Enrico Martin

126 Chapter 6 Seven studies reported on the tumor SUVmax to liver SUVmean ratio (T/L- ratio). 16,22,37,44,46,52,53 Thresholds varied from 1.4-3.0. Pooled sensitivity and specificity were 0.93 (95% CI: 0.87-0.97) and 0.79 (95% CI: 0.70-0.86) respectively. Pooled pLR was 4.69 (2.89-7.41) and nLR was 0.09 (0.03-0.18). There was moderate heterogeneity between studies, but no source for heterogeneity was found. Five studies reported on qualitative FDG-PET analysis. 15,54–57 Pooled sensitivity and specificity were 0.94 (95% CI: 0.88-0.98) and 0.82 (95% CI: 0.71-0.91) respectively. Pooled pLR was 5.86 (3.00-11.24) and nLR was 0.07 (0.02-0.16). There was moderate heterogeneity between studies. Higher sensitivity was found in studies including a smaller sample of lesions. Higher specificity was found in studies which included symptomatic lesions only. Eleven studies reported individual patient-level data of SUVmax on 246 patients. 21,22,62,47,48,50,54,55,59–61 Highest sensitivities were found for thresholds at 3.0 and 3.5 (0.99) and highest specificity was found for a threshold at 4.5 (0.88, Table 2 ). Accuracy was not significantly different between thresholds of 3.0, 3.5, and 4.0. However, sensitivity at a threshold of 3.5 was non-significantly higher than 4.0 (0.99 vs. 0.97) and specificity was higher at 3.5 compared to 3.0 (0.75 vs. 0.61). There was substantial heterogeneity between studies ( Figure 3 & Supplementary Figure 2 ). Sensitivity was higher in studies including a larger amount of lesions and a higher proportion of MPNST ( Supplementary Table 3 ). Sensitivity was lower in studies that included symptomatic lesions only. Specificity was higher in studies including a smaller amount of lesions and symptomatic lesions only. Specificity was lower for studies including histologically proven lesions only. Liquid biopsies Three studies reported on liquid biopsies, identifying 4 potential circulating biomarkers. One study used microarray analysis to identify genes that encode putative secreted proteins in 22 patients with benign and/or malignant peripheral nerve sheath tumors. 63 They found elevated serum levels of adrenomedullin (ADM) as a potential biomarker for malignant transformation of PNST with significantly higher mean ADM concentrations in NF1 patients with MPNST compared to NF1 patients with plexiform neurofibromas only (0.24 vs. 0.18 ng/mL; p=0.03). The diagnostic accuracy was not provided. A second study found that soluble AXL (sAXL) serum levels were higher in NF1 patients with plexiform neurofibromas and MPNSTs compared to those with dermal neurofibromas only, sAXL could not differentiate MPNST from others. 64 A third study performed screening for 56 potential serum biomarkers in 104 NF1 patients (with and without MPNST) compared with 41 controls. 65 Insulin-growth factor binding protein 1 (IGFBP1) was elevated in MPNST patients and was able to discriminate them with a sensitivity of 0.90 and specificity of 0.50. Regulated upon Activation Normal T-cell Expressed and Secreted (RANTES) was also elevated and had a sensitivity of 0.90 and a specificity of 0.26 to discriminate MPNST patients.