BMC Nephrology (Feb 2020)

Noninvasive indices for predicting nonalcoholic fatty liver disease in patients with chronic kidney disease

  • A Reum Choe,
  • Dong-Ryeol Ryu,
  • Hwi Young Kim,
  • Hye Ah Lee,
  • Jiyoung Lim,
  • Jin Sil Kim,
  • Jeong Kyong Lee,
  • Tae Hun Kim,
  • Kwon Yoo

DOI
https://doi.org/10.1186/s12882-020-01718-8
Journal volume & issue
Vol. 21, no. 1
pp. 1 – 11

Abstract

Read online

Abstract Background Data on clinical characteristics of nonalcoholic fatty liver disease (NAFLD) in patients with chronic kidney disease (CKD) are scarce. We investigated the clinical features and risk factors of NAFLD using noninvasive serum markers in CKD patients and attempted the temporal validation of a predictive model for CKD based on NAFLD. Methods This retrospective cross-sectional study was conducted in a single tertiary center. We enrolled 819 CKD patients and evaluated the predictive performance of relevant clinical and laboratory markers for the presence of NAFLD in both derivation (data from 2011 to 2014, n = 567) and validation (data from 2015 to 2016, n = 252) groups. Results In the derivation group, NAFLD was observed in 89 patients (15.7%; mean body mass index (BMI), 24.6 kg/m2; median estimated glomerular filtration rate (eGFR), 28.0 ml/min). BMI, hemoglobin, serum alanine aminotransferase, eGFR, and triglyceride-glucose index were used to derive a prediction model for the presence of NAFLD. Using the cutoff value of 0.146, the area under the receiver operating characteristic curve (AUROC) for the prediction of NAFLD was 0.850. In the validation group, NAFLD was observed in 51 patients (20.2%; mean BMI, 25.4 kg/m2; median eGFR, 36.0 ml/min). Using the same prediction model and cutoff value, the AUROC was 0.842. NAFLD prevalence in CKD patients was comparable to that in the general population, increasing over time. Conclusions Our model using BMI, renal function, triglyceride-glucose index, serum alanine aminotransferase, and hemoglobin accurately predicted the presence of NAFLD in CKD patients.

Keywords