Author
|
Country
|
Histology of pre-/malignant tissue
|
No. of pre-/malignant samples
|
No. of normal (control) samples
|
AUROC
|
Sensitivity (%)
|
Specificity (%)
|
Discriminative strengtha
|
---|
Tidy 2013
|
UK
|
High-grade CIN
|
87
|
109
|
0.827
|
86
|
56
|
Good
|
Balasubramani 2009
|
UK
|
High-grade CIN
|
85
|
681
|
0.800
|
79
|
73
|
Good
|
Brown 2000
|
UK
|
High-grade CIN
|
126
|
370
|
0.951
|
92
|
92
|
Good
|
Abdul 2006
|
UK
|
High-grade CIN
|
178
|
680
|
0.652
|
74
|
53
|
Moderate
|
Malvehy 2014
|
Multi-centre
|
Melanoma, NMSC
|
320
|
320
|
NR
|
97
|
34
|
Moderate
|
Mohr 2013
|
Multi-centre
|
Melanoma, NMSC
|
126
|
126
|
NR
|
98
|
NR
|
Insufficient
|
Aberg 2011
|
Multi-centre
|
Melanoma
|
59
|
59
|
0.850
|
95
|
49
|
Good
|
Rocha 2017
|
Multi-centre
|
Melanoma
|
6
|
154
|
NR
|
100
|
70
|
Good
|
Murdoch 2014
|
UK
|
Oral SCC
|
10
|
61
|
0.776
|
65
|
92
|
Good
|
Sarode 2015
|
India
|
Oral SCC
|
50
|
50
|
NR
|
NR
|
NR
|
Insufficient
|
- AUROC: area under receiver operating characteristic curve for discrimination between neoplasia and normal tissue; CIN: cervical intraepithelial neoplasia; NMSC: non-melanomatous skin cancer; Melanoma refers to cutaneous melanoma in all cases. SCC: squamous cell carcinoma; NR: not reported
- aDiscriminative strength is a summary of the reported quantitative indices for discrimination between neoplastic and normal tissue in each study, not taking into account the number of samples. This was rated as follows: “Good”, AUROC > 0.7 or sensitivity and specificity both > 0.75 or Youden index > 0.5; “Moderate”, AUROC > 0.6 or Sensitivity > 0.7 and Youden index > 0.25; “Insufficient”, not meeting the above criteria or insufficient data. Youden index is calculated as the (sensitivity + specificity) minus 1 [11]