Hepatitis C virus (HCV) infection can cause permanent liver damage and
hepatocellular carcinoma, and deaths related to HCV deaths have recently
increased. Chronic HCV infection is often undiagnosed such that the virus
remains infective and transmissible. Identifying HCV infection early is essential
for limiting its spread, but distinguishing individuals who require further HCV
tests is very challenging. Besides identifying high-risk populations, an optimal
subset of indices for routine examination is needed to identify HCV screening
candidates. Therefore, this study analyzed data from 312 randomly chosen blood
donors, including 144 anti-HCV-positive donors and 168 anti-HCV-negative donors. The HCV viral load in each sample was measured by real-time
polymerase chain reaction method. Receiver operating characteristic curves
were used to find the optimal cell blood counts and thrombopoietin
measurements for screening purposes. Correlations with values for key indices
and viral load were also determined. Strong predictors of HCV infection were
found by using receiver operating characteristics curves to analyze the optimal
subsets among red blood cells, monocytes, platelet counts, platelet large cell
ratios, and mean corpuscular hemoglobin concentrations. Sensitivity, specificity,
and area under the receiver operator characteristic curve
According to the World Health Organization, deaths from primary hepatocellular carcinoma (HCC) exceeded 1 million in 2010. The leading risk factors for HCC are hepatitis B virus (HBV) and hepatitis C virus (HCV) infections [
In a high percentage (54%–86%) of cases, HCV infection persists for many decades and ultimately causes liver cirrhosis or HCC [
A complete blood count (CBC) is one of the most commonly performed blood tests. Since it reveals peripheral blood changes, the CBC is routinely performed in health examinations, even in asymptomatic patients. However, there is no evaluation showing the screen for HCV potential infection by CBC data. The objective of this study was to identify an optimal subset of routinely obtained haematological indices that can be used to discriminate potential HCV infection cases from the general population. Further, the change of TPO levels in apparently healthy people was also examined.
Blood samples were obtained from the Kaohsiung Blood Center between January 2008 and December 2009. Before transfusion, all blood donors were required to complete a “Blood Donor Registration Form”
Anti-HCV-positive cases were identified by a Murex anti-HCV (Version 4.0) enzyme immunoassay (Abbott, South Africa) in enzyme-linked immunosorbent assay (ELISA). The following measurements were performed according to the manufacturer instructions: CBC counts were measured with a Sysmex XT-1800i autoanalyser (Sysmex, Japan), alanine aminotransferase (ALT) levels were measured with an AU7200 autoanalyser (Beckman Coulter, USA), and serum TPO levels were measured by ELISA (Quantikine; R&D Systems Europe, Oxfordshire, UK). The HCV RNA viral loads were measured by real-time PCR using the COBAS AmpliPrep/COBAS TaqMan HCV Test (Roche Molecular Systems, USA). A RIBA (Chiron RIBA HCV 3.0 Strip Immunoblot Assay, Novartis Vaccines and Diagnostics, USA) was performed to verify positive responses to anti-HCV.
Statistical analyses were performed using JMP software (Version 9.0, SAS Institute Inc., Cary, NC, USA). Chi-square test and Student’s
Table
Comparison of demographic characteristics and clinical measurements in the HCV-infected group and in the negative control group.
Variable | HCV-infected group |
Negative control group |
|
ANCOVA |
---|---|---|---|---|
Gender |
||||
Male | 83 (57.6) | 38 (22.6) | ||
Female | 61 (42.4) | 130 (77.4) | <0.001 | |
Age mean (sd) | 39.3 ± 10.8 | 37.4 ± 7.3 | 0.071 | |
WBC (×103 |
6.8 ± 1.9 | 6.0 ± 1.7 | <0.001 | 0.006 |
RBC (×106 |
4.9 ± 0.7 | 4.5 ± 0.5 | <0.001 | 0.024 |
HB (g/dl) | 14.7 ± 1.5 | 13.4 ± 1.8 | <0.001 | <0.001 |
HCT (%) | 43.0 ± 3.9 | 39.8 ± 4.4 | <0.001 | 0.001 |
MCV (fl) | 89.1 ± 8.0 | 88.1 ± 6.3 | 0.243 | 0.122 |
MCH (pg) | 30.5 ± 3.1 | 29.6 ± 2.8 | 0.008 | 0.030 |
MCHC (g/dl) | 34.2 ± 1.0 | 33.5 ± 1.3 | <0.001 | 0.005 |
PLT (×103 |
222.3 ± 58.8 | 270.5 ± 61.8 | <0.001 | <0.001 |
NEU (%) | 56.4 ± 9.0 | 58.3 ± 8.2 | 0.052 | <0.001 |
LYM (%) | 34.0 ± 8.2 | 33.1 ± 7.7 | 0.270 | 0.648 |
MONO (%) | 6.6 ± 1.8 | 5.7 ± 1.4 | <0.001 | <0.001 |
EOS (%) | 2.5 ± 1.6 | 2.4 ± 1.7 | 0.630 | 0.174 |
BAS (%) | 0.5 ± 0.3 | 0.5 ± 0.3 | 0.140 | 0.230 |
RDW-SD (fl) | 41.8 ± 2.9 | 42.6 ± 2.9 | 0.013 | 0.055 |
RDW-CV (%) | 13.3 ± 1.5 | 13.6 ± 1.5 | 0.040 | 0.034 |
PDW (fl) | 13.2 ± 2.0 | 12.1 ± 1.9 | <0.001 | <0.001 |
MPV (fl) | 10.8 ± 0.8 | 10.5 ± 0.9 | <0.001 | 0.001 |
P-LCR (%) | 31.7 ± 6.2 | 28.5 ± 6.4 | <0.001 | <0.001 |
PCT (%) | 0.2 ± 0.1 | 0.3 ± 0.1 | <0.001 | <0.001 |
NEUT (×103 |
3.9 ± 1.4 | 3.6 ± 1.3 | 0.034 | 0.090 |
LYMPH (×103 |
2.3 ± 0.7 | 1.9 ± 0.6 | <0.001 | 0.004 |
MONO (×103 |
0.4 ± 0.1 | 0.3 ± 0.1 | <0.001 | <0.001 |
EOS (×103 |
0.2 ± 0.1 | 0.1 ± 0.1 | 0.083 | 0.561 |
BAS (×103 |
0.03 ± 0.02 | 0.03 ± 0.02 | 0.463 | 0.789 |
TPO (pg/ml) | 74.4 ± 66.3 | 42.0 ± 37.9 | <0.001 | <0.001 |
ALT (IU/L) | 35.5 ± 45.0 | 23.3 ± 29.5 | 0.006 | 0.339 |
HCV: hepatitis C virus; ANCOVA: analysis of covariance; WBC: white blood cell count; RBC: red blood cell count; Hb: haemoglobin; HCT: haematocrit; MCV: mean corpuscular volume; MCH: mean corpuscular haemoglobin; MCHC: mean corpuscular haemoglobin concentration; RDW: RBC distribution width; PLT: platelet count; PCT: plateletcrit; PDW: platelet distribution width; MPV: mean platelet volume; P-LCR: platelet-large cell ratio; NEU: neutrophil; LYM: lymphocyte; MON: monocyte; EOS: eosinophil; BAS: basophil; TPO: thrombopoietin; ALT: alanine aminotransferase.
Significance level:
Table
Compared with the negative control group, the HCV-infected group showed significantly higher red blood cell counts (RBC), haemoglobin (Hb) levels, and haematocrit (HCT) levels (Table
Mean TPO levels were significantly higher in the HCV-infected group than in the negative control group (
Stepwise multiple logistic regression analysis was performed to identify haematological indices, ALT factors, and TPO factors that predict HCV infection. The HCV infection status was used as the dependent variable; haematological indices, ALT, and TPO were used as independent variables. The model revealed seven significant predictors: mean corpuscular haemoglobin concentration (MCHC), RBC, PLT, MPV, P-LCR, MONO, and TPO (Table
Results of multivariate stepwise regression analysis.
Variables | Odds Ratio | 95% CI |
|
---|---|---|---|
RBC ≥ 4.76 (×106 |
2.043 | 1.104–3.810 | 0.023 |
MCHC ≥ 33.9 (g/dl) | 2.792 | 1.532–5.189 | 0.001 |
PLT ≤ 258 (×103 |
3.124 | 1.708–5.809 | <0.001 |
MPV ≥ 10.6 (fl) | 0.532 | 0.091–2.331 | 0.437 |
P-LCR ≥ 28.9 (%) | 5.458 | 1.238–31.722 | 0.037 |
MONO ≥ 0.38 (×103 |
3.504 | 1.926–6.478 | <0.001 |
TPO ≥ 42.071 (pg/ml) | 4.673 | 2.620–8.525 | <0.001 |
Significance level:
Correlations between HCV viral load and values for RBC, MCHC, PLT, MPV, P-LCR, MONO, and TPO.
Variables |
|
|
---|---|---|
RBC (×106 |
0.279 | <0.001 |
MCHC (g/dl) | 0.217 | <0.001 |
PLT (×103 |
−0.333 | <0.001 |
P-LCR (%) | 0.194 | 0.001 |
MONO (×103 |
0.370 | <0.001 |
TPO (pg/ml) | 0.351 | <0.001 |
Significance level:
An ROC analysis was then performed to determine the optimal cut-offs for the factors in a dummy-variable logistic regression model. Binary variables were coded as 1 or 0. A stepwise procedure was used to select an optimal subset of dummy regressors and point scores. Finally, six dummy variables that showed significant differences (RBC, MCHC, PLT, P-LCR, MONO, and TPO) were identified as predictors of HCV infection and were used to construct the AUROC (range = 0.651–0.741) (Table
Prediction performance of haematological indices and TPO.
Variables | Cut-off point | AUC | Specificity | Sensitivity |
|
Score |
---|---|---|---|---|---|---|
RBC (×106 |
4.76 | 0.664 | 0.738 | 0.521 | <0.001 | 1 |
MCHC (g/dl) | 33.9 | 0.651 | 0.601 | 0.632 | <0.001 | 1 |
PLT (×103 |
258 | 0.730 | 0.583 | 0.771 | <0.001 | 1 |
P-LCR (%) | 28.9 | 0.655 | 0.577 | 0.722 | <0.001 | 1 |
MONO (×103 |
0.38 | 0.736 | 0.720 | 0.667 | <0.001 | 1 |
TPO (pg/ml) | 42.07 | 0.741 | 0.643 | 0.757 | <0.001 | 2 |
AUC: area under the curve value; for other abbreviations, see Table
This method accounted for the point score whereas the final score (range = 1–7) was the sum of the parameters. A score of 4, which was considered optimal, yielded a sensitivity of 75.6% and a specificity of 78.5%. The AUROC for HCV infection was 0.859 (
The ROC curve analysis of scores with best prognostic power for predicting HCV infection.
Since previous studies indicate that thrombocytopenia results from chronic liver disease, we speculated that a haematological comparison between a healthy blood donor and a donor with HCV might reveal the impact of HCV on PLT and TPO; an improved understanding of this impact could help determine whether a donor has HCV. This hypothesis was tested by investigating the relationships among haematological indices and TPO and HCV viral loads. The haematological indices and TPO were also evaluated in terms of predictive performance. Because of the varying consent given by the participants, the negative control group and HCV-infected group were not matched by age or gender. Therefore, ANCOVA was used to adjust the statistical analysis for age and gender.
The data analysis indicated that ALT levels were not significantly associated with HCV infection. The ALT levels were abnormal in most hepatitis patients [
In a previous study, an HCV-infected group, which also had chronic hepatitis and cirrhosis, had a mean PLT greater than
Unlike liver cirrhosis, the mechanism of decreased PLT in HCV is decreased TPO secretion [
The ROC curve analysis is widely used to measure the discriminatory power of diagnostic or prognostic tests [
Hepatitis C virus is the main risk factor for HCC in Western Europe, North America, and Asia. Almost all HCV-associated HCCs occur in patients with cirrhosis. Antiviral treatment is the only available option for preventing or delaying the occurrence of HCC in patients with chronic HCV infection. In the early stages of HCC, malignant behavior may not have a strong correlation with histological appearance. The use of improved HCV screening methods that can detect infection in early stage not only limits further spread but also reduces the overall number of chronic HCV cases and substantially reduces the incidence of HCC.
Currently, the most urgent tasks are identifying potential markers for screening or early diagnosis of HCC among high-risk individuals with chronic hepatitis C and identifying target molecules for the treatment and prevention of HCV-associated HCC. The analytical results of this study suggest that cases in which RBC, MCHC, PLT, P-LCR, and MONO exceed the optimal cut-off values require further confirmation such as by HCV NAT. The data obtained in this study can be used to improve accuracy in screening the general population for potential cases of HCV infection. To enhance the early detection of HCV infection, further studies are needed to modify and improve existing screening procedures and to develop convenient supplemental screening flowcharts. Until then, the findings of this study should be applied cautiously.
Alanine transaminase
Analysis of covariance
Complete blood count
Enzyme-linked immunosorbent assay
Hemoglobin
Hepatocellular carcinoma
Hematocrit
Hepatitis B virus
Hepatitis C virus
Mean corpuscular haemoglobin concentration
Monocyte count
Nucleic acid test
Mean platelet volume
Polymerase chain reaction
Platelet-large cell ratio
Plateletcrit
Platelet distribution width
Platelet count
Red blood cell count
Recombinant immunoblot assay
Receiver operating characteristic
Area under ROC curve
Thrombopoietin
White blood cell count.
No author has competing interests in the publication of this study.
Pei-Yu Chu contributed to study design and paper revision. Mei-Hua Tsai and Bintou Sanno-Duanda contributed to analysis, interpretation of data, and paper drafting. Kuei-Hsiang Lin performed conception and overall coordination of study and paper revision. Kuan-Tsou Lin performed data analysis and experiments. Chi-Ming Hung and Hung-Shiang Cheng carried out experiments. Yu-Chang Tyan, Ming-Hui Yang, and Shyng-Shiou Yuan contributed to interpretation of data. All authors have read and approved the final paper. Mei-Hua Tsai and Kuei-Hsiang Lin equal contributed to this work.
The authors gratefully acknowledge Dr. Ren-Chin Jang for his dedication and commitment to this collaborative study. The Statistical Analysis Laboratory, Department of Medical Research, Kaohsiung Medical University Hospital, Kaohsiung Medical University, is greatly appreciated for helpful suggestions and guidance throughout the paper revision process. This study was funded by the Taiwan Blood Services Foundation (97-01-KS), Kaohsiung Medical University (KMU-M110018, Aim for the Top 500 Universities Grant KMU-O104003, and Aim for the Top Journal Grant KMU-DT103010), Ministry of Health and Welfare (MOHW103-TD-B-111-05), and the National Science Council, Taiwan (NSC-102-2320-B-037-017). This study was approved by the Ethics Committee of the Taiwan Blood Services Foundation.