Page 39 - Read Online

Page 10 of 22 Guerriero et al. Hepatoma Res 2019;5:6 I http://dx.doi.org/10.20517/2394-5079.2018.108

miR-30e Down Serum Single Diagnosis HCC vs. HL + CLD 39 vs. 14 + 17 91.7 70.5 0.93 [139]
miR-331-3p Up Serum Single Diagnosis HCC vs. HL + BLD 103 vs. 40 + 79.6 92.6 0.89 [135]
95
miR-331-3p Up Serum Single + AFP Diagnosis HCC vs. HL + BLD 103 vs. 40 + 91.2 92.6 N/A [135]
95
miR-335 Down Serum Single Diagnosis HCC vs. HL + CH 125 vs. 125 + N/A N/A N/A [86]
125
miR-519d Up Exosomes Single Diagnosis HCC vs. LC 87 vs. 31 N/A N/A 0.82 [57]
miR-595 Up Exosomes Single Diagnosis HCC vs. LC 87 vs. 31 N/A N/A 0.92 [57]
miR-638 Down Exosomes Single Diagnosis HCC vs. HL 126 vs. 21 N/A N/A N/A [75]
miR-665 Up Exosomes Single Diagnosis HCC vs. HL 30 vs. 10 N/A N/A N/A [74]
miR-885-5p Up Serum Single Diagnosis HCC + CHB + LC 46 + 23 + 26 90.5 79.2 0.94 [142]
vs. HL vs. 24
miR-939 Up Exosomes Single Diagnosis HCC vs. LC 87 vs. 31 N/A N/A 0.84 [57]
miR-96 Up Serum Single Diagnosis HCC vs. CHB 104 vs. 100 77.9 75.3 0.83 [143]
miR-96 Up Serum Single + AFP Diagnosis HCC vs. CHB 104 vs. 100 83.6 82.4 0.88 [143]
miR-96 Up Serum Single Diagnosis HCC vs. CHB 104 vs. 100 77.9 75.3 0.80 [143]
miR-96 Up Serum Single Diagnosis HCC vs. HL 104 vs. 120 N/A N/A N/A [143]
miR-96 Up Serum Single Diagnosis HCC vs. LC 104 vs. 90 N/A N/A N/A [143]
AUC: area under the receiver-operating characteristic curve; AFP: alpha fetoprotein; BLD: benign liver disease; CH: chronic hepatitis;
CHB: chronic hepatitis B; CHC: chronic hepatitis C; CLD: chronic liver disease; DN: dysplastic nodules; HBV: hepatitis B virus; HCC:
hepatocellular carcinoma; HL: healthy liver; LC: liver cirrhosis; (T): training set; (V): validation set; N/A: not available data

two reports [45,47] , but others reported miR-21 high in plasma of HCC patients [48,49] . Analysis of miR-125-b
and miR-101 levels also displayed variances between serum and plasma: an upregulation of plasma miR-
[43]
125-b levels was reported in HCC patients in comparison with healthy controls , while downregulation
[50]
of the same miRNA was found in serum ; miR-101 levels were found high in plasma of HCC patients in
[51]
comparison with healthy controls , whereas this miRNA was found downregulated in serum of HCC
patients in comparison with healthy controls at least by two reports [52,53] . These several examples support the
concept that differences between plasma and serum are common and should be taken in consideration when
comparing results from different studies.

The use of plasma or serum is not the only source of variability of achieved results. An analysis of published
studies strongly suggests that both pre-analytical and analytical procedures can affect results. Any change
in tissue collection steps (like type of blood tubes, centrifugation strength and sample conservation) can
generate differences in miRNA levels [54,55] . Considering the different hard-to-control sources of variability, it
is not difficult to understand how uneven and sometimes even opposite results can easily derive.

In addition to technical reasons, aspects linked to experimental design can also be added to factors
responsible for heterogeneity of results. The existence of various types of control populations represents
indeed a source of variability and, in some cases, a limitation to the practical value of such results. In fact,
for identifying useful biomarkers for the early detection of HCC it is very important to compare HCC not
only with healthy controls but vs. cirrhotic patients, considering that 80%-90% of HCCs arise in this group
of high-risk patients. A paradigmatic example is miR-122, a liver-specific miRNA whose level was found
increased in serum/plasma of HCC patients in comparison with healthy patients [45,51] , but studies found
no significant differences when HCC patients were compared to cirrhotic or chronic hepatitis patients [51,56] .
These findings indicate that increased circulating miR-122 levels likely reflect liver damage rather than the
presence of an underlying HCC, indicating the importance of controls to draw conclusions. Among studies
[57]
that produced results on differential circulating miRNAs between HCC vs. cirrhotic patients, Fornari et al.
found that serum miR-939, miR-595 and miR-494 could separate cirrhotic patients with and without HCC,
[51]
performing better than AFP. Moshiri et al. showed that the combination of three plasma miRNAs, miR-
101-3p, miR-106b-3p and miR-1246, exhibited a high diagnostic accuracy in discriminating HCC from
cirrhotic patients. Combination of two of the same miRNAs, miR-101-3p and miR-106b-3p, exhibited also an
[58]
excellent diagnostic accuracy in serum of HCC vs. cirrhotic patients. Chen et al. proved that plasma miR-