“…Extending upon this, the number of studies that perform a receiver operating characteristic (ROC) analysis (plotting the relation between sensitivity and specificity) to assess the discriminative power of sHLA as a biomarker is limited. Although only partially informative due to the lack of proper non-malignant control groups, the described area under the curve (AUC) for sHLA-G to discriminate cancer patients from healthy controls was significantly different from chance (AUC = 0.5) in all reporting studies and ranged from 0.6 to 0.842 [ 33 , 35 , 36 , 39 , 44 , 50 ]. From these studies, three report on a good discriminative power of sHLA-G between malignancy and other, non-malignant diseases, although two of those mixed non-malignant control samples with healthy controls for ROC analysis [ 35 , 36 , 39 ].…”