’Original article Comparative diagnostic accuracy of red cell distribution width-to-platelet ratio versus noninvasive fibrosis scores for the diagnosis of liver fibrosis in biopsy-proven nonalcoholic fatty liver disease Mustafa Cengiza and Seren Ozenirlerb Objective Nonalcoholic fatty liver disease (NAFLD) is a chronic liver disease and assessment of liver fibrosis is important. We aimed to investigate the performance of red cell volume distribution width-to-platelet ratio (RPR) in predicting liver fibrosis in patients with NAFLD and to compare it with well-known noninvasive predicting fibrosis scores (alanine aminotransferase ratio, aspartate aminotransferase platelet ratio index, fibrosis index, fibrosis 4, and fibrosis, cirrhosis index). Materials and methods Serum samples of consecutive biopsy-proven NAFLD patients were used to calculate the RPR index. Fibrosis stages were evaluated using the Brunt Criteria. Area under receiver operating characteristics curve was used to calculate predicting performance and compare with other noninvasive fibrosis scores. Results One hundred and twenty-three consecutive patients with biopsy-confirmed NAFLD were recruited; 54 patients (43.9%) were women. The median age of the patients was 49 years. Fibrosis scores were F0–1, F2, F3, and F4 in 79 (64.2%), 27 (22%), 11 (8.9%), and 6 (4.9%) patients, respectively. The median RPR increased as the fibrosis scores progressed: F0, 0.0524; F1, 0.0534; F2, 0.0606; F3, 0.0815; and F4 0.2022. Area under receiver operating characteristics curve of the RPR was 0.69 in predicting significant fibrosis (≥ F2), 0.81 in advanced fibrosis ( ≥ F3), and 0.85 in F4, and all were statistically significant (P 0.05). RPR was correlated with fibrosis r: 0.37, 95% confidence interval: (0.21–0.52), P < 0.001. RPR was an independent predicting factor for identifying both significant and advanced fibrosis in regression analysis (P < 0.05). Conclusion RPR was both correlated and able to predict liver fibrosis and may be suggested to reduce liver biopsy in NAFLD. Eur J Gastroenterol Hepatol 27:1293–1299 Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
Introduction
Nonalcoholic fatty liver disease (NAFLD), a global cause of chronic liver disorders, consists of simple steatosis and nonalcoholic steatohepatitis (NASH), and may progress to significant, advanced fibrosis and as well as cirrhosis [1–3]. Identification and assessment of the degree of liver fibrosis in NAFLD are worthy of attention and also liver biopsy is still the only procedure used for the diagnosis [4]. However, it is not really an ideal approach as a consequence of several potential uncomfortable side effects, for instance, typically the invasive nature of the technique, the possibility of hemorrhage, perforation, insufficient sample dimension, simple mistakes, the high cost, pain, as well as intraobserver and interobserver variability [5–7]. European Journal of Gastroenterology & Hepatology 2015, 27:1293–1299 Keywords: fibrosis, nonalcoholic fatty liver disease, noninvasive marker, red cell volume distribution width-to-platelet ratio a Department of Gastroenterology, A.Y. Ankara Oncology Training and Research Hospital and bDepartment of Gastroenterology, Gazi University Faculty of Medicine, Ankara, Turkey
Correspondence to Mustafa Cengiz, MD, Department of Gastroenterology, Dr A.Y. Ankara Oncology Training and Research Hospital, 06200 Yenimahalle, Ankara, Turkey Tel: + 90 312 3060909; fax: + 90 312 334 03 52; e-mail: [email protected] Received 15 June 2015 Accepted 2 July 2015
Numerous noninvasive fibrosis scores for predicting liver fibrosis have already been suggested. All these scores consist of readily available, appropriate, and repeatable variables in predicting models, for example, aspartate aminotransferase (AST) to alanine aminotransferase (ALT) ratio (AAR), AST to platelet ratio index (APRI), fibrosis index (FI), fibrosis, cirrhosis index (FCI), and fibrosis 4 (FIB-4) scores [8–12]. There is an inevitable requirement of noninvasive fibrosis indicators in predicting liver fibrosis. Although various markers are proposed to be correlated with liver fibrosis, sufficient results in this field are not actually available [13,14]. Considering these natural drawbacks associated with liver biopsy, a perfect test should be reproducible and simple. A new very simple, easy to calculate, noninvasive clinical prediction rule, red cell volume distribution width (RDW)-to-platelet ratio (RPR), with respect to predicting the actual extent of liver fibrosis, along with good overall performance in individuals with chronic hepatitis B, has already been suggested [15,16]. Nevertheless, there are actually no data on the overall performance of this method in the evaluation of liver fibrosis in patients with biopsy-confirmed NAFLD and also this index has not been validated externally as yet. We proposed to investigate whether or not the RPR index is correlated with and capable of predicting liver fibrosis stages and as well as to compare the diagnostic
0954-691X Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
DOI: 10.1097/MEG.0000000000000445
Copyright r 2015 Wolters Kluwer Health, Inc. All rights reserved.
1293
1294
November 2015 • Volume 27 • Number 11
European Journal of Gastroenterology & Hepatology
accuracy of RPR in predicting liver fibrosis with other widely accepted noninvasive fibrosis scores such as AAR, APRI, FI, FIB-4, and FCI utilized for the particular detection of the existence and degree of liver fibrosis in patients with biopsy-proven NAFLD. Materials and methods Patients
All consecutive patients older than 18 years of age with a suspicion of NAFLD on the basis of persistently elevated serum aminotransferase levels and imaging studies that showed a diffusely hyperechogenic liver associated with fatty liver, confirmed with a final diagnosis of NAFLD upon liver biopsy, were included in the study. Individuals who had a diagnosis of viral hepatitis, primary biliary cirrhosis, sclerosing cholangitis, autoimmune hepatitis, hemochromatosis, malignancy, Wilson’s disease, alpha-1-antitrypsin deficiency, and drug-induced liver disease were excluded from the research. Other exclusion criteria were the consumption of alcohol more than 20 g/day for men and 10 g/day for women, the presence of infectious diseases on admission, chronic renal diseases, significant cardiovascular disorders, collagens diseases, hematological disorders, including iron deficiency, deficiency of B12 or folate, hemoglobinopathies, presence of blood transfusions, bone marrow depression, and use of anticoagulant drugs, NSAIDs, and hepatotoxic drugs that might impair red cell production and increase red cell destruction. These conditions were eliminated depending on the evidence-based laboratory, clinic, and histological criteria. Patients who had diabetes mellitus were on oral antidiabetics and/or insulin therapy; those who had hypertension were all using angiotensin-converting enzyme inhibitors and their hypertension was under control. Laboratory data
Demographical, medical, and laboratory records were collected and registered in a computer system database by a clinician in a blinded manner. Dividing weight in kilograms (kg) by the square of height in meters (m) as (kg/m2) yielded the BMI. On the day of the liver biopsy, venous blood samples were obtained from all patients after fasting for at least 8–12 h between 8.30 and 10.00 a.m. A complete blood count (CBC) containing RDW, hemoglobin, white blood cell, neutrophils, lymphocytes, and platelets was performed using a Beckman Coulter Gen-S automated analyzer (High Wycombe, Buckinghamshire, UK). RDW was calculated as dividing the standard deviation (SD) by the mean corpuscle volume (MCV) and then multiplying that result by 100 (RDW = SD/MCV × 100%).The SD represented the volume of erythrocytes or red blood cells that were in the blood smear. Roche Modular System auto analyzer (Roche Cobas Integra, Indianapolis, Indiana, USA) was used to assess albumin, ALT, AST, gammaglutamyl aminotransferase, and alkaline phosphatase levels. The RPR index was calculated using the following equation: RPR = RDW (%)/platelet count (109/l) [15]. Frequently known noninvasive clinical prediction rules in predicting the degree of liver fibrosis, AAR, APRI, FCI, FI,
and FIB-4, were calculated as in previously reported studies [8,12,17,18]. Liver histopathologic evaluation
Percutaneous liver biopsy was performed by an experienced clinician using a 16 G nonreusable needle. All liver biopsy examples containing a minimum of 12 complete portal tracts and also longer than 20–25 mm were considered for the study. Liver tissue specimens were evaluated by the same skilled hepatopathologist in a blinded manner. Hematoxylin and eosin and Masson trichrome stains were used for histopathological evaluation of formalin-fixed and paraffin-embedded tissues. The diagnosis was evaluated on the basis of Brunt’s criteria [19]. Histological features were graded depending on the NAFLD scoring system recommended by the National Institute of Diabetes and Digestive and Kidney Diseases NASH Clinical Research Network [20]. Steatosis was graded from 0 to 3 at 5%, ≥ 33–66%, and ≥ 66%, respectively. Lobular inflammation was scaled at 0 = no; 1 = ≤ 2 foci; 2 = 2–4 foci; and 3 = ≥ 4 foci and ballooning was graded as 0 = none; 1 = few ballooning cells; and 2 = many ballooning cells, respectively. Depending on the recommendations of Brunt’s criteria, steatosis (0–3), lobular inflammation (0–3), and ballooning (0–2) were then combined to establish the NAFLD activity score (0–8). Fibrosis was classified as 0 = no fibrosis; 1 = periportal or perisinusoidal fibrosis; 2 = perisinusoidal and portal/periportal fibrosis; 3 = bridging fibrosis; and 4 = cirrhosis. These pathologic protocols for liver fibrosis may lack data evaluating their interobserver variability and their inability to predict fibrosis-related mortality. The liver biopsy does not clearly reflect the fibrotic changes occurring in the entire liver. The liver fibrosis process is not linear, and biopsies from different areas may show different stages of fibrosis. In early liver fibrosis, liver function tests may be normal and the patients may not have any complaint. Patients were divided into significant or advanced fibrosis groups depending on the fibrosis scores. The significant fibrosis group included patients with fibrosis scores ranging from 2 to 4 (≥ F2) and the advanced fibrosis group included patients with scores 3–4 ( ≥ F3). Ethics
The research was basically carried out within the agreement of ethical recommendations of the 1975 Declaration of Helsinki that was modified in 2008 and was also approved by the Institutional Review Board. Statistical analysis
All statistical analyses were carried out using SPSS, version 21 (SPSS Inc., Chicago, Illinois, USA) and MedCalc, version 14 (MedCalc, Mariekerke, Belgium). The normality of variables was investigated using the Kolmogorov–Smirnov/ Shapiro–Wilk tests properly. Variables with a normal distribution were represented as mean ± SD, those with a nonnormal distribution and ordinal variables were described as median ± SE or interquartile range. Spearman rank correlation coefficients were determined to evaluate the correlations between RPR and histopathological, clinical, and laboratory features of patients with NAFLD. Determination
Copyright r 2015 Wolters Kluwer Health, Inc. All rights reserved.
Noninvasive marker and liver fibrosis Cengiz and Ozenirler
Results
One hundred and twenty-three consecutive biopsyconfirmed NAFLD patients were enrolled in this study. The median age of the patients was 49 years (18–72), also, 54 (43.9%) patients were women. The fibrosis stages in the study were F0–1 in 79 (64.2%), F2 in 27 (22%), F3 in 11 (8.9%), and F4 in 6 (4.9%) patients. The median RDW and ± SE was 13.84 ±0.16 and for platelet, it was 230.07 ± 76.84. The median BMI of the patients was 29.55 ± 0.58. The median levels of ALT and AST were 53 ± 5.8 and 38.5 ± 3.6, respectively. The median value of RPR was 0.0569 ± 0.0048 among the patients. The typically primary demographic, medical, and histopathological properties of the cohort population are summarized and presented in Table 1. The median values of the RPR in F0 to F4 were 0.0524, 0.0534, 0.0606, 0.0815, and 0.2022, respectively. The results showed that the median values of RPR were progressively increased as the fibrosis scores enhanced, and for all comparisons P < 0.001. The median results for RPR and other noninvasive fibrosis scoring systems (AAR, APRI, FI, FCI, and FIB-4) are presented in Table 2. The comparison of RPR values of fibrosis stages is presented in Fig. 1. The RPR index was correlated with fibrosis scores as r: 0.323, 95% confidence interval (CI) (0.21–0.52), and P < 0.001. This index was also correlated with albumin level, age, total bilirubin, international normalized ratio, and ballooning feature of histopathological evaluation (P < 0.05). Correlations of RPR with clinical and histopathological features of patients with biopsy-proven NAFLD are shown in Table 3. The correlation of RPR with fibrosis stages is also shown (Fig. 2). We also investigated whether RPR is an independent predicting factor for fibrosis using univariate logistic regression analysis in the identification of ≥ F2 and ≥ F3. We found that the RPR index was an independent predicting factor in identifying the severity of liver fibrosis using 1000 sample bootstrapping models. For ≥ F2 B (coefficient): 11.676,
1295
Table 1. Baseline characteristics of the study population (n = 123) Factors
Values
Age (years) (minimum–maximum) Female sex [n (%)] BMI (kg/m2) Red cell volume distribution width (%) Platelet counta (109/l) Prothrombin time, INR Total bilirubin (mg/dl) Serum albumin (g/dl) Aspartate aminotransferase (IU/l) Alanine aminotransferase (IU/l) Steatosis Inflammation Ballooning NAS Fibrosis stage [n (%)] F0–1 F2 F3 F4 Noninvasive models RPR AAR APRI FCI FI FIB-4
49 (18–72) 54 (43.9) 29.55 ± 0.58 13.84 ± 0.16 230.07 ± 76.84 0.97 ± 0.08 0.63 ± 0.08 4.4 ± 0.05 38.5 ± 3.6 53 ± 5.8 2.33 ± 0.74 1.45 ± 0.67 1.35 ± 0.56 5.15 ± 1.43 79 (64.2) 27 (22) 11 (8.9) 6 (4.9) 0.0569 ± 0.0048 0.707 ± 0.046 0.393 ± 0.054 0.057 ± 0.181 1.245 ± 0.099 0.935 ± 0.201
Results are determined as median ± SE. AAR, alanine aminotransferase ratio; APRI, aspartate aminotransferase platelet ratio index; FCI, fibrosis, cirrhosis index; FI, fibrosis index; FIB-4, fibrosis 4; INR, international normalized ratio; NAS, nonalcoholic fatty liver disease activity score; RPR, red cell volume distribution width-to-platelet ratio. a Results are represented as mean ± SD.
Table 2. Median values of noninvasive fibrosis scores depending on fibrosis stages Methods RPR AAR APRI FCI FI FIB-4
F0
F1
F2
F3
≥ F2
≥ F3
F4
0.0524 0.750 0.2530 0.0715 1.1250 0.7683
0.0534 0.6512 0.3931 0.0454 1.0240 0.8939
0.0606 0.6429 0.3711 0.0534 1.3300 1.1732
0.0815 1.1579 0.7746 0.1148 2.2200 2.7259
0.0673 0.8903 0.5044 0.0721 1.5775 1.6174
0.1102 1.1579 0.9507 0.1722 2.5300 2.7836
0.2022 1.2656 1.2012 0.2352 2.9450 3.0694
AAR, alanine aminotransferase ratio; APRI, aspartate aminotransferase platelet ratio index; FCI, fibrosis, cirrhosis index; FI, fibrosis index; FIB-4, fibrosis 4; RPR, red cell volume distribution width-to-platelet ratio.
0.4000 0.3000 RPR
of whether or not RPR is an independent predictive factor in identifying the presence of significant and advanced liver fibrosis scores was carried out using univariate logistic regression analysis. The abilities of the RPR index and other well-recognized liver fibrosis-predicting models in detecting fibrosis scores were assessed by area under receiver operating characteristics (AUROC) curve evaluation. The level that had the best sensitivity and specificity in AUROCs (Youden Index) analysis was considered the best cut-off value. After establishing the best cut-off values for RPR in the identification of the significant and advanced fibrosis groups AUROCs, SE, sensitivity, specificity, positive predictive value, and negative predictive value were calculated. The AUROCs produced for RPR in predicting stages of liver fibrosis were created to choose the final method. AUROCs for RPR and other well-known noninvasive scoring methods such as AAR, APRI, FCI, FI, and FIB-4 were compared with one another by multiple comparisons of AUROCs using the Hanley–McNeil methods [21]. P less than 0.05 was considered statistically significant for all analyses.
www.eurojgh.com
∗ ∗
0.2000 0.1000
∗ ∗
∗ ∗
0.0000 0
1
2 3 Fibrosis stages
4
Fig. 1. Comparison of RPR values of fibrosis stages. RPR, red cell volume distribution width-to-platelet ratio.
95% CI (4.35–36.29) and P = 0.034 and for ≥ F3 B: 13.9, 95% CI: (5.82–40.19), and P = 0.005. According to the diagnostic overall performance of the RPR index, AUROC for significant fibrosis (≥ F2) was 0.685; 95% CI: (0.579–0.785), and in advanced fibrosis
Copyright r 2015 Wolters Kluwer Health, Inc. All rights reserved.
1296
November 2015 • Volume 27 • Number 11
European Journal of Gastroenterology & Hepatology
Discussion
Table 3. Correlation of RPR with clinical and histopathological parameters of NAFLD patients Spearman P
95% CI
P
0.323 0.27 − 0.11 0.04 0.19 0.12 − 0.317 0.407 0.13 0.428 − 0.077 0.110 0.189
0.21–0.52 0.09–0.43 − 0.28 to 0.07 − 0.14 to 0.21 0.01–0.36 − 0.06 to 0.29 − 0.468 to − 0.148 0.24–0.541 − 0.142 to 0.385 0.27–0.56 − 0.251 to 0.101 − 0.07 to 0.28 0.01–0.35
< 0.001 0.002 0.24 0.66 0.034 0.18 0.0004 < 0.001 0.34 < 0.001 0.395 0.227 0.037
Fibrosis score Age ALT level AST level ALP level GGT level Albumin level Total bilirubin BMI INR Steatosis Inflammation Ballooning
ALP, alkaline phosphatase; ALT, alanine aminotransferase; AST, aspartate aminotransferase; CI, confidence interval; GGT, gamma-glutamyl aminotransferase; INR, international normalized ratio; NAFLD, nonalcoholic fatty liver disease; RPR, red cell distribution width-to-platelet ratio.
0.4000
:0.323, P < 0.001
RPR
0.3000 0.2000 0.1000 0.0000 0
1
2
3
4
Fibrosis scores Fig. 2. Correlation of RPR with fibrosis stages. RPR, red cell volume distribution width-to-platelet ratio.
( ≥ F3), it was 0.807, 95% CI (0.679–0.932) and P
Introduction
Nonalcoholic fatty liver disease (NAFLD), a global cause of chronic liver disorders, consists of simple steatosis and nonalcoholic steatohepatitis (NASH), and may progress to significant, advanced fibrosis and as well as cirrhosis [1–3]. Identification and assessment of the degree of liver fibrosis in NAFLD are worthy of attention and also liver biopsy is still the only procedure used for the diagnosis [4]. However, it is not really an ideal approach as a consequence of several potential uncomfortable side effects, for instance, typically the invasive nature of the technique, the possibility of hemorrhage, perforation, insufficient sample dimension, simple mistakes, the high cost, pain, as well as intraobserver and interobserver variability [5–7]. European Journal of Gastroenterology & Hepatology 2015, 27:1293–1299 Keywords: fibrosis, nonalcoholic fatty liver disease, noninvasive marker, red cell volume distribution width-to-platelet ratio a Department of Gastroenterology, A.Y. Ankara Oncology Training and Research Hospital and bDepartment of Gastroenterology, Gazi University Faculty of Medicine, Ankara, Turkey
Correspondence to Mustafa Cengiz, MD, Department of Gastroenterology, Dr A.Y. Ankara Oncology Training and Research Hospital, 06200 Yenimahalle, Ankara, Turkey Tel: + 90 312 3060909; fax: + 90 312 334 03 52; e-mail: [email protected] Received 15 June 2015 Accepted 2 July 2015
Numerous noninvasive fibrosis scores for predicting liver fibrosis have already been suggested. All these scores consist of readily available, appropriate, and repeatable variables in predicting models, for example, aspartate aminotransferase (AST) to alanine aminotransferase (ALT) ratio (AAR), AST to platelet ratio index (APRI), fibrosis index (FI), fibrosis, cirrhosis index (FCI), and fibrosis 4 (FIB-4) scores [8–12]. There is an inevitable requirement of noninvasive fibrosis indicators in predicting liver fibrosis. Although various markers are proposed to be correlated with liver fibrosis, sufficient results in this field are not actually available [13,14]. Considering these natural drawbacks associated with liver biopsy, a perfect test should be reproducible and simple. A new very simple, easy to calculate, noninvasive clinical prediction rule, red cell volume distribution width (RDW)-to-platelet ratio (RPR), with respect to predicting the actual extent of liver fibrosis, along with good overall performance in individuals with chronic hepatitis B, has already been suggested [15,16]. Nevertheless, there are actually no data on the overall performance of this method in the evaluation of liver fibrosis in patients with biopsy-confirmed NAFLD and also this index has not been validated externally as yet. We proposed to investigate whether or not the RPR index is correlated with and capable of predicting liver fibrosis stages and as well as to compare the diagnostic
0954-691X Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
DOI: 10.1097/MEG.0000000000000445
Copyright r 2015 Wolters Kluwer Health, Inc. All rights reserved.
1293
1294
November 2015 • Volume 27 • Number 11
European Journal of Gastroenterology & Hepatology
accuracy of RPR in predicting liver fibrosis with other widely accepted noninvasive fibrosis scores such as AAR, APRI, FI, FIB-4, and FCI utilized for the particular detection of the existence and degree of liver fibrosis in patients with biopsy-proven NAFLD. Materials and methods Patients
All consecutive patients older than 18 years of age with a suspicion of NAFLD on the basis of persistently elevated serum aminotransferase levels and imaging studies that showed a diffusely hyperechogenic liver associated with fatty liver, confirmed with a final diagnosis of NAFLD upon liver biopsy, were included in the study. Individuals who had a diagnosis of viral hepatitis, primary biliary cirrhosis, sclerosing cholangitis, autoimmune hepatitis, hemochromatosis, malignancy, Wilson’s disease, alpha-1-antitrypsin deficiency, and drug-induced liver disease were excluded from the research. Other exclusion criteria were the consumption of alcohol more than 20 g/day for men and 10 g/day for women, the presence of infectious diseases on admission, chronic renal diseases, significant cardiovascular disorders, collagens diseases, hematological disorders, including iron deficiency, deficiency of B12 or folate, hemoglobinopathies, presence of blood transfusions, bone marrow depression, and use of anticoagulant drugs, NSAIDs, and hepatotoxic drugs that might impair red cell production and increase red cell destruction. These conditions were eliminated depending on the evidence-based laboratory, clinic, and histological criteria. Patients who had diabetes mellitus were on oral antidiabetics and/or insulin therapy; those who had hypertension were all using angiotensin-converting enzyme inhibitors and their hypertension was under control. Laboratory data
Demographical, medical, and laboratory records were collected and registered in a computer system database by a clinician in a blinded manner. Dividing weight in kilograms (kg) by the square of height in meters (m) as (kg/m2) yielded the BMI. On the day of the liver biopsy, venous blood samples were obtained from all patients after fasting for at least 8–12 h between 8.30 and 10.00 a.m. A complete blood count (CBC) containing RDW, hemoglobin, white blood cell, neutrophils, lymphocytes, and platelets was performed using a Beckman Coulter Gen-S automated analyzer (High Wycombe, Buckinghamshire, UK). RDW was calculated as dividing the standard deviation (SD) by the mean corpuscle volume (MCV) and then multiplying that result by 100 (RDW = SD/MCV × 100%).The SD represented the volume of erythrocytes or red blood cells that were in the blood smear. Roche Modular System auto analyzer (Roche Cobas Integra, Indianapolis, Indiana, USA) was used to assess albumin, ALT, AST, gammaglutamyl aminotransferase, and alkaline phosphatase levels. The RPR index was calculated using the following equation: RPR = RDW (%)/platelet count (109/l) [15]. Frequently known noninvasive clinical prediction rules in predicting the degree of liver fibrosis, AAR, APRI, FCI, FI,
and FIB-4, were calculated as in previously reported studies [8,12,17,18]. Liver histopathologic evaluation
Percutaneous liver biopsy was performed by an experienced clinician using a 16 G nonreusable needle. All liver biopsy examples containing a minimum of 12 complete portal tracts and also longer than 20–25 mm were considered for the study. Liver tissue specimens were evaluated by the same skilled hepatopathologist in a blinded manner. Hematoxylin and eosin and Masson trichrome stains were used for histopathological evaluation of formalin-fixed and paraffin-embedded tissues. The diagnosis was evaluated on the basis of Brunt’s criteria [19]. Histological features were graded depending on the NAFLD scoring system recommended by the National Institute of Diabetes and Digestive and Kidney Diseases NASH Clinical Research Network [20]. Steatosis was graded from 0 to 3 at 5%, ≥ 33–66%, and ≥ 66%, respectively. Lobular inflammation was scaled at 0 = no; 1 = ≤ 2 foci; 2 = 2–4 foci; and 3 = ≥ 4 foci and ballooning was graded as 0 = none; 1 = few ballooning cells; and 2 = many ballooning cells, respectively. Depending on the recommendations of Brunt’s criteria, steatosis (0–3), lobular inflammation (0–3), and ballooning (0–2) were then combined to establish the NAFLD activity score (0–8). Fibrosis was classified as 0 = no fibrosis; 1 = periportal or perisinusoidal fibrosis; 2 = perisinusoidal and portal/periportal fibrosis; 3 = bridging fibrosis; and 4 = cirrhosis. These pathologic protocols for liver fibrosis may lack data evaluating their interobserver variability and their inability to predict fibrosis-related mortality. The liver biopsy does not clearly reflect the fibrotic changes occurring in the entire liver. The liver fibrosis process is not linear, and biopsies from different areas may show different stages of fibrosis. In early liver fibrosis, liver function tests may be normal and the patients may not have any complaint. Patients were divided into significant or advanced fibrosis groups depending on the fibrosis scores. The significant fibrosis group included patients with fibrosis scores ranging from 2 to 4 (≥ F2) and the advanced fibrosis group included patients with scores 3–4 ( ≥ F3). Ethics
The research was basically carried out within the agreement of ethical recommendations of the 1975 Declaration of Helsinki that was modified in 2008 and was also approved by the Institutional Review Board. Statistical analysis
All statistical analyses were carried out using SPSS, version 21 (SPSS Inc., Chicago, Illinois, USA) and MedCalc, version 14 (MedCalc, Mariekerke, Belgium). The normality of variables was investigated using the Kolmogorov–Smirnov/ Shapiro–Wilk tests properly. Variables with a normal distribution were represented as mean ± SD, those with a nonnormal distribution and ordinal variables were described as median ± SE or interquartile range. Spearman rank correlation coefficients were determined to evaluate the correlations between RPR and histopathological, clinical, and laboratory features of patients with NAFLD. Determination
Copyright r 2015 Wolters Kluwer Health, Inc. All rights reserved.
Noninvasive marker and liver fibrosis Cengiz and Ozenirler
Results
One hundred and twenty-three consecutive biopsyconfirmed NAFLD patients were enrolled in this study. The median age of the patients was 49 years (18–72), also, 54 (43.9%) patients were women. The fibrosis stages in the study were F0–1 in 79 (64.2%), F2 in 27 (22%), F3 in 11 (8.9%), and F4 in 6 (4.9%) patients. The median RDW and ± SE was 13.84 ±0.16 and for platelet, it was 230.07 ± 76.84. The median BMI of the patients was 29.55 ± 0.58. The median levels of ALT and AST were 53 ± 5.8 and 38.5 ± 3.6, respectively. The median value of RPR was 0.0569 ± 0.0048 among the patients. The typically primary demographic, medical, and histopathological properties of the cohort population are summarized and presented in Table 1. The median values of the RPR in F0 to F4 were 0.0524, 0.0534, 0.0606, 0.0815, and 0.2022, respectively. The results showed that the median values of RPR were progressively increased as the fibrosis scores enhanced, and for all comparisons P < 0.001. The median results for RPR and other noninvasive fibrosis scoring systems (AAR, APRI, FI, FCI, and FIB-4) are presented in Table 2. The comparison of RPR values of fibrosis stages is presented in Fig. 1. The RPR index was correlated with fibrosis scores as r: 0.323, 95% confidence interval (CI) (0.21–0.52), and P < 0.001. This index was also correlated with albumin level, age, total bilirubin, international normalized ratio, and ballooning feature of histopathological evaluation (P < 0.05). Correlations of RPR with clinical and histopathological features of patients with biopsy-proven NAFLD are shown in Table 3. The correlation of RPR with fibrosis stages is also shown (Fig. 2). We also investigated whether RPR is an independent predicting factor for fibrosis using univariate logistic regression analysis in the identification of ≥ F2 and ≥ F3. We found that the RPR index was an independent predicting factor in identifying the severity of liver fibrosis using 1000 sample bootstrapping models. For ≥ F2 B (coefficient): 11.676,
1295
Table 1. Baseline characteristics of the study population (n = 123) Factors
Values
Age (years) (minimum–maximum) Female sex [n (%)] BMI (kg/m2) Red cell volume distribution width (%) Platelet counta (109/l) Prothrombin time, INR Total bilirubin (mg/dl) Serum albumin (g/dl) Aspartate aminotransferase (IU/l) Alanine aminotransferase (IU/l) Steatosis Inflammation Ballooning NAS Fibrosis stage [n (%)] F0–1 F2 F3 F4 Noninvasive models RPR AAR APRI FCI FI FIB-4
49 (18–72) 54 (43.9) 29.55 ± 0.58 13.84 ± 0.16 230.07 ± 76.84 0.97 ± 0.08 0.63 ± 0.08 4.4 ± 0.05 38.5 ± 3.6 53 ± 5.8 2.33 ± 0.74 1.45 ± 0.67 1.35 ± 0.56 5.15 ± 1.43 79 (64.2) 27 (22) 11 (8.9) 6 (4.9) 0.0569 ± 0.0048 0.707 ± 0.046 0.393 ± 0.054 0.057 ± 0.181 1.245 ± 0.099 0.935 ± 0.201
Results are determined as median ± SE. AAR, alanine aminotransferase ratio; APRI, aspartate aminotransferase platelet ratio index; FCI, fibrosis, cirrhosis index; FI, fibrosis index; FIB-4, fibrosis 4; INR, international normalized ratio; NAS, nonalcoholic fatty liver disease activity score; RPR, red cell volume distribution width-to-platelet ratio. a Results are represented as mean ± SD.
Table 2. Median values of noninvasive fibrosis scores depending on fibrosis stages Methods RPR AAR APRI FCI FI FIB-4
F0
F1
F2
F3
≥ F2
≥ F3
F4
0.0524 0.750 0.2530 0.0715 1.1250 0.7683
0.0534 0.6512 0.3931 0.0454 1.0240 0.8939
0.0606 0.6429 0.3711 0.0534 1.3300 1.1732
0.0815 1.1579 0.7746 0.1148 2.2200 2.7259
0.0673 0.8903 0.5044 0.0721 1.5775 1.6174
0.1102 1.1579 0.9507 0.1722 2.5300 2.7836
0.2022 1.2656 1.2012 0.2352 2.9450 3.0694
AAR, alanine aminotransferase ratio; APRI, aspartate aminotransferase platelet ratio index; FCI, fibrosis, cirrhosis index; FI, fibrosis index; FIB-4, fibrosis 4; RPR, red cell volume distribution width-to-platelet ratio.
0.4000 0.3000 RPR
of whether or not RPR is an independent predictive factor in identifying the presence of significant and advanced liver fibrosis scores was carried out using univariate logistic regression analysis. The abilities of the RPR index and other well-recognized liver fibrosis-predicting models in detecting fibrosis scores were assessed by area under receiver operating characteristics (AUROC) curve evaluation. The level that had the best sensitivity and specificity in AUROCs (Youden Index) analysis was considered the best cut-off value. After establishing the best cut-off values for RPR in the identification of the significant and advanced fibrosis groups AUROCs, SE, sensitivity, specificity, positive predictive value, and negative predictive value were calculated. The AUROCs produced for RPR in predicting stages of liver fibrosis were created to choose the final method. AUROCs for RPR and other well-known noninvasive scoring methods such as AAR, APRI, FCI, FI, and FIB-4 were compared with one another by multiple comparisons of AUROCs using the Hanley–McNeil methods [21]. P less than 0.05 was considered statistically significant for all analyses.
www.eurojgh.com
∗ ∗
0.2000 0.1000
∗ ∗
∗ ∗
0.0000 0
1
2 3 Fibrosis stages
4
Fig. 1. Comparison of RPR values of fibrosis stages. RPR, red cell volume distribution width-to-platelet ratio.
95% CI (4.35–36.29) and P = 0.034 and for ≥ F3 B: 13.9, 95% CI: (5.82–40.19), and P = 0.005. According to the diagnostic overall performance of the RPR index, AUROC for significant fibrosis (≥ F2) was 0.685; 95% CI: (0.579–0.785), and in advanced fibrosis
Copyright r 2015 Wolters Kluwer Health, Inc. All rights reserved.
1296
November 2015 • Volume 27 • Number 11
European Journal of Gastroenterology & Hepatology
Discussion
Table 3. Correlation of RPR with clinical and histopathological parameters of NAFLD patients Spearman P
95% CI
P
0.323 0.27 − 0.11 0.04 0.19 0.12 − 0.317 0.407 0.13 0.428 − 0.077 0.110 0.189
0.21–0.52 0.09–0.43 − 0.28 to 0.07 − 0.14 to 0.21 0.01–0.36 − 0.06 to 0.29 − 0.468 to − 0.148 0.24–0.541 − 0.142 to 0.385 0.27–0.56 − 0.251 to 0.101 − 0.07 to 0.28 0.01–0.35
< 0.001 0.002 0.24 0.66 0.034 0.18 0.0004 < 0.001 0.34 < 0.001 0.395 0.227 0.037
Fibrosis score Age ALT level AST level ALP level GGT level Albumin level Total bilirubin BMI INR Steatosis Inflammation Ballooning
ALP, alkaline phosphatase; ALT, alanine aminotransferase; AST, aspartate aminotransferase; CI, confidence interval; GGT, gamma-glutamyl aminotransferase; INR, international normalized ratio; NAFLD, nonalcoholic fatty liver disease; RPR, red cell distribution width-to-platelet ratio.
0.4000
:0.323, P < 0.001
RPR
0.3000 0.2000 0.1000 0.0000 0
1
2
3
4
Fibrosis scores Fig. 2. Correlation of RPR with fibrosis stages. RPR, red cell volume distribution width-to-platelet ratio.
( ≥ F3), it was 0.807, 95% CI (0.679–0.932) and P
