• Potential conflict of interest: Nothing to report.


We thank Prof. Dogru and colleagues for their interest in our recent article on the importance of adipocyte fatty acid binding protein (AFABP) in nonalcoholic fatty liver disease (NAFLD).1 It is now well accepted that NAFLD is the hepatic manifestation of the metabolic syndrome and as such is intimately associated with insulin resistance, visceral obesity, and dyslipidemia.2 Insulin resistance, which is a key characteristic of both conditions, has also been associated with NAFLD progression from simple steatosis to nonalcoholic steatohepatitis.3 We used the homeostasis model assessment of insulin resistance (HOMA-IR) to reflect the spectrum of insulin sensitivity. This index has been shown to correlate with the results of euglycemic-hyperinsulinemic clamp in patients without diabetes and with type 2 diabetes, including those treated with metformin and other oral hypoglycemic agents.4, 5 We were careful to exclude those taking thiazolidenediones, which have been shown to significantly affect circulating adipokine levels,6, 7 in contrast to sulfonylureas and metformin, which have not.7, 8 Type 2 diabetes eventually ensues in many subjects with increasing insulin resistance and is associated with more progressive fatty liver disease9; therefore, we decided not to exclude subjects with type 2 diabetes in our study. Although we agree it would be interesting to further subclassify subjects by glucose dysregulation status, the use of post hoc subset analysis is far less robust statistically and prone to type 2 error.

We agree that metabolic variables and, in particular, measures of insulin resistance are important to consider when interpreting data on adipocytokines, which are intimately related to these factors. Indeed, our data showed the close association between both AFABP and lipocalin-2 to insulin resistance, body mass index, and waist circumference. To ensure our findings were independent of key confounders, we performed multivariate analysis for each histological endpoint in NAFLD using those factors significant on univariate analysis.1 We demonstrated that the association between AFABP and necroinflammatory and fibrotic activity is independent of abdominal obesity (the waist-hip ratio), cholesterol, high-density lipoprotein, and insulin resistance. For further clarity, as suggested by Prof. Dogru, we provide in Table 1 the relationship between AFABP and disease severity in NAFLD, directly controlled for insulin resistance and the key metabolic variables of body mass index, low-density lipoprotein, high-density lipoprotein, triglycerides, and glucose. This conclusively demonstrates that AFABP plays an important role in the pathogenesis of NAFLD independent of metabolic confounders.

Table 1. 
FactorAdjusted for HOMA-IRBest Fitting Model*
Odds Ratio (95% CI)P ValueOdds Ratio (95% CI)P Value
  • *

    Input variables: AFABP, HOMA-IR, LDL, HDL, triglycerides, BMI, and glucose.

Lobular Inflammatory Grade    
 AFABP (per 10 ng/mL increase)1.4 (1.0–1.8)0.031.4 (1.0–1.9)0.02
 LDL (per 1 mmol/L increase)1.6 (1.1–2.5)0.03
 HOMA-IR (per 1 unit increase)1.1 (1.0–1.2)0.06
Ballooning Grade    
 AFABP (per 10 ng/mL increase)1.6 (1.2–2.1)0.011.6 (1.2–2.1)0.01
 HOMA-IR (per 1 unit increase)1.1 (1.0–1.2)0.06
Fibrosis Stage    
 AFABP (per 10ng/mL increase)1.3 (1.0–1.8)0.021.3 (1.0–1.7)0.03
 Glucose (per 1 mmol/L increase)1.3 (1.0–1.5)0.01

David van der Poorten*, Kerry-Lee Milner†, Donald J. Chisholm†, Jacob George*, * Storr Liver Unit, Westmead Millennium Institute, University of Sydney, Sydney, Australia, † Garvan Institute of Medical Research, University of New South Wales, Sydney, Australia.