28 patients [14 female; mean age 20 (2-44) years] with neurofibromatosis type 1 were included.
15 patients were children below the age of 18 years.
6/28 patients presented with one target lesion,
5/28 patients presented with two target lesions,
8/28 patients presented with three target lesions,
4/28 patients with four target lesions,
3/28 patients with five target lesions and 2/28 patients with six target lesions,
respectively.
In the 28 patients cohort,
83 lesions were evaluated,
of which 75 were rated as benign target lesions,
53/75 lesions were histologically examined after surgical resection and 22/75 lesions were not resected due to completely stable clinical course and imaging follow-ups (reference standard).
In 6 patients 8 histologically proven MPNSTs were found,
5/6 patients had one MPNST and one patient had three MPNSTs.
PET/MRI imaging
23/83 target lesions were localized in the extremities,
25/83 target lesions were localized subcutaneously at the abdominal body trunk,
intraabdominally or in the pelvis,
25/83 target lesions were localized thoracocervically and 10/83 target lesions were localized paravertebrally.
Image examples of a benign target lesion and a MPNST are illustrated in Fig.2.
The mean effective radiation dose patients received in our study was 3.24±1.65 mGy.
The estimated mean examination time was 95±21 minutes per patient including five NF1 patients who had to be repositioned in order to achieve whole body coverage.
Quantitative PNF lesion measurements
The mean size of benign PNF lesions was 2.65±1.83cm,
whereas the mean size of MPNSTs was 3.54±1.11cm (P<.01).
The measured PET SUVmean of histologically proven MPNSTs was significantly higher than the SUVmean of benign PNF lesions (3.84±3.98 [MPNSTs] vs.
1.85±1.03 [PNF]; P<.01) (Fig.3a).
The SUVmax of histologically proven MPNSTs was similarly higher than the SUVmax of benign PNF (5.84±6.10 [MPNSTs] vs.
3.03±1.92 [PNF]; P<.01) (Fig.3b).
Similarly,
the lesion SUVmean-to-liver SUVmean ratio significantly differed between MPNSTs and PNF lesions (3.20±2.70 [MPNSTs] vs.
1.23±0.61 [PNF]; P<.01) (Fig.4).
As significant cut-off values for differentiation between still benign PNF and MPNSTs we calculated SUVmax ≥ 2.78 (sensitivity 0.88; specificity 0.73) and 1.45 for lesion SUVmean-to-liver SUVmean ratio (sensitivity 0.88; specificity 0.79).
MPNSTs showed only a tendency for higher diffusion restriction in large ROI analysis,
which however did not reach statistical significance (ADCmean values of benign PNF lesions and MPNSTs (1.87±0.24 x10−3 mm2/s [PNF] vs.
1.76±0.11 x10−3 mm2/s [MPNSTs]; P=1.0].
(Fig.5a).
Similarly,
small ROI analysis in lesion areas with the highest 18F-FDG-uptake showed a tendency towards lower ADCmean/min values in MPNSTs compared to benign PNF with SUVmean > 2 without statistical significance (P>.05) (Fig.5b).
24/28 patients had both a previous (mean time interval: 21.63±13.38 months) and follow-up (mean time interval: 13.26±7.94 months) PET/MRI or MRI examination.
The estimated lesion growth rate correlated significantly with an increased glucose consumption as measured by means of PET SUVmean (rs=.41; P=.003) (Fig.6a),
whereas no significant correlation was found between the lesion growth rate and the MRI ACDmean in large ROI analysis (rs=.07; P=.67) (Fig.6b).
25/75 benign PNF lesions presenting with a SUVmean value below that of bloodpool SUVmean (0.95±0.30) showed a significant lower growth rate per month than 50/75 benign PNF lesions with a relatively elevated SUVmean value (-0.32±1.00%/month [SUVmean < bloodpool] vs.
1.10±2.25%/month [SUVmean > bloodpool]; P <.05) (Fig.7).
Qualitative radiological evaluation
49/75 benign PNF lesions and 6/8 MPNSTs showed an enhancement of contrast media,
whereas 26/75 benign PNF lesions did not enhance (P=.50).
35/75 benign PNF lesions and 4/8 MPNSTs had a positive target sign (P=.86).
8/28 patients did not get an examination of the brain in PET/MRI because of an inconspicuous brain MRI examination performed shortly before the PET/MRI scan.
The PET/MRI of 20/28 patients included a dedicated T2 weighted FLAIR brain protocol for screening purposes as described above.
Out of these 20 patients,
14 patients (70%) had T2-hyperintense focal areas of signal intensity (FASI) in brain parenchyma.
Further,
we found changes compatible with optic nerve gliomas in 8 patients (40%) (Fig.8).