Fatty liver disease (FLD) is a common disease, and can be further subdivided according to its cause into either alcoholic fatty liver disease (a type of alcoholic liver disease [ALD]) or nonalcoholic fatty liver disease (NAFLD) [1,2]. ALD and NAFLD are serious threats to the health of people worldwide, with incidences rising to about 4.5%, 15% respectively in China [3,4].

In clinical practice, it is crucial to distinguishing alcohol basis from nonalcoholic basis of hepatic steatosis, as the diagnosis relates to the selection of treatment, priority for liver transplantation and organ allocation [5,6]. However, after reviewing the literature, we found few articles regarding the differentiation of ALD and NAFLD. A liver biopsy is considered the “gold standard” to establish the diagnosis, but it is an invasive procedure accompanied by certain risks and deficiencies [7] and therefore has limited clinical application. Furthermore, unreliable drinking history [8], lack of a sensitivity and specificity of single biochemical index [9], and environmental factors (e.g., obesity, diet) have yielded difficulties for the differential diagnosis of ALD and NAFLD. Dunn et al. [10] adopt adopted a multivariate analysis method to analyze the standard risk factors for NAFLD, laboratory abnormalities associated with ALD, and a series of laboratory and histological variables related to the severity of the disease. They then determined that the ratio of aspartate aminotransferase (AST)/alanine transaminase (ALT), mean corpuscular volume (MCV), body mass index (BMI), and gender were independent predictors of ALD. Finally, they developed a new diagnostic model called the ALD/NAFLD index (ANI), to distinguish ALD from NAFLD. Though γ-glutamyl transferase (GGT) has a high predictive value for ALD [11,12], it has not been incorporated into the multivariate analysis due to limited available data.

The purpose of this study was to verify the reliability of ANI as a noninvasive approach to the differential diagnosis of ALD and NAFLD, and to investigate whether ANI combined with GGT will further enhance diagnosis accuracy of diagnosis in China. The goal is to provide a reliable and convenient tool for the clinician to differentiate diagnoses of ALD and NAFLD.

In clinical practice, it is of great importance to identify whether the etiology of hepatic steatosis is alcoholic or non-alcoholic, as it relates to treatment options and priorities for liver transplantation and organ allocation [5,6]. Although some serological markers such as mitochondrial aspartate aminotransferase isoenzymes (mAST), carbohydrate-deficient transferrin (CDT), and protein kinase C ε (PKC-ε) have been used for the diagnosis of ALD or NAFLD, none of them has sufficient sensitivity and specificity [16-19]. A meta-analysis performed by Hernaez et al. [20] shows that ultrasound is an accurate, reliable imaging technique for the detection of fatty liver, but it is difficult to distinguish between steatosis and steatohepatitis and the cause of steatosis [21]. The CAP as a novel physical parameter to detect hepatic steatosis is measured on the Fibroscan (vibration-controlled transient elastography machine). Recent studies have testified its high diagnostic accuracy in detecting hepatic steatosis in patients with chronic liver disease [22,23]. In our study, the CAP values of the ALD and NAFLD group was no statistical difference (242.87 ± 28.15 vs. 254.23 ± 50.46, p = 0.091). It suggested that CAP might not be helpful in identifying causes of fatty liver.

In recent years, the ANI diagnostic model proposed by Dunn et al. [10] has high accuracy for identification of ALD and NAFLD. This model has combined the relevant parameters of risk factors and taken into account the effects of obesity and alcohol intake.

As a result of lower gastric alcohol dehydrogenase activity in females and the high proportion of body fat tissue, women are more vulnerable to the impact of alcohol hepatotoxicity, with increasing alcohol intake are more likely to develop into alcoholic disease [24]. However, ALD is more common in men, this is most likely because men’s alcohol consumption tends to be twice or more than women [25], which is consistent with our results: the incidence of ALD in male was higher than female (Table 1).

NAFLD is closely related to obesity, insulin resistance, hypertension, and dyslipidemia, and is considered a hepatic manifestation of metabolic syndrome (MetS) [26]. Moreover, a high BMI and visceral obesity are important risk factors for NAFLD [27]. In our study, the BMI of NAFLD group was significantly higher than ALD group (p < 0.001) (Table 1).

The results showed that MCV in ALD group was significantly higher than those the NAFLD group (98.84 ± 8.64 vs. 89.60 [6.40], p < 0.001) (Table 1). The main reason for alcoholics increased MCV is the direct toxic effects of alcohol on hematopoietic stem cells, followed by the reduced intake or malabsorption of vitamin B12 and folic acid [28-30].

Except for MCV, serum transaminase level also can be influenced by the toxicity of alcohol. Studies have shown that AST/ALT is an independent predictor of ALD [9,31]. The level of serum ALT is usually higher than AST in other causes of liver damage, while ALD patients often appear with AST increased mainly. This is because AST is a mitochondrial enzyme, acetaldehyde and intermediates (such as free radicals) produced during ethanol metabolism can lead to oxidative stress and lipid peroxidation, which result to serum AST increased caused by mitochondria injury. Moreover, phosphate pyridoxine deficiency caused by chronic alcoholism can further give rise to AST and ALT of liver cells reduction. Its impact on ALT is more obvious, so AST/ALT ratio is increased [32,33]. Our findings were consistent with these findings, so AST/ALT is also included into ANI scoring system alongside MCV for differentiating diagnosis of ALD and NAFLD.

GGT is a transmembrane protein present in the microsome, and its function related to cellular uptake of amino acid [34]. Serum GGT mainly from the liver of healthy person, alcohol abuse cause liver cell damage resulting microsomes GGT released into the blood leaving the elevation of its serum concentration [35,36]. Studies have shown that GGT has a high predictive value of ALD [11,12]. In our study, the serum GGT level was significantly higher in the ALD than the NAFLD group (245.0 [602.50] vs. 58.0 [41.60], p < 0.001) (Table 1).

ANI is a reliable noninvasive diagnostic index to identify ALD and NAFLD. ANI was provided with high sensitivity (87.1%) and specificity (92.5%) in diagnosis of ALD when the cut-off toke –0.22. As ANI > –0.22 confirms the alcoholic etiology, while ANI < –0.22, NAFLD should be considered (Table 2, Fig 4). Study from Dunn et al. [10] finds that ANI > 0 ALD is highly suspected, and ANI < 0 is likely to be NAFLD, which is not consistent with our results. Such differences may result from genetic and environmental factors or the complex interaction of them. According to the cut-off value of ANI, ANI greater than –0.22 was found in 87.21% patients in ALD group, while patients with ANI less than –0.22 accounted for 90.57% of the NAFLD group (Fig. 5). When the cut-off was –0.22, ANI for the diagnosis of ALD of AUROC was 0.934 (95% CI, 0.879 to 0.969), which was significantly higher than AST/ALT, MCV, and GGT (these were 0.826 [95% CI, 0.752 to 0.885], 0.814 [95% CI, 0.739 to 0.875], and 0.815 [95% CI, 0.740 to 0.876], respectively; all p < 0.001). When combined with ANI and GGT, the diagnosis efficiency of ALD is further improved with AUROC was 0.976 (95% CI, 0.934 to 0.994; p = 0.016). Meanwhile, the AUROC between AST/ALT, MCV, and GGT was not exist significant difference (all p > 0.05) (Table 2). All this indicated that ANI showed a better performance in differentiating diagnosis of ALD and NAFLD, and the ability was further improve when ANI combined with GGT.

Although ANI scoring system shows high value for differentiating ALD and NAFLD, but there are still some limitations. Thus, we should fully understand the characteristics of ANI scoring system before using it. First of all, ANI is a continuous variable, statistical analysis requires determining a threshold for comparison results, but also need to confirm the diagnosis in combination with other indicators. There is no impact of short-term reduce alcohol intake on ANI. Therefore, the differential diagnosis accuracy of ANI will not be influenced no matter whether excessive alcohol consumption occurred recently or not. Secondly, although ANI > –0.22 indicated alcoholic etiology, this does not exclude the possibility of concomitant MetS. Moreover, the use of ANI scoring system requires except for other liver diseases, since a number of indicators used to calculate the ANI may be affected by other causes. Finally, when patients have occurred cirrhosis or model for end-stage liver disease (MELD) score greater than 20, the reliability of ANI will be reduced. In addition, ANI cannot quantitative calculation of alcohol intake in patients with ALD.

The limitations of ANI scoring system make its clinical application subject to certain restrictions, but ANI is still effective as a very simple and practical tool to assist the differential diagnosis of ALD and NAFLD. Especially combined detection of serum GGT, its value is further improved. Other indicators such as mAST, CDT, or PKC-ε used concurrently with ANI scoring system may further improve its diagnostic value .This would be worth of further study.

In conclusion, our study validated the ANI scoring systems ability to identify ALD in patients with hepatic steatosis and NAFLD with high accuracy in Chinese population. When ANI > –0.22, the diagnosis of ALD is supported, when ANI < –0.22, the diagnosis was more likely to be NAFLD. We should seriously consider the above diagnosis after except for other etiology that may result in liver steatosis. Meanwhile, when GGT is combined with ANI, its accuracy of differentiating diagnosis had been further improved. Although the ANI scoring system itself has certain limitations, and still cannot replace histopathologic examination, it might be a reliable and convenient tool for the clinician to differential diagnosis of ALD and NAFLD, which can help in triaging patients for liver biopsy and deciding about candidacy for liver transplantation.