Clinical usage of serum ferritin to assess liver fibrosis in patients with non-alcoholic fatty liver disease: Proceed with caution

Authors


  • Conflict of interest: None to declare.

Abstract

Serum ferritin was recently reported to have low diagnostic accuracy for the detection of advanced fibrosis in patients with non-alcoholic fatty liver disease (NAFLD). To corroborate these findings, we investigated the diagnostic accuracy of serum ferritin levels for detecting liver fibrosis in NAFLD patients utilizing a large Japanese cohort database. A total 1201 biopsy-proven NAFLD patients, seen between 2001 and 2013, were enrolled into the Japan Study Group of NAFLD. Analysis was performed on data from this cohort comparing between serum ferritin levels and hepatic histology. Serum ferritin increased with increasing histological grade of steatosis, lobular inflammation and ballooning. Multivariate analyses revealed that sex differences, steatotic grade and fibrotic stage were independently associated with serum ferritin levels (P < 0.0001, <0.0001, 0.0248, respectively). However, statistical analyses performed using serum ferritin levels demonstrated that the area under the receiver–operator curve for detecting fibrosis was not adequate for rigorous prediction. Several factors including sex differences, steatosis and fibrosis were found to correlate with serum ferritin levels. Therefore, serum ferritin may have low diagnostic accuracy for specifically detecting liver fibrosis in NAFLD patients due to the involvement of multiple hepatocellular processes.

Introduction

Non-alcoholic fatty liver disease (NAFLD) is an important cause of chronic liver injury in many countries around the world.[1] NAFLD represents a spectrum of conditions that are histologically characterized by macrovesicular hepatic steatosis and a diagnosis is made in patients who have not consumed alcohol in amounts sufficient to be considered harmful to the liver. The histological changes range over a wide spectrum, extending from non-alcoholic fatty liver (NAFL), which is generally non-progressive, to non-alcoholic steatohepatitis (NASH), liver cirrhosis, liver failure, and sometimes even hepatocellular carcinoma.[2] In patients with suspected NAFLD, the degree of liver fibrosis must be assessed to determine the prognosis and optimal treatment as for other liver diseases such as viral hepatitis.

Iron is considered one of the putative elements that interact with oxygen radicals to induce liver damage and fibrosis.[3] Ferritin is the primary iron-storage protein and serum ferritin concentration has historically been used to predict severe fibrosis in chronic liver diseases.[4, 5] However, Angulo et al. recently reported that serum ferritin levels have low diagnostic accuracy for the detection of advanced fibrosis in patients with NAFLD.[6] In their paper, they concluded that serum ferritin levels were significantly associated with the presence and severity of liver fibrosis, but the area under the receiver–operator curve (AUROC) was less than 0.60 for the presence of fibrosis or any stage of liver diseases. This result is clinically important, but it is controversial because serum ferritin is routinely measured in the USA and is considered to be one of several clinical indicators of NASH.[5] To build on this work, we have investigated the diagnostic accuracy of serum ferritin levels for detecting liver fibrosis in NAFLD patients utilizing the Japanese Society Group (JSG)-NAFLD database,[7] considered to be one of the largest cohorts in the world.

Methods

Patients population

A total 1201 biopsy-proven NAFLD patients, seen between 2001 and 2013, were enrolled from institutes affiliated with the JSG-NAFLD. This study group is represented by the following nine hepatology centers in Japan: Yokohama City University, Kyoto Prefectural University of Medicine, Hiroshima University, Kochi Medical School, Saga Medical School, Osaka City University, Nara City Hospital, Kurume University and Saiseikai Suita Hospital. Histological grading and staging was classified according to Brunt et al. and Kleiner et al., as previously reported.[8, 9] Presence of fibrosis, severe fibrosis and advanced fibrosis were classified as stages 1–4, 2–4 and 3–4, respectively.

Result

Using the JSG-NAFLD database, data from a total of 1201 biopsy-proven NAFLD patients was retrospectively analyzed. In this cohort, 641 patients were male and the mean age was 50.8 ± 15.0 years old. Based on our analysis, serum ferritin increased with increasing histological grade of steatosis, lobular inflammation and ballooning (P < 0.0001, 0.0215, 0.0347, respectively) (Table 1). Serum ferritin levels stratified by the fibrotic stage were as follows: stage 0, 180.6 ± 31.4 (n = 228); stage 1, 238.0 ± 24.0 (n = 389); stage 2, 332.1 ± 26.7 (n = 315); stage 3, 290.1 ± 31.8 (n = 222); and stage 4, 205.6 ± 69.1 (n = 47) (Table 1). In addition, we performed multiple regression analysis using sex differences and histopathological parameters including grade of steatosis, necroinflammation, ballooning and fibrotic stage. The sex differences, steatotic grade and fibrotic stage were significantly associated with serum ferritin levels by multiple regression analysis (Table 1). We subsequently calculated the AUROC to estimate the diagnostic performance of serum ferritin for detecting presence of fibrosis. For stage 1–4 liver diseases, it was found to be 0.617 (optimal cut-off value, 208.8 ng/mL; sensitivity, 49.2%; specificity, 69.7%; positive predictive value, 87.4%; negative predictive value, 24.3%) (Fig. 1a). The AUROC calculated to estimate the diagnostic performance of the serum ferritin for detecting severe fibrosis (stage 2–4) in NAFLD patients was 0.573 (optimal cut-off value, 295.5 ng/mL; sensitivity, 34.1%; specificity, 72.1%; positive predictive value, 59.1%; negative predictive value, 55.4%) (Fig. 1b). Finally, the AUROC for detecting advanced fibrosis (stages 3, 4) was 0.554 (optimal cut-off value, 301.0 ng/mL; sensitivity, 33.5%; specificity, 74.8%; positive predictive value, 27.7%; negative predictive value, 79.6%) (Fig. 1c).

Table 1. Serum ferritin levels and gender, histopathological grade of steatosis, lobular inflammation, ballooning hepatocyte and fibrotic stage
VariablesScorenSerum ferritin (ng/mL)Univariate analysis, P-valueMultivariate analysis, P-value
  1. aStatistically significant.
Sex     
 Male641298.4 ± 259.1<0.0001<0.0001a
 Female560216.2 ± 223.9
Steatosis     
 Grade 1502224.0 ± 190.3  
 Grade 2432268.3 ± 211.1<0.0001a<0.0001a
 Grade 3267315.1 ± 288.6  
Lobular inflammation     
 Grade 0136216.5 ± 167.6  
 Grade 1666258.6 ± 242.10.0215a0.9838
 Grade 2331273.1 ± 206.7
 Grade 368297.3 ± 190.3  
Ballooning hepatocyte     
 Grade 0200227.0 ± 154.0  
 Grade 1559255.2 ± 226.90.0347a0.9227
 Grade 2442281.7 ± 249.0  
Fibrotic stage     
 Stage 0228180.6 ± 31.4  
 Stage 1389238.0 ± 24.0  
 Stage 2315332.1 ± 26.70.0030a0.0248a
 Stage 3222290.1 ± 31.8  
 Stage 447205.6 ± 69.1  
Figure 1.

Area under the receiver–operator curve (AUROC) for detecting diagnostic accuracy based on the serum ferritin levels. (a) Presence of fibrosis (stages 1–4). (b) Severe fibrosis (stages 2–4). (c) Advanced fibrosis (Stages 3, 4).

Discussion

In this study, similar to data presented in Angulo et al., the sensitivity and positive predictive value were not high enough to predict severe and advanced fibrosis in NAFLD patients utilizing serum ferritin alone. We previously reported that serum ferritin concentration was significantly higher in patients with NASH as compared to patients with NAFL. However, we also demonstrated that the sensitivity was not high enough to rule out NASH utilizing serum ferritin alone.[4] Therefore, we developed a new scoring system that includes ferritin and two other additional clinical laboratory parameters.[10] The results presented here reconfirm that measurement of serum ferritin levels alone demonstrate low diagnostic accuracy (AUROC, <0.60) for detecting severe or advanced fibrosis even if patients have significantly high serum ferritin levels.

Ferritin is reported to be associated with systemic inflammation, and often it is associated with chronic inflammatory disease states such as diabetes and obesity.[11] Furthermore, we reported that serum ferritin is associated with visceral fat area, subcutaneous fat area and degree of fatty liver.[12] In this study, several factors such as sex differences, steatosis, inflammation and ballooning hepatocytes as well as fibrotic stage are suggested to affect the serum ferritin levels. In general, unlike viral hepatitis, NAFLD may have two aspects: steatosis and fibrosis. Therefore, in NAFLD patients, it may be difficult to assess liver fibrosis by serum ferritin levels alone. Because the incidence of NAFLD is rising rapidly in both adults and children, simple non-invasive methods for detecting fibrosis in these patients is of major clinical interest. However, we assert that because some clinicians use ferritin as a biomarker for the severity of fibrosis, they should be vigilant in its appropriate use to avoid missing subsequent progression of liver disease. In conclusion, we hope the contents of this manuscript are useful to support the cautious use of serum ferritin alone to diagnose the severity of liver disease in patients with NAFLD.

Ancillary

Advertisement