi-
mab (Avastin). To confirm the differential growth pattern
across these two cohorts, volumetric measurements of tumor
size were made for the preceding clinical scan, the study MR
scan, and a follow-up scan 1 year later (see Table 1). Volumetric
measurements were made on T
1
-weighted (T1W) postcontrast
imaging using the semiautomatic segmentation tool within the
Brainlab iPlan software (Brainlab AG Germany) and the results
FLT AND FDG PET IN NF2 VESTIBULAR SCHWANNOMA 827
Otology & Neurotology, Vol. 40, No. 6, 2019
of segmentation were reviewed and, where necessary, edited by
an experienced neuroradiologist (I.D.).
PET Data Acquisition
Patients were scanned using FDG and FLT on two separate
occasions, less than a week apart. For both tracers 200 MBq was
the target injected activity. Patients were scanned using both a
conventional PET-CT scanner, the Truepoint TrueV Biograph
PET-CT scanner (Siemens), with a spatial resolution of approx-
imately 4.5 mm full width at half maximum (30); and with a
brain dedicated scanner the (HRRT, Siemens) with spatial
resolution of approximately 2.5 mm full width at half maximum
(29). For each radiotracer, the scan sequence followed a 60-gap-
30-gap-30 minute structure with alternated order of scanners,
i.e., three scans for each radiotracer injection alternating
between the PET-CT and the HRRT (with sequence shown
in supplementary Figure 1C, http://links.lww.com/MAO/
A784). Patients were placed on one of the two scanners (with
the initial scanner altering between patients), injected with the
radiotracer, and data acquired for 60 minutes (scan 1). Follow-
ing a short break of between 10 and 20 minutes, patients were
placed on the other scanner, with data acquired for 30 minutes
(scan 2). Finally, following a second short break the patient was
placed on the original scanner with data acquired for a further
30 minutes (scan 3).
This scan sequence was devised to allow assessment of tracer
uptake during scan 2 at approximately 75 minutes postinjection
on both scanners, either from direct measurement or from linear
interpolation of data from scans 1 and 3 (radioactivity concen-
trations on the VS followed an approximately linear relation-
ship during this period for both tracers). For attenuation
correction, a 6-minute transmission scan was acquired when
using the HRRT (preinjection for scan 1 and postemission
acquisition for scans 2 and 3) and a pre-emission CT scan
when using the TrueV PET/CT scanner.
Image Reconstruction
Data from both scanners were reconstructed using imple-
mentations of three-dimensional iterative Ordinary Poisson
Ordered Subset Expectation Maximisation (31) without (No-
RM) and with resolution modeling (RM), reconstructing the
data during the last 30 minutes of scan 1 and the data for scans 2
and 3, each into three 10 minute frames. For the TrueV scanner,
the Siemens offline reconstruction package ‘‘e7_tools’’ was
used with an image zoom of two resulting in images with a
voxel size of 1.33 mm 1.33 mm 2.03 mm and an image grid
dimension of 256 256 107 voxels. HRRT data was recon-
structed using HRRT user community software generating
images consisting of 256 256 207 voxels each of size
1.22 mm 1.22 mm 1.22 mm. In both cases, 10 and 12 iter-
ations for No-RM and for RM respectively were conducted
using 16 subsets for HRRT and 21 for the TrueV. RM recon-
struction is referred to as HD for the TrueV PET (32) while for
the HRRT user community software was used (33). The iterations
and subsets selected reflect our standard image reconstruction
protocols. Postreconstruction smoothing using Gaussian filters,
which can be used to reduce image noise, was not performed
since it could worsen image resolution, which was considered to
be critical for this clinical application.
Reconstructions for both scanners were performed with full
corrections including scatter and attenuation. In the case of
HRRT, attenuation correction was calculated from a recon-
structed and segmented m-map image using the total variation
TXTV method (34). To minimize the effects of patient motion
TABLE 1. Patient demographics, tumor features, and clinical outcome at 1-year follow-up
Patient Location
Patient
Age
c
and
Sex
Growth
Classification
Volume on
Preceding
Clinical Scan
(cm
3
)
Volume at
Time of
PET Scan
(cm
3
)
Volume 1 Yr
Following
PET Scan
(cm
3
)
Annual
Adjusted
Volume Change
(cm
3
/yr)
Status of VS 1 Yr
Following the
PET Scan
A Right 33
Female
RG 0.29 0.36
a
Resected 0.09 Resected—cochlear
preserving surgery
Left RG 1.36 1.60 2.09 0.43 Continued growth.
Patient qualified for
Bevacizumab treatment.
B Right 48
Male
Indolent 0.95 0.84 0.87 0.04 Monitoring
Left Indolent 2.22 2.23 2.30 0.04 Monitoring
C
b
Right 32
Male
Indolent 0.85 0.79
a
0.82 0.02 Monitoring.
D Right 21
Female
RG 1.34 1.42 3.30 0.99 Continued growth.
Patient qualified for
Bevacizumab treatment.
Left Indolent 0.22 0.23
a
0.28 0.03 Monitoring
E Right 59
Male
RG 0.24 0.67
a
Resected 0.46 Resected- cochlear
preserving surgery
Left Indolent 0.04 0.07
a
0.07 0.02 Monitoring
F Right 55 Female Indolent Very small enhancing intracanalicular
nodule
N/A Monitoring
Left Indolent 0.26 0.23
a
0.20 0.03 Monitoring
a
Intracanalicular lesion at the time of PET scan.
b
Patient C had a large left VS removed 1 year before study.
c
Age at the time the PET scans took place.
RG indicates rapid growing.
828 J. M. ANTON-RODRIGUEZ ET AL.
Otology & Neurotology, Vol. 40, No. 6, 2019
particularly the deterioration of image resolution, image-based
motion correction using frame-by-frame realignment for each
10 minute frame was used for both scanners (35).
Delineation of Tumor VOI for PET Quantification
Tumor volumes of interest (VOI) for PET analysis were
manually drawn on contrast enhanced T1W MR images (voxel
size 0.9 mm 0.9 mm 0.8 mm), acquired as part of the study
MRI. Regions were drawn to the edge of the enhancing tumor
(care was taken when delineating the tumor to avoid partial
volume effects from nearby structures or surrounding CSF) and
subsequently were modestly eroded using a single iteration and a
331 erosion kernel. All manual outlining was done using
Analyze version 11 and was performed under the supervision of
AJ and ID, consultant neuroradiologists with over 40 years of
combined experience. The study MRI was acquired on the same
day as one of the PET scans for all the patients and therefore
within 1 week of both PET scans. Using SPM 8 (http://www.fi-
l.ion.ucl.ac.uk/spm), contrast enhanced T1W MRIs were core-
gistered to the 30minutes motion corrected PET images from
each of the three scans, and the manually drawn VOIs were re-
sliced to PET space using the rigid body transformations calcu-
lated from this coregistration and nearest-neighbor interpolation.
PET quantification was performed using the standardized
uptake value (SUV), whereby the radiotracer concentration at
75 minutes posttracer injection within each voxel was normal-
ized by the injected radioactivity dose and patient weight (36).
The tumor VOIs were then applied to the PET data to calculate
SUV
mean
(reflecting the overall regional tracer distribution),
SUV
max
(max value of the tracer distribution), and SUV
peak
within each tumor. The latter is considered to be less sensitive to
the VOI boundary and the uptake distribution (37).
Statistical Analysis
SPSS version 23 was used for all statistical analyses. The
normality and homogeneity of variance for derived values was
assessed using the Shapiro–Wilk and Levene test respectively.
Intergroup differences in growth rate, SUV
mean
, SUV
max
, and
SUV
peak
between indolent and growing tumors were compared
using a Student’s ttest. Linear regression analysis was under-
taken to assess intertumor relationship between standardized
uptake values for both FDG and FLT using each scanner with
and without RM. Finally, the ability of each tracer to classify
VS as rapidly growing was assessed using the area under the
curve (AUC) of the receiver operator characteristic (ROC)
curve for each SUV parameter, using the multidisciplinary
defined growth classification as the truth.
RESULTS
Patient Demographics
Six patients with NF2 participated in this study, three
males and three females with an age range of 21 to 59
years. Five patients had bilateral VS with the remaining
patient having undergone previous surgical removal of a
left-sided VS. Six tumors were intracanalicular at the time
of the PET study and among the 11 VS, 4 were classified as
rapidly growing while the rest were indolent (see Table 1).
Confirmatory measurements of tumor volume change
between the preceding clinical MRI and the study MRI
demonstrated that compared with the indolent tumor group
rapidly growing tumors displayed a higher annual adjusted
growth rate (0.00 versus 0.49 cm
3
/yr, p¼0.01, two-tailed
ttest). Patient demographics, tumor growth pattern, and
the clinical outcome for each VS at 1-year follow-up are
shown in Table 1. Mean injected tracer activities were
203 2 MBq (range 202 –210) of FLT and 206 4 MBq
(range 201– 211) of FDG.
Visual Inspection of Uptake
Uptake of both FDG and FLT was seen in all tumors,
using both scanners with and without RM. SUV mean,
maximum, and peak of both tracers at approximately 75
to 105 minutes after injection are shown for the TrueV
scanner in supplementary Tables 1C and 2C, http://link-
s.lww.com/MAO/A785; and for the HRRT scanner in
supplementary Tables 3C and 4C, http://links.lww.com/
MAO/A785.
Figure 1 shows axial coregistered T
1
-weighted contrast
enhanced MRI, FDG PET, and FLT PET image sections
for two patients(A and D) with bilateral tumors. All of the
PET images shown were acquired using the TrueV scanner
and show decay corrected SUV (g/ml) atapproximately 75
to 105 minutes postinjection with the FDG images win-
dowed to saturatethe high brain uptake. Patient A (top row)
had bilateral rapidly growing tumors, with the right smaller
VS scheduled for surgical removal at the time of the PET
scans. High uptake of both tracers is observed in the larger
left-sidedVS, with a small area of focal FDG uptake within
the right-sided tumor. Patient D (bottom row) also had
bilateral VS, with the right-sided tumor classified as
rapidly growing and the left-sided tumor classified as slow
growing (indolent). For both tracers clear uptake is
observed for the right-sided rapidly growing tumor while
little uptake is observed for the left-sided tumor.
Group Comparison
Intergroup differences in tumor SUV
mean
,SUV
max
, and
SUV
peak
between rapidly growing and indolent tumors for
both FDG and FLT are shown in Table 2. The group
comparison between FDG and FLT for both scanners
using RM is presented in Figure 2. Rapidly growing
tumors displayed significantly higher FDG SUV
mean
and SUV
peak
compared with indolent tumors using both
scanners, with and without RM ( p<0.05, one-tailed
ttest). With the exception of values derived using the
HRRT scanner without RM, the FDG SUV
max
values were
also significantly higher in the rapidly growing tumor
group ( p<0.05).
While use of the TrueV scanner without RM did not
demonstrate a significant difference in FLT uptakebetween
rapidly growing and indolent tumors ( p>0.05), use of the
TrueV with RM did demonstrate significantly higher FLT
SUV
mean
values in the rapidly growing tumors (p<0.05,
one-tailed ttest). Similarly, use of the HRRT scanner with
and without RM also demonstrated significantly higher
SUV
mean
and SUV
peak
values compared with indolent
tumors ( p<0.05, one-tailed ttest).
Scatter Plots
Scatter plots of SUV for FDG against FLT for the
TrueV and HRRT scanners are shown in Figure 3. Each
FLT AND FDG PET IN NF2 VESTIBULAR SCHWANNOMA 829
Otology & Neurotology, Vol. 40, No. 6, 2019
point of the graph represents one of the VS with data
shown for the SUV
mean
with and without RM (rows).
Lines of best fit for linear relationships are shown,
together with the fit equation and R-squared values.
VS classified as rapidly growing are plotted as a solid
circle, while indolent tumors are plotted as a square.
Visual inspection of the scatter plots in Figure 3
suggests that FDG and FLT are related to each other
in a proportional manner with the use of the higher
resolution HRRT scanner and/or RM improving the
correlation between FDG and FLT SUV
mean
values
(TrueV: adjusted R
2
value of 0.67 vs 0.73 with RM,
HRRT: adjusted R
2
value of 0.85 vs 0.86 with RM).
Similar plots for both SUV
max
and SUV
peak
without and
with RM can be found in supplementary Figures 1C and
2C, http://links.lww.com/MAO/A784, for the TrueV and
HRRT scanner respectively.
In supplementary Figure 4C, http://links.lww.com/
MAO/A784, scatter plots of the SUV
mean
for FDG and
FLT versus tumor volume for the TrueV scanner without
RM are shown. A weak positive correlation between
SUV
mean
and tumor volume is observed with adjusted R
2
values of 0.18 ( p¼0.11) and 0.08 ( p¼0.17) for FDG
and FLT respectively.
Area Under the Curve of the Receiver Operator
Characteristic Curve
AUC of the ROC curves for SUV mean, maximum,
and peak, for both tracers, and for both scanners are
shown in Table 3. Values ranged from 0.714 to 0.857
with SUV
mean
and from 0.786 to 0.821 with SUV
peak
,
suggesting a good ability of FDG and FLT SUV values to
discriminate indolent from rapidly growing tumors. Use
of RM for both scanners generally increased the AUC
FIG. 1. MRI and FDG/FLT PET images. From left to right: coregisteredaxial contrast enhanced MRI slices through cerebellopontine angle;
axial PET FDG images taken at 30 minutes using TrueV PET-CTscanner without RM; axial PET FLT images taken at 30 minutes using TrueV
PET-CT scanner without RM. Top row—33-year-old female (patient A) with bilateral growing VS. High uptake of both FDG and FLT is
observed in the larger left-sided VS, with a small area of focal FDG uptake within the right-sided tumor. Bottom row—21-year-old female
(patient D) with right-sided growing VS and left-sided slow growing (indolent) tumor. The right lesion shows uptake of both FLTand FDG while
the left lesion showed minimal uptake of either tracer. All PET images show the summed activity at approximately 75 to 105 minutes
postinjection.
830 J. M. ANTON-RODRIGUEZ ET AL.
Otology & Neurotology, Vol. 40, No. 6, 2019
values. Overall, FDG displayed higher AUC values than
FLT (0.750–0.893 vs 0.643– 0.857) with the exception of
SUV
mean
when using the HRRT scanner with RM, where
FLT displayed greater discriminatory power (AUC 0.857
vs 0.821). Both FDG and FLT outperformed tumor
volume in discriminating between rapidly growing and
indolent tumors with AUC
ROC
value of 0.601.
DISCUSSION
In this pilot study we have demonstrated for the first
time that there is uptake of two commercially available
PET radiotracers, FDG and FLT, within NF2 related VS
and that uptake of these tracers has the potential ability to
discriminate rapidly growing VS from more indolent
tumors. This was established through a complex study
design to elucidate the relative contributions of tracer,
noise, and spatial resolution to the PET signal. The data
demonstrates, however, that a short PET acquisition with
clinically available tracers on a standard scanner can
yield clinically relevant information on tumor growth.
The finding of growth-dependant uptake of FDG in
NF2 related VS is in clear contrast to previous inconclu-
sive results with FDG seen in sporadic VS (20,21). While
differences in experimental design may partly underlie
this discordance, greater uptake of FDG within NF2
related VS may also reflect fundamental biological dif-
ferences between these two tumor groups at both the
macroscopic and microscopic level. While sporadic VS
are generally found as a single tumor arising from the
vestibular nerve at the porus acousticus (38), NF2 related
tumors are often multilobulated, originating from multi-
ple sites on both the vestibular and cochlear nerve (26).
At the cellular level, NF2 related VS display higher
cellularity (27) and greater immunostaining for cellular
proliferation indices (e.g., Ki-67, MIB-1) compared with
sporadic tumors (28,39). Furthermore, there is evidence
that pathophysiological mechanisms other than cellular
proliferation such as cyst formation (40,41), intratumoral
hemorrhage (42–44), and inflammation (44 – 47) may
play a greater role in the growth of sporadic VS.
While uptake of FDG and FLT represent differing
underlying biological processes, the uptake of both these
tracers within NF2 related VS was strongly correlated in
our study. One interpretation is that the uptake of FDG
and FLT relates to a common factor or process such as
tumor size or vascularity, but the correlation between
tracer uptake and tumor volume was, however, compar-
atively weaker than the relationship between FDG and
FLT uptake itself. Similarly while increased neovascu-
larization within growing tumors may result in greater
early tracer delivery (48,49), with the later PET measure-
ments (75–105 min) used in this study these effects
would be minimal. As such, the increased uptake of both
FLT and FDG seen in this study likely represents that
within growing NF2 related VS there is both concurrent
cellular proliferation and increased metabolic demand.
Imaging VS with FDG and FLT has been previously
viewed as challenging due to the limited spatial
TABLE 2. Intertumor comparison of derived mean, maximum, and peak SUV values (g/ml) between slow growing (indolent) and fast growing tumors following the injection
of FDG and FLT
FDG—Intragroup Mean (þ/S.D)
TrueV No-RM TrueV RM HRRT No-RM HRRT RM
N SUV Mean SUV Max SUV Peak SUV Mean SUV Max SUV Peak SUV Mean SUV Max SUV Peak SUV Mean SUV Max SUV Peak
Slow growing (indolent) 7 2.11 (0.88) 4.74 (1.90) 2.57 (1.01) 2.01 (0.94) 4.42 (2.09) 2.56 (1.25) 1.81 (0.87) 10.58 (5.04) 2.28 (1.07) 1.71 (1.00) 5.42 (3.25) 2.19 (1.27)
Fast growing 4 3.49 (1.26) 7.21 (2.16) 4.09 (1.50) 3.61 (1.43) 8.40 (3.57) 4.41 (1.82) 2.96 (0.93) 15.00 (3.97) 3.82 (1.21) 2.95 (0.90) 9.00 (2.25) 3.86 (1.18)
pvalue p<0.05 p<0.05 p<0.05 p<0.05 p<0.05 p<0.05 p<0.05 p¼0.08 p<0.05 p<0.05 p<0.05 p<0.05
FLT- Intragroup Mean (þ/SD)
TrueV No-RM TrueV RM HRRT No-RM HRRT RM
N SUV Mean SUV Max SUV Peak SUV Mean SUV Max SUV Peak SUV Mean SUV Max SUV Peak SUV Mean SUV Max SUV Peak
Slow growing (indolent) 7 0.86 (0.49) 2.47 (1.34) 1.07 (0.60) 0.94 (0.65) 2.47 (1.81) 1.21 (0.83) 0.68 0.37) 7.36 (3.23) 0.97 (0.47) 0.74 (0.48) 4.39 (3.03) 1.08 (0.67)
Fast growing 4 1.27 (0.37) 3.47 (1.17) 1.63 (0.57) 1.62 (0.39) 4.43 (1.77) 2.13 (0.74) 1.13 (0.39) 10.18 (4.26) 1.61 (0.66) 1.34 (0.39) 7.14 (2.43 1.92 (0.70)
pvalue p¼0.09 p¼0.12 p¼0.08 p<0.05 p¼0.06 p¼0.05 p<0.05 p¼0.12 p<0.05 p<0.05 p¼0.08 p<0.05
Displayed data—intragroup mean SUV (SD). All SUV values derived at approximately 75 to 105 minutes postinjection.
FLT AND FDG PET IN NF2 VESTIBULAR SCHWANNOMA 831
Otology & Neurotology, Vol. 40, No. 6, 2019
resolution of conventional PET, leading to potential
contamination of tumor uptake from surrounding brain
and bony marrow respectively (50). To assess this we
used a complex scanning regime which incorporated two
different PET scanners with different spatial resolution,
both with and without RM reconstruction, and without
any postreconstruction image smoothing. One conse-
quence of this approach is that noise in the images is
increased and this may explain the reduced discrimina-
tory power of SUV
max
when compared with SUV
mean
and
SUV
peak
. Use of either RM or the higher spatial resolu-
tion HRRT scanner improved the proportional relation-
ship between FDG and FLT suggesting that when tumor
uptake contamination from neighboring tissues is
reduced, a better correlation between the two imaged
biological processes is observed. Use of the HRRT
scanner or RM, however, resulted in only small improve-
ments in AUC
ROC
values suggesting that the degree of
contamination from neighboring structures is small in
comparison with the tumor uptake range, and that
increased spatial resolution has only a modest effect
on tumor growth classification. As such use of more
clinically available lower spatial resolution PET scanners
such as the TrueV PET-CT scanner may still show good
ability to discriminate growing VS.
The results of this study demonstrate that both FDG
and FLT uptake has merit to discriminate between
rapidly growing and slow growing (indolent) tumors,
and that this discriminatory ability exceeds that of
tumor volume alone. While standard clinical practice
in many institutions is radiological surveillance of
tumor growth with serial MRI, there is a danger of
missing significant tumor growth between interval
scans, with the complication rate and difficulty of
surgery increasing as tumors become larger (51,52).
In many cases this strategy will be acceptable but
select patients exist in whom the ability to predict
rather than detect growth may be valuable. The above
results suggest that assessment of tumor proliferative
and metabolic activity using FDG or FLT PET
may have future clinical utility in allowing more
timely identification of tumors requiring surgical inter-
vention.
A limitation of this study is that the number of included
patients was low, due in part to patient concerns regard-
ing additional radiation exposure and the complexity of
the scanning regime. Future, larger studies, which incor-
porate just one scanner and a single tracer injection of
either FDG or FLT, should be performed. These studies
could be performed on new generation PET-MR
FIG. 2. Intertumor comparison boxplots of derived mean and peak SUV values (g/ml) between slow growing (indolent) and fast growing VS
following the injection of FDG and FLT. Using the TrueV PET-CTscanner with RM for FDG (A) and FLT (C); using the HRRTscanner with RM
for FDG (B) and FLT (D).
p<0.05—one tailed Student’s ttest, comparison between slow growing/indolent and fast growing VS for each
SUV parameter.
832 J. M. ANTON-RODRIGUEZ ET AL.
Otology & Neurotology, Vol. 40, No. 6, 2019
scanners, which allow for both simultaneous MR image
acquisition and also potentially for reductions in the
injected radioactive dose due to improved scanner sen-
sitivity (53). Evaluation of FDG and FLT PET as pre-
dictive markers of future tumor growth is limited in part
in this study due to loss of growth follow-up in resected
tumors. It is, nonetheless, interesting to note that within
this study the two non-resected rapidly growing tumors
with high FDG and FLT uptake continued to demonstrate
rapid growth and larger, prospective studies should be
undertaken to further evaluate the role of these tracers as
growth predictors.
CONCLUSIONS
Data from 6 NF2 patients, with a total of 11 VS,
indicate that for both FLT and FDG an uptake signal
above background can be detected and that this uptake
shows promise in providing additional and complemen-
tary information to serial MRI measurements for the
classification of VS which are rapidly growing. Further
studies should be undertaken to assess FLT and FDG
PET as predictors of tumor growth, and as a clinical
imaging tool for early identification of tumors requiring
consideration of early treatment.
FIG. 3. Intertumor scatterplot comparison of FDG versus FLTuptake. Scatter plots of the mean tumor SUV values of FDG against FLT for
both the TrueV (left column) and HRRTscanner (right column) without RM and with RM. For each graph, a line of best fit for propor tionality is
shown, along with the equation and R
2
values. Lesions classified as rapid growing (growing) are shown as solid circles, whereas slow
growing/indolent lesions are shown as an open square.
TABLE 3. Receiver operator characteristic curve (ROC) area under the curve (AUC) values when using volume of lesion (top), and
mean, maximum, and peak SUV values (g/ml) of FDG and FLT within contrast enhanced VS lesions to classify lesion growth at
approximately 75 to 105 minutes following the injection
0.601
TrueV No-RM TrueV RM HRRT No-RM HRRT RM
Volume Tracer SUV
Mean
SUV
Max
SUV
Peak
SUV
Mean
SUV
Max
SUV
Peak
SUV
Mean
SUV
Max
SUV
Peak
SUV
Mean
SUV
Max
SUV
Peak
FDG 0.821 0.786 0.786 0.821 0.821 0.821 0.821 0.750 0.821 0.821 0.893 0.821
FLT 0.714 0.714 0.786 0.821 0.750 0.786 0.786 0.643 0.786 0.857 0.750 0.821
Combined FDG and FLT 0.786 0.821 0.821 0.821 0.821 0.821 0.821 0.679 0.821 0.857 0.857 0.821
Data shown for both the TrueV PET-CT and HRRT PET scanners with and without RM.
FLT AND FDG PET IN NF2 VESTIBULAR SCHWANNOMA 833
Otology & Neurotology, Vol. 40, No. 6, 2019
Acknowledgments: The authors thank Patricia Braithwaite and
Raji Anup for efforts and help in the recruitment of patients at
the Highly Specialised Commissioned NF2 service, Central
Manchester University Hospital NHS Foundation trust. The
authors also thank the staff of the Wolfson Molecular Imaging
Centre for help and support to set and run this project, in
particular Prof Karl Herholz with his assistance with the
ARSAC license and Sarah Wood with the submission of the
project to ethics committee.
REFERENCES
1. Evans DG, Howard E, Giblin C, et al. Birth incidence and
prevalence of tumor-prone syndromes: Estimates from a UK
family genetic register service. Am J Med Genet 2010;152A:
327– 32.
2. Evans DGR. Neurofibromatosis type 2 (NF2): A clinical and
molecular review. Orphanet J Rare Dis 2009;4:16.
3. Evans DGR, Baser ME, O’Reilly B, et al. Management of the
patient and family with neurofibromatosis 2: A consensus confer-
ence statement. Br JNeurosurg 2005;19:5– 12.
4. Dirks MS, Butman JA, Kim HJ, et al. Long-term natural history of
neurofibromatosis Type 2-associated intracranial tumors. J Neuro-
surg 2012;117:109– 17.
5. Hexter A, Jones A, Joe H, et al. Clinical and molecular predictors of
mortality in neurofibromatosis 2: A UK national analysis of 1192
patients. J Med Genet 2015;52:699– 705.
6. Chen L-H, Zhang H-T, Xu R-X, et al. Microsurgery for patients
diagnosed with neurofibromatosis type 2 complicated by vestibular
schwannomas: Clinical experience and strategy for treatments.
Medicine 2018;97:e0270– 280.
7. Chen L, Chen L, Liu L, et al. Vestibular schwannoma microsurgery
with special reference to facial nerve preservation. Clin Neurol
Neurosurg 2009;111:47– 53.
8. Nowak A, Dziedzic T, Czernicki T, et al. Strategy for the surgical
treatment of vestibular schwannomas in patients with neurofibro-
matosis type 2. Neurol Neurochir Pol 2015;49:295– 301.
9. Brackmann DE, Fayad JN, Slattery WH 3rd, et al. Early proactive
management of vestibular schwannomas in neurofibromatosis type
2. Neurosurgery 2001;49:274– 80.
10. van de Langenberg R, de Bondt BJ, Nelemans PJ, et al. Follow-up
assessment of vestibular schwannomas: Volume quantification
versus two-dimensional measurements. Neuroradiology 2009;51:
517– 24.
11. Harris GJ, Plotkin SR, Maccollin M, et al. Three-dimensional
volumetrics for tracking vestibular schwannoma growth in neuro-
fibromatosis type II. Neurosurgery 2008;62:1314– 9.
12. Baser ME, Mautner V-F, Parry DM, et al. Methodological issues in
longitudinal studies: Vestibular schwannoma growth rates in neuro-
fibromatosis 2. J Med Genet 2005;42:903– 6.
13. Solares CA, Panizza B. Vestibular schwannoma: An understanding
of growth should influence management decisions. Otol Neurotol
2008;29:829– 34.
14. Reivich M, Kuhl D, Wolf A, et al. The [18F]fluorodeoxyglucose
method for the measurement of local cerebral glucose utilization in
man. Cir Res 1979;44:127– 37.
15. Hsu PP, Sabatini DM. Cancer cell metabolism: Warburg and
beyond. Cell 2008;134:703– 7.
16. Warburg O. On the origin of cancer cells. Science 1956;123:309–
14.
17. Endo K, Oriuchi N, Higuchi T, et al. PET and PET/CT using 18F-
FDG in the diagnosis and management of cancer patients. Int J Clin
Oncol 2006;11:286– 96.
18. Boellaard R, Delgado-Bolton R, Oyen WJG, et al. FDG PET/CT:
EANM procedure guidelines for tumour imaging: Version 2.0. Eur
J Nucl Med Mol Imaging 2015;42:328– 54.
19. Shields AF, Grierson JR, Dohmen BM, et al. Imaging proliferation
in vivo with [F-18]FLT and positron emission tomography. Nat
Med 1998;4:1334– 6.
20. Chen JM, Houle S, Ang LC, et al. A study of vestibular schwan-
nomas using positron emission tomography and monoclonal anti-
body Ki-67. Am J Otol 1998;19:840– 5.
21. Sakamoto H, Nakai Y, Matsuda M, et al. Positron emission tomo-
graphic imaging of acoustic neuromas. Acta Otolaryngol Suppl
2000;542:18– 21.
22. Muzi M, Spence AM, O’Sullivan F, et al. Kinetic analysis of 30-
deoxy-30-18F-fluorothymidine in patients with gliomas. J Nucl Med
2006;47:1612– 21.
23. Schiepers C, Chen W, Dahlbom M, et al. 18F-fluorothymidine
kinetics of malignant brain tumors. Eur J Nucl Med Mol Imaging
2007;34:1003– 11.
24. Jacobs AH, Thomas A, Kracht LW, et al. 18F-fluoro-L-thymidine
and 11C-methylmethionine as markers of increased transport and
proliferation in brain tumors. J Nucl Med 2005;46:1948– 58.
25. Chen W, Cloughesy T, Kamdar N, et al. Imaging proliferation in
brain tumors with 18F-FLT PET: Comparison with 18F-FDG.
J Nucl Med 2005;46:945– 52.
26. Stivaros SM, Stemmer-Rachamimov AO, Alston R, et al. Multiple
synchronous sites of origin of vestibular schwannomas in neurofi-
bromatosis Type 2. J Med Genet 2015;52:557–62.
27. Sobel RA, Wang Y. Vestibular (Acoustic) schwannomas: Histo-
logic features in neurofibromatosis 2 and in unilateral cases.
J Neuropathol Exp Neurol 1993;52:106– 13.
28. Aguiar PH, Tatagiba M, Samii M, et al. The comparison between
the growth fraction of bilateral vestibular schwannomas in neuro-
fibromatosis 2 (NF2) and unilateral vestibular schwannomas using
the monoclonal antibody MIB 1. Acta Neurochir 1995;134:40– 5.
29. de Jong HW, van Velden FH, Kloet RW, et al. Performance
evaluation of the ECAT HRRT: An LSO-LYSO double layer
high resolution, high sensitivity scanner. Phys Med Biol 2007;52:
1505– 26.
30. Jakoby BW, Bercier Y, Watson CC, et al. Physical performance and
clinical workflow of a new LSO HI-REZ PET/CT scanner. IEEE
Nucl Sci Symp Medical Imaging Conf Rec 2007:3130–4.
31. Hudson HM, Larkin RS. Accelerated image reconstruction using
ordered subsets of projection data. IEEE Trans Med Imaging
1994;13:601– 9.
32. Panin VY, Kehren F, Michel C, et al. Fully 3-D PET reconstruction
with system matrix derived from point source measurements. IEEE
Trans Med Imaging 2006;25:907– 21.
33. Comtat C, Sureau FC, Sibomana M, et al. Image based resolution
modeling for the HRRT OSEM reconstructions software. IEEE
Nucl Sci Symp Medical Imaging Conf Rec 2008:4120–3.
34. Keller SH, Svarer C, Sibomana M. Attenuation correction for the
HRRT PET-scanner using transmission scatter correction and total
variation regularization. IEEE Trans Med Imaging 2013;32:
1611– 21.
35. Anton-Rodriguez JM, Sibomana M, Walker MD, et al. Investigation
of motion induced errors in scatter correction for the HRRT
brain scanner. IEEE Nucl Sci Symp Med Imaging Conf Rec
2010:2935– 40.
36. Boellaard R, Krak NC, Hoekstra OS, et al. Effects of noise, image
resolution, and ROI definition on the accuracy of standard uptake
values: A simulation study. J Nucl Med 2004;45:1519–27.
37. Wahl RL, Jacene H, Kasamon Y, et al. From RECIST to PERCIST:
Evolving considerations for PET response criteria in solid tumors. J
Nucl Med 2009;50 (suppl 1):122S– 50S.
38. Neely JG. Gross and microscopic anatomy of the eighth cranial
nerve in relationship to the solitary schwannoma. Laryngoscope
1981;91:1512– 31.
39. Antinheimo J, Haapasalo H, Seppala M, et al. Proliferative potential
of sporadic and neurofibromatosis 2-associated schwannomas as
studied by MIB-1 (Ki-67) and PCNA labeling. J Neuropathol Exp
Neurol 1995;54:776– 82.
40. Charabi S, Mantoni M, Tos M, et al. Cystic vestibular schwanno-
mas: Neuroimaging and growth rate. J Laryngol Otol 1994;
108:375– 9.
41. Sinha S, Sharma BS. Cystic acoustic neuromas: Surgical outcome in
a series of 58 patients. J Clin Neurosci 2008;15:511–5.
834 J. M. ANTON-RODRIGUEZ ET AL.
Otology & Neurotology, Vol. 40, No. 6, 2019
42. Gomez-Brouchet A, Delisle MB, Cognard C, et al. Vestibular
schwannomas: Correlations between magnetic resonance imaging
and histopathologic appearance. Otol Neurotol 2001;22:79– 86.
43. Park CK, Kim DC, Park SH, et al. Microhemorrhage, a possible
mechanism for cyst formation in vestibular schwannomas. J Neuro-
surg 2006;105:576– 80.
44. de Vries M, Hogendoorn PC, Briaire-de Bruyn I, et al. Intratumoral
hemorrhage, vessel density, and the inflammatory reaction contrib-
ute to volume increase of sporadic vestibular schwannomas. Virch-
ows Arch 2012;460:629– 36.
45. Lewis D, Roncaroli F, Agushi E, et al. Inflammation and vascular
permeability correlate with growth in sporadic vestibular schwan-
noma. Neuro Oncol 2018. [Epub ahead of print].
46. de Vries M, Briaire-de Bruijn I, Malessy MJ, et al. Tumor-associ-
ated macrophages are related to volumetric growth of vestibular
schwannomas. Otol Neurotol 2013;34:347– 52.
47. de Vries M, van der Mey AG, Hogendoorn PC. Tumor biology of
vestibular schwannoma: A review of experimental data on the
determinants of tumor genesis and growth characteristics. Otol
Neurotol 2015;36:1128– 36.
48. Plotkin SR, Stemmer-Rachamimov AO, Barker FG, et al. Hearing
improvement after bevacizumab in patients with neurofibromatosis
type 2. N Engl J Med 2009;361:358–67.
49. Wong HK, Lahdenranta J, Kamoun WS, et al. Anti-vascular
endothelial growth factor therapies as a novel therapeutic approach
to treating neurofibromatosis-related tumors. Cancer Res
2010;70:3483– 93.
50. Soret M, Bacharach SL, Buvat I. Partial-volume effect in PET
tumor imaging. J Nucl Med 2007;48:932– 45.
51. Odat HA, Piccirillo E, Sequino G, et al. Management strategy of
vestibular schwannoma in neurofibromatosis type 2. Otol Neurotol
2011;32:1163– 70.
52. Zhao F, Wang B, Yang Z, et al. Surgical treatment of large
vestibular schwannomas in patients with neurofibromatosis type
2: Outcomes on facial nerve function and hearing preservation. J
Neurooncol 2018;138:417– 24.
53. Grant AM, Deller TW, Khalighi MM, et al. NEMA NU 2-2012
performance studies for the SiPM-based ToF-PET component
of the GE SIGNA PET/MR system. Med Phys 2016;43:
2334– 43.
FLT AND FDG PET IN NF2 VESTIBULAR SCHWANNOMA 835
Otology & Neurotology, Vol. 40, No. 6, 2019
