Abstract
Quantification of the number of noninhibited responses (lures) in the inhibitory control task (ICT) has been proposed for the diagnosis of minimal hepatic encephalopathy (MHE). We assessed the efficacy of ICT compared with recommended diagnostic standards.
We studied patients with cirrhosis and healthy individuals (controls) who underwent the ICT at 2 centers (center A: n = 51 patients and 41 controls, center B: n = 24 patients and 14 controls). Subjects were evaluated for MHE by psychometric hepatic encephalopathy score (PHES). Patients from center B also were assessed for MHE by critical flicker frequency and spectral electroencephalogram analyses.
Patients with cirrhosis had higher ICT lures (23.2 ± 12.8 vs 12.9 ± 5.8, respectively, P < .01) and lower ICT target accuracy (0.88 ± 0.17 vs 0.96 ± 0.03, respectively, P < .01) compared with controls. However, lures were comparable (25.2 ± 12.5 vs 21.4 ± 13.9, respectively, P = .32) among patients with/without altered PHES (center A). There was a reverse, U-shaped relationship between ICT lure and target accuracy; a variable adjusting lures was devised based on target accuracy (weighted lures at center B). This variable differed between patients with and without MHE. The variable weighted lures was then validated from data collected at center A by receiver operator characteristic curve analysis; it discriminated between patients with and without PHES alterations (area under the curve = 0.71 ± 0.07). However, target accuracy alone was as effective as a stand-alone variable (area under the curve = 0.81 ± 0.06).
The ICT is not useful for the diagnosis of MHE, unless adjusted by target accuracy. Testing inhibition (lures) does not seem to be superior to testing attention (target accuracy) for the detection of MHE.