Hepatic encephalopathy (HE) is a common complication of cirrhosis that affects quality of life, increases the risk of accidents, and is an independent predictor of poor outcome [¹,²]. When neurological deficits are subtle but the neurological clinical examination is normal, a condition referred to as minimal HE [³], patients are exposed to a risk of developing clinical episodes of HE over time [⁴]. The presence of HE in cirrhosis is a prognostic marker of severity and a valid indication for liver transplantation, although it is not considered in the model for end-stage liver disease (MELD) score on which organ distribution is based in most liver transplant centres [⁵].

Neurological alterations observed in HE are postulated to result from the exposure of the brain to abnormally elevated concentrations of ammonia present in the general circulation in response to liver insufficiency and portosystemic collaterals [⁶]. Accordingly, high ammonia levels have been associated with large portosystemic collaterals such as esophageal varices in patients with cirrhosis [⁷]. However, ammonia determination is not currently accepted as a reliable marker to identify patients with HE [⁸]. Hyperammonemia arises from the production by colonic bacteria and the small intestine through an increased intestinal glutaminase activity [⁹]. Although the pathogenesis of HE is still incompletely elucidated, the ammonia hypothesis remain central [¹⁰] and a large number of experimental data support the role of hyperammonemia in the direct and indirect alterations of brain function that characterize HE [¹¹].

Making a diagnosis of HE may be straightforward when a patient with cirrhosis presents with obvious neurological deficits such as altered consciousness, but it is much more challenging in the presence of more subtle neuropsychological or personality changes that are not uncommon in an outpatient population of cirrhotics (up to 62% in a recent report [¹²]). In fact, it is recommended to search for minimal HE in patients who complain of cognitive alterations, a disturbed sleep [¹³], or are exposed to an accident risk while driving or at their work-place. As neurological deficits associated with minimal HE are clinically subtle, this complication may be underdiagnosed and may negatively impact patients' management. Accordingly, minimal HE can be accessible to medical therapy that may improve quality of life and prevent the development of clinical episodes of overt HE.

In clinical practice, the available tools for the diagnosis of HE include clinical scales to assess the mental status, such as the West-Haven scale [¹⁴], and a number of psychometric tests to assess the presence of congnitive deficits [¹⁵]. Neuroradiological imaging is mostly directed at excluding other neurological disorders. Blood ammonia concentration in the context of HE is difficult to interpret, as the correlation between neurological symptoms and ammonia blood levels is variable, with a wide overlap across different stages of HE [⁸]. Some [¹⁶,¹⁷], but not all [⁷,¹⁸] studies report a closer correlation with arterial as compared to venous ammonia blood levels. The diagnosis of minimal HE is therefore challenging for the clinician who has to choose between a suboptimal biological test or a number of neuropsychological tests that may be time consuming, need to be adjusted for several parameters, and are subject to learning biases [³].

To improve the performance of blood ammonia for the diagnosis of minimal HE, we hypothesized that measuring ammonia in arterialized blood following oral glutamine-induced hyperammonemia and consecutive cognitive deterioration would increase the diagnostic yield of this biological test. Therefore, we explored the value of capillary blood ammonia measured at bedside following an oral glutamine challenge to unmask HE in patients with cirrhosis.

Statistical tests were performed using Statistical Package for the Social Sciences SPSS version 10.0 (SPSS Inc. Chicago, IL, USA) Continuous variables were expressed as mean and standard deviation or median value (interquartile range [IQR]) as appropriate. The results of psychometric tests were transformed into Z-scores, which allows to normalize the value to a standard distribution curve. We used the Wilcoxon-signed rank, and chi-squared test as appropriate. To test for blood ammonia changes over time, we used the one way repeated measures ANOVA test with Dunnett's test. We constructed receiver-operating characteristics (ROC) curves to assess for overall accuracy of capillary blood ammonia levels and to identify optimal cutoffs. To identify independent predictors of development of HE episodes during follow-up, we performed a multivariate logistic regression analysis using a Cox proportional hazard model. A 2-sided p value of less than 0.05 was considered statistically significant.

The study protocol was approved by the Ethics Committee of the Geneva University Hospitals and by the Swiss Agency for Therapeutic Products (SwissMedic) authorities. The study was conducted in accordance with the ethical principles of the Declaration of Helsinki. All patients and controls gave written, informed consent to participate.

Making a diagnosis of minimal HE is challenging for the clinician, who needs a sensitive, reliable, and easy-to-use diagnostic tool. However, at present, neuropsychological evaluation and electrophysiological tests do not fulfill these requirements. This aspect may explain in part why only a minority of hepatologists are screening for minimal HE in daily practice [²⁵]. Therefore, a simple test would be welcome and greatly facilitate the diagnosis and thus the management of HE. In our study, we tried to improve the diagnostic accuracy of blood ammonia, a key substance in the pathogenesis of HE, for the presence of minimal HE. To do so, we used capillary blood as an equivalent to the arterial compartment, and induced hyperammonemia via small intestine ammonia production to unmask HE. Finally, we designed the test as a bedside procedure fulfilling the definition of POCT to improve patients' and clinician's acceptance.

First of all, using our short test battery, we confirmed the high prevalence of minimal HE (44%) in our patients with advanced cirrhosis (82% of Child B and C), consistent with published data [³,⁴]. Secondly, we demonstrated that an amino acid load using oral glutamine at a higher dose than used in previous studies [²⁶,¹⁹] is well tolerated in patients with moderate to severe liver insufficiency. We also observed that the present test didn't precipitate any clinical episodes of HE in the 4 patients with TIPS, as this situation worsens portal-systemic shunting and makes the patient more vulnerable to the HE [²⁷]. Our results also demonstrate that capillary blood ammonia is not an accurate biological marker of minimal HE in patients with cirrhosis, nor a reliable predictor of future episodes of overt HE during follow-up. Nevertheless, we show that the diagnostic value of capillary blood ammonia can be substantially improved following an oral aminoacid load, raising the AUROC value from 0.541 at baseline to 0.727 at 60 minutes after the oral glutamine challenge. This significant and rapid elevation in blood ammonia after the test is consistent with the significant contribution of the small intestine to the production of ammonia [²⁸], in addition to the well-accepted role of colonic bacteria. The presence and severity of portosystemic venous collaterals is another important parameter that determines blood ammonia levels [⁷]. Ortiz et al. [²⁹] reported a similar increase in capillary blood ammonia following an oral glutamine challenge in patients with congenital portosystemic shunts as compared to patients with cirrhosis. In our study, most patients had evidence of portosystemic collateralization visible on different imaging modalities preventing an accurate analysis, therefore we did not include this parameter nor the value of portal pressure in the statistical model.

Induced hyperammonemia using either glutamine alone [²²] or combined with other amino acids [³⁰] is associated with electroencephalographic, brain magnetic resonance and biochemical alterations, the neuropsychological consequences of which are inconstant, showing either absent [³¹], partial [²²] or generalized [³²] deterioration of tests' performance. In the present study, we observed changes limited to the Trail A and B tests, both commonly used components of standard test battery for the detection of minimal HE [³], suggesting that visuo-spatial skills and visuo-motor coordination are affected by hyperammonemia. We also explored whether induced hyperammonemia could predict the development of overt HE during follow-up, and found that post test capillary blood ammonia was a poor predictor of events. This is in contradiction with results from the study by Romero-Gomez et al. [¹⁹] who reported that an abnormal ammonia response after oral glutamine was predictive of the development of overt HE during follow-up. The distribution of liver disease severity in the 2 studies (Child A patients: 75% versus 17% in our study) and the absence of any treatment for HE that may influence the ammonia response following a glutamine challenge [³³,³⁴], could explain these differences.

Our study is in line with previous results which showed that determination of blood ammonia levels, either a single venous dosage [¹⁸] or in capillary blood after induced hyperammonemia, correlates only poorly with symptoms of HE. Whether determination of ammonia in a arterialized-venous blood, as in our study, shows higher values as compared to venous blood has been proposed [¹⁶] but not confirmed in our small subgroup of 8 patients. Thus, additional parameters such as size and extension of portosystemic shunts [²⁹,¹⁵], hyponatremia [³⁵], nutritional status [³⁶], the presence of hepatitis C infection [³⁷], together with the presence of a systemic inflammatory reaction syndrome (SIRS) [³⁸] that may modulate ammonia neurotoxicity at the cerebral level, are factors that influence the development of HE. We were however not able to examine all these parameters in our cohort, as values of serum creatinine, sodium, and C-reactive protein were available only in a limited number of patients.

In conclusion, we demonstrated that determination of capillary blood ammonia after a bedside oral glutamine challenge is feasible, well tolerated and superior to basal levels for the diagnosis of minimal HE in patients with cirrhosis and moderate to severe liver failure. The increase in blood ammonia, absent in healthy subjects, correlated imperfectly with the deterioration in psychometric tests. Thus, in spite of a recognized important pathogenic role in HE, blood ammonia, even in a capillary, arterialized vascular bed, remains an approximate marker for the presence of HE. According to our data in unselected patients with cirrhosis, ammonia determination in capillary blood after an oral glutamine challenge is not a valid tool for the diagnosis of HE and a poor predictor of future clinical episodes of HE.

HE: hepatic encephalopathy; MELD: model for endstage liver disease; TIPS: transjugular intrahepatic shunt; HVPG: hepatic venous pressure gradient; POCT: point-of-care testing; AUC: area under the curve.

The authors declare that they do have nothing to disclose regarding funding or conflict of interest with respect to this manuscript.

SD: participated in the development of the study, inclusion and follow-up of patients, performance of psychometric tests and blood ammonia dosage, and drafting of the manuscript. EG: participated in the selection of patients and drafting of the manuscript, and performed the statistical analysis. PRB: participated in the development of the study, supervised the psychometric tests, and drafted the manuscript. NG: participated in the development of the study and drafting of the manuscript. GM: participated in the selection and inclusion of patients. AH: participated in the development of the study, selection of patients and drafted the manuscript. LS: designed the study, included patients, performed psychometric tests and ammonia dosage, and drafted the manuscript.

All authors read and approved the final version of the manuscript.

The pre-publication history for this paper can be accessed here:

http://www.biomedcentral.com/1471-230X/11/134/prepub