Digestive Diseases and Sciences, Vol. 50, No. 10 (October 2005), pp. 17801784 ( C 2005) DOI: 10.

1007/s10620-005-2937-6

Breath Ammonia Testing for Diagnosis of Hepatic Encephalopathy

SUJA DuBOIS, MD,* SUE ENG, MD,* RENUKA BHATTACHARYA, MD,* STEVE RULYAK, MD,* TODD HUBBARD, DAVID PUTNAM, PhD, and DAVID J. KEARNEY, MD*

Measurement of arterial ammonia has been used as a diagnostic test for hepatic encephalopathy, but obtaining an arterial specimen is an invasive procedure. The aim of this study was to evaluate the ability of a minimally invasive, highly sensitive optical sensing device to detect ammonia in the breath of patients with end-stage liver disease and to evaluate the correlation of breath ammonia levels, arterial ammonia levels, and psychometric testing. Fifteen subjects with liver cirrhosis and clinical evidence of hepatic encephalopathy underwent mini-mental status examination, number connection test, focused neurological examination, and arterial ammonia testing. On the same day, breath ammonia testing was performed using an apparatus that consists of a sensor (a thin membrane embedded with a pH-sensitive dye) attached to a beroptic apparatus that detects optical absorption. Helicobacter pylori testing was performed using the 14 C urea breath test. A positive correlation was found between arterial ammonia level and time to complete the number connection test (r = 0.31, P = 0.03). However, a negative correlation was found between breath ammonia level and number connection testing (r = 0.55, P = 0.03). Furthermore, no correlation was found between breath and arterial ammonia levels (r = 0.005, P = 0.98). There is a signicant correlation between the trailmaking test and arterial ammonia levels in patients with cirrhosis. However, no correlation was found between breath and arterial ammonia levels using the beroptic ammonia sensor apparatus in this small study.
KEY WORDS: encephalopathy; cirrhosis; ammonia.

Hepatic encephalopathy is a neuropsychiatric consequence of advanced liver disease. Although the exact pathogenesis is unknown, accumulation of ammonia from poor hepatic function and portosystemic shunting has been implicated as a primary factor (1). A pathogenic role for ammonia is supported by the effectiveness of treatments that facilitate a reduction in the production and absorption of ammonia, such as lactulose and neomycin (2). Thus far, there is no one simple, reliable, noninvasive, repro-

Manuscript received June 2, 2004; accepted February 18, 2005. From the *Department of Medicine, Division of Gastroenterology, University of Washington School of Medicine, Seattle, Primary and Specialty Medical Care Service, VA Puget Sound Health Care System, Seattle, and Pacic Technologies, Kirkland, Washington, USA. Address for reprint requests: David J. Kearney, MD, Seattle VAMC 111GI, 1660 South Columbian Way, Seattle, Washington 98108, USA; kearney@u.washington.edu.

ducible test to quantify the degree of hepatic encephalopathy. Clinical testing involves evaluation for psychiatric and neuromuscular disturbances, such as personality changes, disorientation, and asterixis, and may be graded using the West Haven criteria for altered mental state or the Glasgow coma scale (3). Biochemical testing may consist of determination of blood ammonia level, though accuracy may vary depending on whether venous or arterial blood is tested and on specimen handling, transport, and measurement techniques (4). When clinical encephalopathy is not readily apparent, various psychometric and electrophysiological tests may be employed to assess for minimal encephalopathy, such as the widely used trailmaking test (5). Since blood ammonia tests are invasive and subject to technical difculties in measurement, we sought to evaluate breath ammonia testing with a novel beroptic ammonia sensor as a potential diagnostic test for hepatic
Digestive Diseases and Sciences, Vol. 50, No. 10 (October 2005)

1780
0163-2116/05/1000-1780/0
C

2005 Springer Science+Business Media, Inc.

BREATH AMMONIA TESTING FOR HEPATIC ENCEPHALOPATHY DIAGNOSIS

encephalopathy. Breath ammonia results are compared with arterial ammonia levels and with the trailmaking test. MATERIALS AND METHODS
Cirrhotic patients who met minimal listing criteria for liver transplant evaluation (MELD score 10, ChildPugh score 7) with a clinical history of hepatic encephalopathy were asked to participate in the study. The study was approved by the Human Subjects Committee of the University of Washington. All subjects gave written informed consent prior to inclusion in the study protocol. The presence and degree of encephalopathy were evaluated by a mini-mental status examination, psychometric testing with the trailmaking test (number connection tests A and B), and focused neurologic exam. Patients were excluded if they had renal insufciency (BUN >40 mg/dl or creatinine >2.0 mg/dl within the past 3 months), a history of surgical shunt procedure or transjugular intrahepatic portosystemic shunt (TIPSS), severe chronic obstructive pulmonary disease or asthma, sedatives or narcotics within the 48 hr prior to enrollment, a history of a neurologic disorder, Wilsons disease, or diabetes mellitus requiring treatment with insulin. Patients were also excluded if they had used a proton pump inhibitor or bismuth compounds within the past 2 weeks or antibiotics within the past month, so as not to interfere with the 14 C urea breath test. Patients had an arterial blood gas obtained as part of routine pretransplant evaluation, with an additional 4 ml of arterial blood obtained for ammonia measurement. Patients underwent breath ammonia testing and testing for Helicobacter pylori on the same day. All patients had fasted for at least 6 hr at the time of the study procedures. All subjects were off of proton pump inhibitors, sucralfate, or H2 receptor antagonists for at least 2 weeks prior to study procedures. Trailmaking Test. The conventional trailmaking test has two parts (Ref 10, No. 3168). The rst portion (A) consists of 25 circles numbered consecutively from 1 to 25 on a sheet of paper. The subject is required to connect the circles in consecutive order by drawing a line between circles as quickly as possible. The second portion (B) consists of 25 circles, 13 of which are numbered consecutively, and 12 alphabetically labeled from A to L. The subject is asked to connect the numbers and letters in alternating fashion (i.e., 1, A, 2, B, . . .). If an error is made, the proctor informs the subject of the error and the subject is required to correct it. The score is the time in seconds to completion, including the time to correct errors. Ammonia Breath Test Instrument and Materials. The ammonia breath testing system is currently under development as a new experimental device. The ammonia sensor consists of a thin membrane, xed to one end of a 250-m beroptic cable and embedded with a pH dye that is sensitive to ammonia gas. As gaseous ammonia enters the membrane, the dye shifts from its acidic to its basic form as it is deprotonated; this results in an optical absorbance change.The sensor is responsive to gaseous ammonia but unaffected by dissolved ions such as ammonium and unaffected by normal constituents of the breath, such as CO2 . The optical measuring instrument consists of three separate opto-electronic modules (to measure three beroptic sensors at one time) and control software running on an attached PC. The instrument measures the differential intensity of the ammonia and reference wavelengths reected from the ammonia sensor and calculates the ratio of these two optical signals. Two types of
Digestive Diseases and Sciences, Vol. 50, No. 10 (October 2005)

ammonia sensors with different sensitivity levels were utilized in this study: a more sensitive AS-1 sensor, capable of detecting ammonia in the range of 50 to 2000 ppb (parts per billion), and an AS-2 sensor, most effective at detecting ammonia in the 500to 5000-ppb range. Breath Ammonia Protocol. Fiberoptic sensors (held at 100% relative humidity and room temperature) were attached to the optical measurement instrument, inserted into the T-tube, and monitored for at least 5 min in air. At the end of this test period, the data le was annotated with an air event marker. Subjects were then asked to breathe normally into the device for 5 min, and this data le was annotated with a baseline event marker, representing the basal breath ammonia measurement. Following each test, sensors were exposed to known ammonia standards for calculation of calibration tting parameters used in posttest analysis of breath ammonia levels. Urea Breath Testing. A 14 C breath test was performed during the same visit as a diagnostic test for H. pylori infection. Patients had fasted for at least 6 hr and been off of proton pump inhibitors, sucralfate, and H2 receptor antagonists for at least 2 weeks and off of antibiotics or bismuth preparations for at least 4 weeks. Patients ingested one capsule containing 1 Ci of 14 C-labeled urea (Ballard TriMed, Draper, UT, USA) with 30 ml of water. A breath sample was obtained 10 min later and analyzed with a scintillation counter for the presence of 14 C. A positive urea breath test was identied as breath 14 CO2 excretion >200 dpm, an indeterminate test as 50200 dpm, and a negative test as <50 dpm. Statistical Analysis. All continuous variables are expressed as mean standard deviation. A correlation coefcient was calculated comparing arterial ammonium concentration and results of the number connection test. Correlation coefcients were also calculated comparing breath ammonia concentration and the number connection test and comparing breath ammonia and arterial ammonium concentrations. It was calculated that 16 subjects would be needed in order to detect.

RESULTS Seventeen patients were enrolled in the study. One subject was withdrawn because his blood was found to be too lipemic to perform arterial ammonia measurement and one subject was withdrawn because he did not keep his appointment to have an arterial ammonia level drawn. Fifteen patients with hepatic cirrhosis undergoing evaluation for liver transplantation completed the study procedures and are included in this analysis. Demographic data are presented in Table 1. Twelve of fteen subjects were male, and the average age was 55 years, with a range of 46 to 60 years.. All subjects were found to be H. pylori negative according to the 14 C urea breath test. A wide range of results was found for the standard and expanded number connection tests (Table 1). A positive correlation was found between arterial ammonia level and time to complete the number connection test (r = 0.31, P = 0.03; Figure 1). However, a negative correlation was found between breath ammonia levels and time to complete the number connection test (r = 0.55,

1781

DuBOIS ET AL.
TABLE 1. PATIENT CHARACTERISTICS Sex (M/F) Age (years) Mean SD Range Medications Lactulose Lactulose + neomycin None NCT (sec), mean (range) Standard Expanded Arterial NH4 (g/dl), mean (range) Breath NH4 (ppm), mean (range) Note. NCT, number connection test. 12 (80%)/3 (2%) 55 4.6 4660 14 (93%) 1 (7%) 1 (7%) 57 (29110) 143 (80345) 155 (33306) 0.54 (00.96)

P = 0.03; Figure 2). No correlation was found between breath ammonia levels and arterial ammonia levels (r = 0.005, P = 0.98; Figure 3). DISCUSSION The correlation between plasma ammonia levels and severity of hepatic encephalopathy is not consistent in the literature, and this may be due to differences in sampling sites and assays used (68). Venous ammonia measurement is subject to variability due to a number of factors including contamination of the laboratory with ammoniacontaining compounds, smoking by patient or technician, delay or turbulence in transport, and patient factors such as muscle mass, urinary pH, and hypokalemia (4, 9). Arterial ammonia is thought to provide a more accurate assessment of the amount of ammonia at the bloodbrain barrier and fewer false-negative results compared to ve-

nous ammonia measurements (8). The partial pressure of ammonia (pNH3 ) can be calculated from the total ammonia and pH, and has been found to correlate more strongly with stage of hepatic encephalopathy than total arterial ammonia (6). However, Ong et al. found that total venous ammonia levels correlated with severity of hepatic encephalopathy just as well as arterial ammonia and partial pressure of ammonia (7). This study conrmed the previously established correlation between the trailmaking test and arterial ammonia levels in cirrhotic patients (10). As many as 60% of patients with liver cirrhosis and no clinical signs of encephalopathy will have abnormalities in number connection testing (11). Time to complete the number connection test has also been found to be signicantly prolonged in cirrhotics compared with controls (P < 0.01) (5). Improved sensitivity is found using number connection test B compared to A, but disadvantages include the signicant inuence of age and education on test results (12). In our study we did not nd a positive correlation between breath ammonia levels and trailmaking testing or a signicant correlation between breath and blood ammonia levels, as was reported in two other publications. Shimamoto et al. showed a positive correlation between blood ammonia and breath ammonia in 20 patients with cirrhosis (r = 0.798, P < 0.001), and signicantly higher breath ammonia levels in cirrhotics compared with healthy controls (13). However, ammonia measurement was from venous sampling, and 66% of cirrhotics with elevated blood ammonia were H. pylori seropositive. It is possible that venous ammonia may be better correlated with breath ammonia than arterial ammonia, since the majority

Fig 1. Correlation between arterial NH4 and number connection test.

1782

Digestive Diseases and Sciences, Vol. 50, No. 10 (October 2005)

BREATH AMMONIA TESTING FOR HEPATIC ENCEPHALOPATHY DIAGNOSIS

Fig 2. Correlation between breath NH4 and number connection test.

of breath ammonia would be expected to come from the pulmonary arterial circulation (with a small contribution from the bronchial artery). Wakabayashi et al. also evaluated the utility of expiratory ammonia concentration in assessing hepatic encephalopathy and found a positive correlation between blood ammonia concentration and expiratory NH3 concentration (P < 0.05) (14). H. pylori status was not mentioned. All patients in our study were

H. pylori negative by 14 C urea breath testing. Helicobacter pylori has been implicated as a potential source of ammonia in the stomachs of patients with cirrhosis and is considered a possible risk factor for encephalopathy in these patients. In an initial pilot study, breath ammonia measurement using the beroptic technique showed distinct patterns for H. pylori-positive and -negative subjects (15).

Fig 3. Lack of correlation between breath NH4 and arterial NH4 . Digestive Diseases and Sciences, Vol. 50, No. 10 (October 2005)

1783

DuBOIS ET AL.

Our study in fact showed a negative correlation between breath ammonia and psychometric testing, and no correlation between blood and breath ammonia measurements. This raises concern regarding the accuracy of the breath ammonia measurements by our technique. This may be explained by technical factors such as loss of ammonia in the passage space, i.e., upper respiratory tract, mouth, or detector tube. Additionally, it has been shown that glutaminase activity is expressed in human airway epithelium and is increased in response to acidic stress and downregulated by cytokines such as TNF- and interferon- (16). It is theoretically possible that decreased pulmonary glutaminase activity could account for decreased liberation of ammonia in the breath of these patients. Furthermore, various pulmonary function abnormalities may be seen in patients with liver disease that may have affected breath ammonia results (17). Patients with ascites may have decreased tidal volumes, and perhaps those with a greater degree of encephalopathy may experience a decreased respiratory rate from depressed respiratory drive. The majority of patients in this study had almostnormal mini-mental status exams, suggesting minimal encephalopathy in these patients. Although most had elevated arterial ammonia and abnormalities in number connection testing, perhaps a patient population with more severe encephalopathy would have shown a positive correlation between breath ammonia and objective measures of encephalopathy. In summary, this study conrmed the correlation between the trailmaking test and arterial ammonia levels in patients with cirrhosis. However, a positive correlation was not found between breath and arterial ammonia levels using the beroptic ammonia sensor in this small study.

REFERENCES
1. Schiff E, Sorrell M, Maddrey W: Diseases of the Liver. Philadelphia, LippincottRaven, 1999 2. Riordan S, Williams R: Treatment of hepatic encephalopathy. N Engl J Med 337:473479, 1997 3. Blei A, Cordoba J: Hepatic encephalopathy. Am J Gastroenterol 96:19681975, 2001 4. Yurdaydin C: Blood ammonia determination in cirrhosis: Still confusing after all these years? Hepatology 38:13081309, 2003 5. Saxena N, et al.: Electrophysiological and neuropsychological tests for the diagnosis of subclinical hepatic encephalopathy and prediction of overt encephalopathy. Liver 22:190197, 2002 6. Kramer L, et al.: Partial pressure of ammonia versus ammonia in hepatic encephalopathy. Hepatology 31:3034, 2000 7. Ong J, et al.: Correlation between ammonia levels and the severity of hepatic encephalopathy. Am J Med 114:188193, 2003 8. Stahl J: Studies of the blood ammonia in liver disease. Ann Intern Med 58:122, 1963 9. Gerron G, et al.: Technical pitfalls in measurement of venous plasma NH3 concentration. Clin Chem 22:663666, 1976 10. Conn H: Trailmaking and number-connection tests in the assessment of the mental state in portal systemic encephalopathy. Dig Dis 22:541550, 1977 11. Rikkers L, et al.: Subclinical hepatic encephalopathy: Detection, prevalence and relationship to nitrogen metabolizm. Gastroenterology 75:462469, 1978 12. Weissenborn K, et al.: The number connection tests A and B: interindividual variability and use for the assessment of early hepatic encephalopathy. J Hepatol 28(4):646653, 1998 13. Shimamoto C, Hirata I, Katsu K: Breath and blood ammonia in liver cirrhosis. Hepato-Gastroenterology 47:443445, 2000 14. Wakabayashi H, et al.: Measurement of the expiratory ammonia concentration and its clinical signicance. Metab Brain Dis 12:161 169, 1997 15. Kearney D, Hubbard T, Putnam D: Breath ammonia testing in Helicobacter pylori infection. Dig Dis Sci 47:25232530, 2002 16. Hunt J, et al.: Expression and activity of pH-regulatory glutaminase in the human airway epithelium. Am J Respir Crit Care Med 165:101107, 2002 17. Fallon M, Abrams G: Pulmonary dysfunction in chronic liver disease. Hepatology 32:859865, 2000

1784

Digestive Diseases and Sciences, Vol. 50, No. 10 (October 2005)