This retrospective study was approved by our local institutional ethical review board and informed consent was waived (Project Number: 345/2018BO).
The underlying study population consisted of all patients with NF1 undergoing FDG-PET/MRI in our institution between September 2012 and February 2018 presenting with a clinical question referring to malignant transformation of known plexiform neurofibromas.
PET/MRI imaging
All combined PET/MRI examinations were performed on an integrated clinical PET/MRI system (Biograph mMR,
Siemens Healthcare GmbH,
Erlangen,
Germany,
software versions B20P and E11P) which is able to acquire PET and MRI simultaneously.
For the generation of a segmentation-based PET attenuation correction map,
a whole-body 3D T1-weighted spoiled gradient-echo sequence in end-expiratory breath-hold with Dixon-based fat-water separation was acquired.
In patients newly examined after 01/2017 atlas-based bone-estimation was additionally available and performed for the purpose of attenuation correction.
In all PET/MRI examinations the following MR measurements were performed: a transversal and coronal T2-weighted turbo spin echo (TSE) sequence,
a coronal whole body short time inversion recovery (STIR) sequence in free breathing,
whole body diffusion weighted imaging (DWI),
whole body T1-weighted volumetric interpolated breath-hold examination (VIBE) sequence after intravenous injection of 0.1 mmol/kg gadolinium-based MRI contrast media (GADOVIST®),
a fluid attenuated inversion recovery (FLAIR) sequence of the head as well as a contrast-enhanced T1-weighted 3D magnetization prepared rapid gradient echo (MPRAGE) sequence of the head.
Patients fasted for at least 6 hours before intravenous injection of [18F] fluorodeoxyglucose (FDG).
The recommended dose for whole body FDG-PET is weight-dependent and ranges between 3.5-7 MBq/kg for a 2-minute scan [16].
As the PET acquisition in PET/MRI is longer (4 min per bed in our case),
we reduced these values by a factor of about 2 based on previous data [17].
The injected dose of 18F-FDG patients received was adjusted to patient body weight (average: 2.5±0.60 MBq/kg).
The corresponding effective doses of PET in pediatrics and adults were calculated from the applied activity,
as described in a previous study [18].
PET acquisition was initiated 60 minutes after tracer injection.
The whole body scan was acquired over 6±2 bed positions.
PET was reconstructed using a 3D ordered-subset expectation-maximization algorithm with 2 iterations,
21 subsets,
matrix size 256 x 256,
Gaussian filtering of 4 mm.
The patient examination times were measured based on the acquisition time stamps that are documented in our Picture Archiving Communication System (PACS) and included an interval of 10 minutes for repositioning patients in order to achieve whole body coverage,
as the scan range was limited to 150 cm until the scanner was updated with an additional scanner coil in 2016.
Quantitative PNF lesion measurements
A maximum of six peripheral nerve target lesions were defined per patient.
Of these,
a maximum of four nerve target lesions per patient with visibly increased FDG uptake above blood pool levels and a maximum of two target lesions in similar anatomical localization and with similar size without visibly increased FDG uptake were selected.
Entirely diffuse configurated plexiform neurofibromas without a definable geometry in MRI,
typically infiltrating skin or muscle,
were excluded from the evaluation.
Image analysis was performed using the software SyngoVia (Siemens,
Erlangen,
Germany).
Lesion size was determined by measurement of the maximum axial diameter of each target lesion using the T1-weighted MRI sequence after intravenous contrast media application.
For all peripheral nerve lesions,
PET quantification was performed measuring the mean,
maximum and peak standardized uptake values (SUVmean/max/peak) based on 50%-isocontour volumes of interests (VOIs).
SUVmax is defined as the highest single-pixel value within a defined volume of interest (VOI),
whereas SUVpeak is defined as an average SUV within a small,
fixed-sized VOI (1 ml) centered on maximum-uptake part of the lesion [19].
For measuring the SUVmean of reference tissues,
we placed a 2 cm-diameter ROI in the right atrium (bloodpool) and a 5 cm-diameter ROI in the liver parenchyma.
In all patients,
lesion SUVmean-to-liver SUVmean ratios were calculated.
In MRI,
we measured the mean and minimum apparent diffusion coefficients (ADC mean; ADCmin) of all target lesions using circular regions of interest (ROI) with a radius comprising the whole of the lesion on the level of its largest transverse cross-section (large ROI analysis).
Additionally,
we applied small ROI measurements as previously described [20] in suspicions PNF parts with SUVmean above 2 and MPNSTs by placing a ROI into the lesion area with the highest 18F-FDG-uptake (small ROI analysis).
To assess the long-term development in size of all included PNF lesions,
we measured the maximum axial lesion size also in available previous and follow-up PET/MRI or MRI examinations in which the same lesions were assessable (Fig.1).
The growth rate was calculated by the quotient of axial diameter change from previous to follow-up examination and the time interval in months.
Qualitative radiological evaluation
For qualitative analysis,
all image data were assessed by two radiologists in consensus.
The morphological characteristics of all target lesions were categorized as target-like or not target-like.
Target-like lesions were defined as centrally hypointense in T2-weighted images with a hyperintense rim resembling a target within PNF of peripheral nerves or large PNF accumulations in certain body areas.
Contrast-medium enhanced target lesions were defined visually as clearly hyperintense lesions in T1-weighted images after intravenous MRI contrast medium injection.
Furthermore,
we evaluated the presence of incidental areas of high signal intensity on T2-weighted FLAIR sequence in the white substance (white matter lesions),
which are typical MR findings in the cerebellum,
brainstem,
basal ganglia and thalami [21].
Also,
the presence of visible optic nerve gliomas was evaluated.
Reference standard
All specimens of resected lesions were histologically examined by our in-house pathology.
For non-resected lesions,
both the clinical course and imaging follow-up were used as clinical reference standard to characterize the lesion as benign or malignant.
The surgical indication was based on a tumor board decision in which all individual cases were discussed.