ABSTRACT
Liver fibrosis may be assessed noninvasively with transient electrography (TE). Data on the performance of TE for detecting liver fibrosis in sub-Saharan Africa are limited. We evaluated the diagnostic accuracy of TE by performing liver biopsies on persons with liver fibrosis indicated by TE. We enrolled HIV-infected and HIV-uninfected participants with TE scores consistent with at least minimal disease (liver stiffness measurement [LSM]≥7.1 kPa). Biopsies were performed and staged using the Ishak scoring system. A concordant result was defined using accepted thresholds for significant fibrosis by TE (LSM ≥ 9.3 kPa) and liver biopsy (Ishak score ≥ 2). We used modified Poisson regression methods to quantify the univariate and adjusted prevalence risk ratios (PRR) of the association between covariates and the concordance status of TE and liver biopsy in defining the presence of liver fibrosis. Of 131 participants with valid liver biopsy and TE data, only 5 participants (3.8%) had Ishak score ≥ 2 of whom 4 had LSM ≥ 9.3 kPa (sensitivity = 80%); of the 126 (96.2%) with Ishak score < 2, 76 had LSM < 9.3 kPa (specificity = 61%). In multivariable analysis, discordance was associated with female gender (adjPRR = 1.80, 95%CI 1.1-2.9; P = .019), herbal medicine use (adjPRR 1.64, 95% CI = 1.0-2.5; P = .022), exposure to lake or river water (adjPRR 2.05, 95% CI = 1.1-3.7; P = .016), and current smoking (adjPRR 1.72, 95%CI 1.0-2.9; P = .045). These data suggest that TE among rural Ugandans has low specificity for detection of histologically confirmed liver fibrosis. Caution should be exercised when using this tool to confirm significant liver fibrosis.
