Available online at www.sciencedirect.com

Reply

Metabolism Clinical and Exp

erimental 59 (2010) E6www.metabolismjournal.com

Microalbuminuria and nonalcoholic fatty liver disease

To the Editor:

We appreciate the comments about our work, whichinvestigated the relationship between microalbuminuria andthe severity of liver histopathology in a sample ofnondiabetic patients with nonalcoholic fatty liver disease(NAFLD). An important finding of our cross-sectional studywas the independent association between microalbuminuriaand higher fibrosis scores after adjustment for age, sex, andhomeostasis model assessment of insulin resistance values.

First, the authors suspected that prediabetes could haveconfounded the observed relationship. Of the 87 patientswith NAFLD in our study, 25 were found to have prediabetesaccording to the results of the oral glucose tolerance test. Ofthe 25 NAFLD patients with prediabetes, 2 (8%) were foundto have microalbuminuria. In contrast, microalbuminuria wasfound in 12 (19.3%) of the 62 NAFLD patients withnormoglycemia. This difference did not reach statisticalsignificance according to the Fisher exact test (P = .33). Theresults of this analysis indicated that prediabetes was evenlydistributed according to the levels of microalbuminuria andwas unlikely to influence the main conclusions. Second, itwas argued that the association between microalbuminuriaand liver fibrosis could have been confounded by hyperten-sion, dyslipidemia, and other components of the metabolicsyndrome. To address these concerns, we have performed aforward stepwise regression analysis with liver fibrosisscores as the dependent variable and age, sex, body massindex, waist circumference, systolic blood pressure, diastolicblood pressure, triglycerides, high-density lipoprotein cho-lesterol, aspartate aminotransferase, alanine aminotransfer-ase, homeostasis model assessment of insulin resistance,ferritin, and microalbuminuria (as a continuous variable) aspredictors. Of note, results showed that microalbuminuria

0026-0495/$ – see front matter © 2010 Elsevier Inc. All rights reserved.

was the only independent predictor of the fibrosis score (β =.29, Pb .05) even when all these variables were forced intothe model. These results clearly indicate that our findingswere not confounded by other potential metabolic factors. Inour study, 32 patients (36.8%) with NAFLD had themetabolic syndrome. Again, the association between micro-albuminuria and liver fibrosis did not change when the modelwas adjusted for age, sex, and the metabolic syndrome (β =.27, Pb .05). Finally, the authors argued that all abnormalurinary albumin excretion test results should be confirmed in2 of 3 samples collected over a 3- to 6-month period becauseof the known day-to-day variability. Although controversystill exists regarding the type of urine specimen to be used toevaluate microalbuminuria, we acknowledge that the lack ofconfirmation of microalbuminuria over a 3- to 6-monthperiod may be a potential caveat of our study. However, asthe authors themselves state, assessment of albuminexcretion rate in timed urine collections (24 hours orovernight) is clinically valuable and remains the most directmeasure of urinary albumin excretion. Taken together, theadditional analyses clearly confirm that microalbuminuria isan independent predictor of liver fibrosis scores in ourpatients with NAFLD even after allowance for the constructof metabolic risk factor clustering.

Yusuf YilmazYesim Ozen Alahdab

Oya YonalNese ImeryuzCem Kalayci

Department of GastroenterologyMarmara University School of Medicine

Altunizade, Istanbul 34662, TurkeyE-mail address: yusufyilmaz@uludag.edu.tr

doi:10.1016/j.metabol.2010.03.005