Diagnostic Accuracy of APRI, AAR, FIB-4, FI, King, Lok, Forns, and FibroIndex Scores in Predicting the Presence of Esophageal Varices in Liver Cirrhosis

Aspartate aminotransferase-to-platelet ratio (APRI), aspartate aminotransferase-to-alanine aminotransferase ratio (AAR), FIB-4, FI, King, Lok, Forns, and FibroIndex scores may be simple and convenient noninvasive diagnostic tests, because they are based on the regular laboratory tests and demographic data. This study aimed to systematically evaluate their diagnostic accuracy for the prediction of varices in liver cirrhosis. All relevant papers were searched via PubMed, EMBASE, CNKI, and Wanfang databases. The area under the summary receiver operating characteristic curve (AUSROC), sensitivity, specificity, positive and negative likelihood ratio (PLR and NLR), and diagnostic odds ratio (DOR) were calculated. Overall, 12, 4, 5, 0, 0, 4, 3, and 1 paper was identified to explore the diagnostic accuracy of APRI, AAR, FIB-4, FI, King, Lok, Forns, and FibroIndex scores, respectively. The AUSROCs of APRI, AAR, FIB-4, Lok, and Forns scores for the prediction of varices were 0.6774, 0.7275, 0.7755, 0.7885, and 0.7517, respectively; and those for the prediction of large varices were 0.7278, 0.7448, 0.7095, 0.7264, and 0.6530, respectively. The diagnostic threshold effects of FIB-4 and Forns scores for the prediction of varices were statistically significant. The sensitivities/ and Xiaozhong Guo, MD, PhD specificities/PLRs/NLRs/DORs of APRI, AAR, FIB-4, Lok, and Forns scores for the prediction of large varices were 0.65/0.66/2.15/0.47/4.97, 0.68/0.58/2.07/0.54/3.93, 0.62/0.64/2.02/0.56/3.57, 0.78/0.63/2.09/ 0.37/5.55, and 0.65/0.61/1.62/0.59/2.75, respectively. APRI, AAR, FIB-4, Lok, and Forns scores had low to moderate diagnostic accuracy in predicting the presence of varices in liver cirrhosis. (Medicine 94(42):e1795) Abbreviations: AAR = aspartate aminotransferase-to-alanine aminotransferase ratio, ALT = alanine aminotransferase, APRI = aspartate aminotransferase-to-platelet ratio, AST = aspartate aminotransferase, AUC = area under curve, AUSROC = area under the summary receiver operating characteristic curve, CI = confidence interval, DOR = diagnostic odds ratio, EV = esophageal varices, FN = false negative, FP = false positive, GGT = gamma glutamyl transpeptidase, GV = gastric varices, INR = international normalized ratio, NLR = negative likelihood ratio, PLR = positive likelihood ratio, PLT = platelets, PSR = platelet count to spleen diameter ratio, SE = standard error, SS = spleen stiffness, TE = transient elastography, TN = true negative, TP = true positive. INTRODUCTION V ariceal bleeding is one of the most lethal portal hypertension-related complications in liver cirrhosis. Early diagnosis and screening of varices should be warranted to improve the prognosis of liver cirrhosis. Upper gastrointestinal endoscopy is the golden diagnostic method for varices. However, given the invasiveness and relatively high cost of endoscopy and poor patients’ adherence, noninvasive diagnostic methods have been developed dramatically in the last decades. Recently, several systematic reviews and meta-analyses have confirmed the diagnostic performances of transient elastography (TE), spleen stiffness (SS), and platelet count to spleen diameter ratio (PSR). First, a meta-analysis of 18 studies by Shi et al found that the summary sensitivity, specificity, and area under curve (AUC) of TE for the prediction of esophageal varices (EV) were 0.87, 0.53, and 0.84, respectively; and those for the prediction of large EV were 0.86, 0.59, and 0.78, respectively. Second, a meta-analysis of 12 studies by Singh et al found that the summary sensitivity and specificity of SS for the prediction of EV were 0.78 and 0.76, respectively; and those for the prediction of large EV were 0.81 and 0.66, respectively. Third, a meta-analysis of 8 studies by Chawla et al found that the summary sensitivity and specificity of PSR EV were 0.89 and 0.74, respectively. of 20 studies by Ying et al also found itivity, specificity, and AUC of PSR for www.md-journal.com | 1 TABLE 1. Formulas of Different Prediction Scores Score Formulas AST to platelets ratio index (APRI) [(AST/ULN) 100]/PLT AST-to-ALT ratio (AAR) AST/ALT FIB-4 (age AST)/(PLT ALT) FI 8–0.01 PLT-ALB King age AST INR/PLT Lok 5.56–0.0089 PLTþ 1.26 AST/ALTþ 5.27 INR Forns 7.811–3.131 ln(PLT)þ0.781 ln(GGT)þ 3.467 ln(age) 0.014 (cholesterol) FibroIndex 1.738–0.064 PLTþ 0.005 AST þ 0.463 gamma globulin ALB1⁄4 albumin, ALT1⁄4 alanine aminotransferase, AST1⁄4 aspartate aminotransferase, BIL1⁄4 bilirubin, FI1⁄4fibrosis index, FIB-41⁄4fibrosis 4 index, GGT1⁄4 gamma glutamyl transpeptidase, HE1⁄4 hepatic encephalopathy, INR1⁄4 international normalized ratio, PLT1⁄4 platelets count, Deng et al the prediction of EV were 0.92, 0.87, and 0.95, respectively. Collectively, these large meta-analyses provided the systematic evidence regarding the values of noninvasive methods for the prediction of varices. In spite of moderate to high diagnostic accuracy, they need high skills in elastography and ultrasound techniques. By comparison, aspartate aminotransferase-to-platelet ratio (APRI), aspartate aminotransferase-to-alanine aminotransferase ratio (AAR), FIB-4, FI, King, Lok, Forns, and FibroIndex scores (Table 1), which are primarily composed of regular laboratory tests and readily available demographic data, do not need any special experiences in imaging techniques. They are more convenient and cheap in clinical practices. To our knowledge, the diagnostic accuracy of APRI, ULN1⁄4 upper limit of normal. AAR, FIB-4, FI, King, Lok, Forns, and FibroIndex scores for the prediction of varices in liver cirrhosis have not been systematically evaluated. METHODS This work is registered on PROSPERO database (registration number: CRD42015017519). Because this work is a systematic review of literatures, the ethical approval and patient consent are not necessary. Literature Search All relevant papers were searched via the PubMed, EMBASE, CNKI, and Wanfang databases. PubMed and EMBASE were two major English-language databases, and CNKI and Wanfang databases were two major Chinese-language databases. As for APRI score, the search items were as follows: ((varices) AND liver cirrhosis) AND ((APRI) OR ((aspartate aminotransferase) AND platelets)). As for AAR score, the search items were as follows: ((varices) AND liver cirrhosis) AND ((AAR) OR ((aspartate aminotransferase) AND alanine aminotransferase)). 2 | www.md-journal.com As for FIB-4 score, the search items were as follows: ((varices) AND liver cirrhosis) AND ((FIB-4 score) OR ((((aspartate aminotransferase) AND alanine aminotransferase) AND platelets) AND age)). As for FI score, the search items were as follows: ((varices) AND liver cirrhosis) AND ((FI score) OR ((albumin) AND platelets)). As for King score, the search items were as follows: ((varices) AND liver cirrhosis) AND ((King score) OR ((((aspartate aminotransferase) AND international normalized ratio) AND platelets) AND age)). As for Lok score, the search items were as follows: ((varices) AND liver cirrhosis) AND ((Lok) OR ((((aspartate aminotransferase) AND alanine aminotransferase) AND platelets) AND international normalized ratio)). As for Forns score, the search items were as follows: ((varices) AND liver cirrhosis) AND ((Forns score) OR ((((gamma-glutamyl transpeptidase) AND cholesterol) AND platelets) AND age)). As for FibroIndex score, the search items were as follows: ((varices) AND liver cirrhosis) AND ((FirbroIndex) OR (((platelets) AND gamma globulin) AND aspartate aminotransferase)). The last search was performed on April 26, 2015. Reference lists were also manually searched. Medicine Volume 94, Number 42, October 2015