Enrico Martin

123 Diagnostic accuracy of non-invasive tests Four studies reported on irregular shape. 31,35,38,39 Pooled sensitivity and specificity were 0.33 (95% CI: 0.04-0.73) and 0.82 (95% CI: 0.71-0.90) respectively. Pooled pLR was 2.03 (0.18-5.42) and nLR was 0.81 (0.26-1.22). There was substantial heterogeneity between studies. Sensitivity was higher in studies including a larger proportion of MPNST and those including NF1 patients only. Three studies reported on tumor size. 21,33,36 Thresholds varied from 4.7-6.3cm. Pooled sensitivity and specificity were 0.72 (95% CI: 0.47-0.92) and 0.85 (95% CI: 0.69-0.94) respectively. Pooled pLR was 5.63 (2.05-13.65) and nLR was 0.34 (0.07-0.67). There was moderate heterogeneity between studies. Sensitivity was higher in studies with a higher proportion of MPNST. Specificity was higher in studies with smaller sample of lesions, lower proportion of MPNST, and those including histologically proven lesions only. Specificity was lower in studies including NF1 patients only. Three studies reported on qualitative MRI assessment. 32,33,41 Pooled sensitivity and specificity were 0.71 (95% CI: 0.53-0.85) and 0.92 (95% CI: 0.81-0.98) respectively. Pooled pLR was 12.44 (2.13-39.05) and nLR was 0.32 (0.15-0.56). There was limited heterogeneity between studies, but no source of heterogeneity was found. Functional MRI characteristics Six studies reported on 16 functional MRI characteristics ( Supplementary Table 2 ). 21,29,31,38,42,43 No characteristic was evaluated in more than 2 different populations. Mean apparent diffusion coefficient (ADCmean) was evaluated in two studies. 38,42 Sensitivity ranged from 0.91-0.92 and specificity from 0.91-0.98. pLR ranged from 10.46- 50.42 and nLR from 0.09-0.10. Minimal ADC (ADCmin) was evaluated in two studies as well. 21,38 Sensitivity ranged from 0.89-0.98 and specificity from 0.93-0.94. pLR ranged from 14.15 to 14.43 and nLR from 0.03-0.12. One study used diffusion coefficient D and perfusion fraction f to investigate a number of characteristics. 38 Sensitivities ranged from 0.81-0.96 and specificity from 0.55-0.98. pLR ranged from 2.11-99.08 and nLR from 0.04-0.22. Dmin and fcenter yielded highest sensitivities (0.96), and dark and Dmargin highest specificity (0.99). One study reported on using the target sign on ADC and diffusion-weighted imaging (DWI). 29 Sensitivity ranged from 0.80-0.97 and specificity from 0.39-0.63. pLR ranged from 1.32-2.64 and nLR from 0.05-0.51. One study evaluated early arterial enhancement on dynamic contrast enhancement MRI. 31 Sensitivity was 0.50 and specificity 0.89, with a pLR of 4.50 and nLR of 0.56. Accuracy was highest when evaluating target sign on ADC mapping with higher specificity compared to static T1 weighted imaging. One study reported on trimethylamine (TMA) peak and TMA fraction. 43 Sensitivity was 0.90 for both and specificity was 0.50 for TMA peak and 0.62 for TMA fraction. pLR ranged from 1.8-2.35 and nLR from 0.16-0.20. 6