Caffeine clearance in cirrhosis

Journalof Hepatology, 1992; 14: 157- 162 @ 1992 Elsevier Science Publishers B.V. All rights reserved. 0168-8278/92/%05.00

157

HEPAT 01003

Frederick W. Lewis and William 6. Universityof Colorado Health Sciences Center and Denver General Hospital Department of Medicine, Division of Gastroenterology, Denver, CO, United Statesof Amc-ice (Received 23 April 1990)

Less complex methods of measuring hepatic metabolic capacity are needed. A simplified caffeine clearance test was evaluated in 23 patients with stable alcoholic liver disease. First, saliva. caffeine concentrations were measured over a 24-h caffeine-free interval. Clearance was calculated from the rate of elimination of caffeine and an assumed volume of distribution and compared with the results of a formal clearance test using sequential plasma and saliva samples following a 300 mg oral dose. The simplified method was then assessed in 11 hospitalized patients with cirrhosis. Saliva caffeine concentrations remained measurable over the interval of study in 82% of patients. Caffeine clearance as determined by the simplified method did not differ from plasma caffeine clearance after an oral dose. Application of this method was achieved in 11 of 12 patients hospitalized for complications of severe liver disease, and revealed markedly diminished clearance. Thus, caffeine clearance can be accurately estimated in patients with severe liver disease using two or more samples of either saliva or plasma. This simplified determination of caffeine elimination rate provtdes a more practical assessment of hepatic metabolic capacity than a formal clearance test. --

-L-herate of elimination of caffeine has been proposed as a quantitative measure of hepatic function (l-3). Previous investigatiorls have demonstrated diminished caffeine metabolism in patients with liver disease (4-Q, measurable concentrations of caffeine in plasma after an overnight fast (g-11), and a relationship between the reduction in caffeine clearance and clinical indicators of the severity of liver disease (12). The majority of patients studied in these investigations have had mild or moderate liver dysfunction (6-12). Clearance studies are cumbersome to perfarm, particularly in severely ill patients, and consequently have not been commonly used to assess liver function for clinical purposes. However, caffeine is present in many beverages and is widely consumed. Estimation of caffeine clearance from the rate of disappearance of caffeine pre-

viously consumed obviates the need for an administered dose. Such a determination is simpler, less invasive and more practical than a formal clearance test (11). This technique could provide a quantitative test of hepatic function more readily applicable to the clinical assessment of patients with liver disease. The purposes of this study were to evaluate simplified methods of determining caffeine clearance in a group of outpatients with mild to moderate liver disease, and to identify factors other than hepatic function which might interfere with the validity of these determinatiGns. Subsequently, these methods were applied as a test of hepafic metabolic function in a group of hospitalized patients with more severe liver dysfunction for whom a formal clearance study proved impractical.

Correspondence: Frederick W. Lewis, M.D., 950 E. Harvard, Suite 540, Denver, CO 80210, U.S.A.

158

F.W. LEWIS et al.

Patients and Methods Patientsand protocols All protocols conformed to the ethical guidelines of the 1975 Declaration of Helsinki and were approved by the Human Subjects Committees of the University of Colorado Health Sciences Center or Denver General Hospital. All patients gave written informed consent. Group I consisted of 23 male outpatients with mild or moderate alcoholic liver disease admitted to the Adult Clinical Research Center of the University of Colorado Health Sciences Center (Table 1). The diagnosis of chronic alcoholic liver disease was made by standard clinical and laboratory data, including liver biopsy where indicated. No patient with acute alcoholic hepatitis or fatty liver alone was included. Patients with recent gastrointestinal hemorrhage or renal insufficiency were excluded, as were patients taking theophylline, cimetidine, phenytoin or other medications known to alter hepatic drug metabo-

TABLE 1 Patient characteristics GroupIa n

Male/female Age (years) Estimatedhome caffeine intake (mglday) Body weight (kg) Ascites present Encephalopathypresent Serum albumin (nl: 3.7-4.8 g/dl) Total bilirubin (nl: Cl.0 mg/dl) Prothrombintime (nl: 11.1-12.6s) Aspartate aminotransferase (nl: C37 NJ/l) Alanine aminotransferase (nl: s40 IUfl) Serum credtinine (nl: 0.6-1.3 mg/dl) Creatinine clearance (mYmin) Modified Child-Pugh class A B C Death before hospital discharge Death within 6 months

23 23/O 42 (28-65) 81 (o-320) 68 (57-102) 9 (39%)

Group IIa 11 912(82%118%) 43 (30-63)

83 (44-116) 6 (55%) 5 (45%)

2.9 (1.7-4.2)

2.1(1.5-3.1)

1.3 (0.4-9.3)

6.5 (0.5-21.4) 16.8 (12.4-19.5)

70 (13-430)

164 (27-1056)

45 (10-176)

51(13-268)

0.8 (0.6-1.1)

1.1(0.6-6.8)

107 (49-224) 10 (44%)

0

7 (30%)

3 (27%)

6 (26%)

8 (73%)

0 3 (13%)

3 (27%) 5 (45%)

Values given are the median values; values shown in parentheses are the rangevalues. a Group I, outpatients studied on a research ward; group II inpatlents studied during an acute hospital admission for complications of advancedliver disease.

lism. A modified Child-Pugh score was calculated (13). Patients were placed on a diet free of caffeine and theobromine for the duration of the study period. The nursing staff was aware of the restriction on caffeine intake and monitored the patients’ compliance. A detailed dietary history was obtained and usual caffeine intake and cigarette use at home estimated. Parotid size was estimated by physical examination, and assigned a value on a scale of O-4. Ascites was documented by physical examination and urinary sodium excretion. Five ml of saliva were obtained by having the patient chew paraffin. Fasting saliva samples were collected 12 and 36 h after admission. On day 3, after an overnight fast, a 300 mg dose of caffeine (Sigma Chemicals, St. Louis, MO) was given in apple juice, and plasma and saliva samples obtained at 0,4,8,12,16,24, and 30 h. Plasma was separated in a refrigerated centrifuge and saliva and plasma were stored at -70 “C until analysis. Group II concluded 11 patients hospitalized at Denver General Hospital for complications of moderate or severe liver disease (Table 1). Most patients were studied during periods of restricted oral intake because of gastrointestinal hemorrhage, hepatic encephalopathy or other serious illness, and those able to consume orally were placed on a caffeine- and theobromine-free diet. Plasma samples were acquired at 24-h intervals at the time of venipuncture for clinical purposes, separated, and stored at -70 “C until analysis. Analyticaltechniques Plasma and saliva caffeine concentrations were determined using high pressure liquid chromatography (14). The method was modified to use l-2 ml of saliva instead of 0.5 ml in some patients in order to reliably detect concentrations in the 0.03-0.15 lug/ml range. Inter- and introday variability were ~7%. Salivary sodium and potassium concentration were determined by flame photometry, and salivary protein concentration was measured using the method of Bradford (15). Calculations To calculate caffeine clearance from fasting samples the natural logarithm of fasting saliva caffeine concentration was plotted against time using the least-squares method (Fig. l), and the slope of the line taken as the elimination rate constant (keti). In group I, volume of distribution (V,) was estimated as 0.6 x body weight (7), and caffeine clearance calculated as (k,lim-VJ. Volume of distribution was estimated in group PI patients specifically for the presen e or absence of ascites, based on ratios of measured caffeine V, to weight in patients in group I (0.69 Ukg for ascites, 0.53 l/kg for no ascites; see below). Caffeine

CAFFEINE

CLEARANCE

IN CIRRHOSIS

159

clearance was calculated from saliva and plasma concentration after the administered dose in a similar fashion, and V, calculated as the dose divided by the difference between the extrapolated caffeine concentration at time zero and the fasting value. The saliva to plasma ratio of caffeine concentration was similar over multiple sampling points in each patient, and an overall mean was calculated for each patient. Statistics Means are expressed as t standard deviation. Data were analyzed using the SAS statistical package (SAS Institute, Cary, NC). Paired and unpaired group comparisons were used where indicated. Linear correlations were made where appropriate, and rank order correlations (rs) where a linear relationship was not evident. A significance value of 0.05 was accepted.

0.000 +

1 .OOG -I

O.lOO-

0.010+ 12 36 TIME WITHOUT CAFFEINE (hours)

Fig. 1. Caffeine elimination determined from fasting saliva caffeine concentra*lon (log scale) in patients with alcoholic liver disease consuming a caffeine-free diet (group I).

In group I, caffeine was measurable in saliva after 12 and 36 h on a caffeine-free diet in 86 and 82% of patients, respectively, and ranged as high as 14.7 pg/ml (Fig. I). Caffeine clearance could not be calculated in three patients because fasting caffeine levels were undetectable at 12 or 36 h. In five patients clearance was not determined because of problems in collecting or processing one of the samples. Fasting saliva caffeine concentration was strongly correlated with measured plasma caffeine clearance (at 12 h, r, = -0.66, p = 0.005; at 36 h, r, = -0.77, p = 0.0002). Caffeine elimination rate constant and clearance calculated from fasting concentrations in group I patients did not differ from kinetic parameters determined in plasma

TABLE

INTAKE

after an oral dose of caffeine (Table 2), and a strong linear relation between the two was observed (Fig. 2). Caffeine clearance, elimination rate constant, and half-life as determined in plasma were both similar to and correlated with values as determined in saliva after an oral dose (Table 2 and Fig. 2). However, volume of distribution calculated using saliva caffeine concentrations was significantly larger than that calculated using plasma concentrations. Factors influencing caffeine distribution into saliva were investigated. The saliva/plasma ratio of caffeine concentration, which varied between 0.70 and 1.05, correlated inversely with serum albumin concentration (Fig. 3) but not with parotid size or other chemical characteristics of saliva (protein, sodium and potassium concentra-

2

Caffeine elimination kinetics determined by three methods in patient group I Fasting caffeine concentrations

n r value for elimination plot Clearance (mUmin) Elimination rate constant (h-l) Half-life (h) Volume of distribution (liters) Median values are given. Values shown in parentheses a Calculated using the formula V, = 0.6 X mass (8).

t f, 12.1

(2.6-59.0)

Saliva caffeine concentrations (oral dose)

Plasma caffeine concentrations (oral dose)

22 0.98

(0.62-1.0)

23 0.99

(0.79-1.0)

30.5

(2.4-226.2)

26.6

(2.1-137.6)

0.014 (O.U04-0.093)

0.040 (O.U03-0.170)

0.031 (U.OU2-U.171)

47.6

(7.4-182.6)

16.2

(4.1-213.0)

21.6

(4.0-325.6)

40.1” (34.1-61.4)

45.0

(32.1-86.3)

42.7

(27.1-59.9)

are the range values.

F.W.

160

LEWIS

et al.

z.% z-200 ~~ LLg 9160 20 6% -J El.20 d5 g

o =narmal $

albumin @ = low albumin

80

$ I

32

40 k! 2

0

2 TIME WITHOUT

0 0

40

20

60

80

100

120

(cral

4

3 CAFFEINE

5 INTAKE

6

7

(days)

140

PLASMA CAFFEINE CLEARANCE (ml/min) ‘;‘ .E

1

dose) 0

I

Fig. 4. Caffeine elimination determined from fasting plasma caffeine concentrations at 24-h intervals in patients hospitalized with complications of moderate or severe liver disease (group II).

r=0.05

p=o.07

TABLE

3

Caffeine elimination kinetics in patient group II

wz =El k5

0 80 albumin 0 =normol e = low albumin

40 5? d ul

0 0

40

20

60

80

100

120

140

PLASMA CAFFEINE CLEARANCE (ml/min) (owl

IL-

4.1: SERUM

Fig. 3.

ALBUMIN

CONCENTRATION

0.97

(0.81-0.99)

3.3

(1.37-10.4)

0.0034 172 2.7

(O.O005-0.018) (37-1527) (0.4-8.0)

Median values are given. Values shown in parentheses are the range values.

dose)

Fig. 2. Comparison of methods in patients with normal and low serum albumin concentration. Plasma caffeine clearance (oral dose) compared to clearance determined using (a) fasting saliva concentrations; (b) saliva concentrations (oral dose).

z?.zL 0

r value for elimination plot Fasting caffeine concentration Wml) Elimination constant, caffeine (h-l) Half-life, caffeine (h) Caffeine clearance (ml/min)

(y/dl)

Saliva to plasma ratio of caffeine concentration varies inversely with serum albumin concentration.

tions, salivary sodium/potassium ratio: r, = -0.29, 0.08, 0.13, -0.23, and 0.30, respectively, all N.S.). Patients in group I with ascites had significantly larger volumes of distribution of caffeine as compared to patients without ascites. The ratio of volume of distribution to body weight was significantly greater in those with ascites (0.69 + 0.07) compared to those without ascites

(0.53 + 0.09, p = 0.0003). For this reason, V,, in group Ii patients was estimated from body weight using these ratios. Of 12 patients evaluated for group II, 11 had detectable plasma caffeine concentration in the fasting state and tifter at least one 24-h period. Caffeine elimination curves were constructed for all 11 patients (Fig. 4), and a strong fit for first-order elimination was observed (mean correlation coefficient for the logarithmic plots with greater than 2 points: r = 0.94 + 0.06). Elimination of caffeine xx markedly diminished in most patients, and mean calculated caffeine half-life was 278 h, or 11.6 days (Table 3). In one patient with intractable ascites, hepatorenal syndrome and encephalopathy, caffeine elimination had virtually ceased (clearance c 1 ml/mm, half-life > 1500 h). Caffeine clearance correlated strongly with modified Child-Pugh score in group I patients (rs = -0.73, p c O.OOOl), and was diminished in those who died within 6 months (17 + 13 ml/min, n = 3) compared to survivors (36 + 35 ml/min, n = 20). Estimated caffeine intake was unrelated to fasting caffeine level at 12 or 36 h, caffeine clearance rate, or modified Child-Pugh score (rs = -0.04, -0.03,0.03, and -0.04, respectively). No significant relationship of caffeine clearance to Child-Pugh score or in-

CAFFEINE

CLEARANCE

IN CIRRHOSIS

hospital mortality was observed in group II, although the patient with a caffeine clearance < 1 mUmin died.

The purpose of this study was to compare several indicators of hepatic metabolic function which could easily be performed on a hospital ward or in an outpatient clinic. Caffeine was chosen as the model compound because of its safety and widespread consumption. The methods evaluated differ as to degree of invasiveness, complexity, and time required for administration and sample collection. Saliva collection is minimaIly invasive; however, a potential disadvantage of saliva analysis is variability of caffeine distribution into saliva. Caffeine in plasma is partially protein bound, limiting its diffusion into saliva. This might be affected by parotid gland function or alterations in plasma protein binding. We found no evidence that parotid dysfunction, which is commonly associated with alcoholic liver disease (l&18), affects caffeine distribution into saliva. However, protein binding of caffeine in plasma emerged as a potentially important variable. The saliva/plasma ratio of caffeine concentrations in our patients (0.70-1.05) varied over a Ader range than the values reported in normal subjects by previous investigators (0.69-0.79, Refs. 11, 19-21). The saliva/plasma ratio approached one in patients with severe liver dysfunction and hypoalbuminemia. Protein binding was not measured. However, the strong inverse correlation observed between serum albumin concentration and saliva/ plasma ratio of caffeine concentration suggests that increased distribution of caffeine from plasma into saliva results from the diminished protein binding of cafeine in hypoalbuminemic patients (2). Regardless of the mechanism, the similarity of plasma and saliva caffeine concentrations in patients with severe liver disease and hypoalbuminemia makes saliva sampling a highIy accurate method of estimating the plasma clearance of caffeine. In patients with milder degrees of liver disedse, salivary caffeine clearance overestimated plasma clearance by a factor of lo-20% as a result of the larger volume of distribution calculated from the lower salivary concentration of caffeine as compared to plasma. Determination of caffeine clearance from fasting samples was simple, noninvasive, and did not require an administered dose. This was easily performed by collecting of only two saliva samples in group I, and by daily plasma samples over a period of 2-7 days in group II. The accuracy of these determinations was confirmed by their similarity to results obtained by a formal clearance test in group I

161

and the high correlation with first-order elimination kinetits observed in group II. These methods, however, were not applicable to most patients with mild liver disease and normal serum albumin concentration because caffeine metabolism was nearly complete prior to the 36 h sampling point. A 12-h overnight test might have been more applicable to this population (11). EIowever, such a test would not be useful for patients with relatively severe liver disease because of insufficient change in caffeine concentration. Indeed, much longer intervalc- were needed for some patients in group II. The interval between samples for determination of caffeine clearance by this method should be tailored to the clinical severity of liver dysfunction. Use of an estimated rather than a calculated volume of distribution is one source of variability for caffeine clearance determined from fasting caffeine concentrations alone. Volumes of distribution of caffeine in plasma ranging from 0.47-0.60 l/kg have been reported in patients with cirrhosis (1,2,6,7,12). No information was available in these reports about the presence or absence of ascites. Patients with sodium retention and ascites would be expected to have a greater portion of body mass composed of water, and our findings confirm this relationship. In our patients with ascites, the volume of distribution of cafkine was best estimated by the formula (0.69 x mass), and in those without ascites by the formula (0.53 x mass). These formulae improve the estimation of V, necessary for calculation of caffeine clearance from fasting concentrations. Hepatic metabolism of caffeine was profoundly impaired in patients hospitalized with complications of liver disease. Symptomatic methyxanthine toxicity has rarely been reported (22,23), but this is the first study in which it has been specifically sought. This is potentially important issue, as the signs and symptoms of methylxanthine toxicity resemble in certain respects and may be confused with those of sepsis, alcohol withdrawal, and hepatic encephalopathy. Despite the severe impairment of hepatic metabolic capacity in our patients, no instances of caffeine intoxication were observed. The mechanism by which such patients avoid caffeine intoxication is unclear. Caffeine intake at home was not estimated in the hospitalized patients, as many were encephalopathic. However, dietary caffeine intake bore no relation to caffeine clearance in stable patients alcoholic liver disease (group I). Thus sug gests that diminished intake is not responsible for the rarity of caffeine intoxication in cirrhosis. Diffusion of caffeine into urine and feces could account for at best a small component of caffeine elimination in these patients. For example, equilibration between plasma and urine in a Patient with a plasma caffeine concentration of lO,uglml who

F.W. LEWIS et al.

162 excreted 1500ml of urine daily would result in a daily caffeine excretion of only 15 tug. Further investigation of the mechanisms by which caffeine intoxication is avoided in patients with severe liver disease is warranted. In this series and others, although a strong correlation was observed between caffeine clearance and clinical indicators of the severity of liver disease, Child-Pugh score predicted mortality as well or better than clearance (12,24-26). However, potential uses for a quantitative measure of liver function include not only prediction of survival, but also assessment of the severity of metabolic impairment to guide appropriate dosing of pharmacotherapeutic agents, delineation of progression of disease, detection of response to therapy and determination of ideal

References

9 10

11

12

13

Renner E, Wietholtz H, Huguenin P, Arnaud MJ, Preisig R. Caffeine:’ a model compound for measuring liver function. Hepatology 1984; 4: 38-46. Desmond PV, Patwardhan RV, Johnson RF, Schenker S. Impaired elimination of caffeine in cirrhosis. Dig Dis Sci 1980; 25: 193-7. Wietholtz H, Voegelin M, Amaud MJ, Bircher J, Preisig R. Assessment of the cytochrome p-448 dependant liver enzyme system by a caffeine breath test. Eur J Clin Pharmacol 1981; 21: 53-9. Scott NR, Stambuk D, Chakraborty J, Marks V, Morgan MY. Caffeine clearance and biotransformation in patients with chronic liver disease. Clin Sci 1988; 74: 377-84. Statland BE. Demas TJ. Serum caffeine half-lives. Am J Clin Patholl980; 75: 390-3. Joeres R, Klinker H, Heusler H, Epping J. Zilly W, Richter E. Influence of smoking on caffeine elimination in healthy volunteers and in patients with alcoholic liver disease. Hepatology 1989; 8: 575-9. Wang T, Kleber G, Stellaard F, Paumgartner G. Caffeine elimination: a test of liver function. Klin Wochenschr 1985; 63: 1124-8. Scott NR, Stambuk D, Chakraborty J, Marks V, Morgan MY. The pharmacokinetics of caffeine and it dimethylxanthine metabolites in patients with chronic liver disease. Br J Clin Pharmaco1 1989; 27: 205-13. Wahllander A, Renner E, Preisig R. Fasting plasma caffeine concentration. Stand J Gastroenterol1985; 20: 1133-41. Marchesini G, Checchia G, Grossi G, et al. Caffeine intake, fasting plasma caffeine and caffeine clearance in patients with liver diseases. Liver 1988; 8: 241-6. Jost G, Wahllander A. von Mandach U, Preisig R. Overnight salivary caffeine clearance: a liver function test suitable for routine use. Hepatology 1987; 7: 338-44. Holstege A, Staiger M, Haag K, Gerok W. Correlation of Caffeine Elimination and Child’s Classification in Liver Cirrhosis. Klin Wochenschr 1989; 67: 6-1.5. Braillon A, Cales P, Valla D, Gaudy D, Geoffery P, Lebrec D. Influence of the degree of liver failure on systemic and splanchnic

time for liver transplantation. The value of caffeine clearance for these uses remains to be tested, but the validity of these simplified measures of caffeine elimination and the ease with which they can be administered to severely ill patients establishes caffeine as an agent with great practical advantage for use as an indicator of hepatic function.

Acknowledgements

The authors would like to thank Georia DeRoche for technical assistance. This work was supported by United States National Institutes of Health grant Nos. l-R29 AA07832 and MOI-RRO0051.

haemodynamics and on response to propranolol in patients with cirrhosis. Gut 1986; 27: ;204--9. 14 DeRoch G, Portz B, Rector WG Jr, Everson GT. Simultaneous determination of caffeine and antipyrine in plasma and saliva using HPLC. J Liquid Chromatogr 1990; 13: 3493-505. 15 Bradford MM. A rapid and sensitive method for the quantitation of microgram quantities of protein utiliig the principle of protein-dye binding. Anal Biochem 1976; 72: 248-51. 16 Dutta SK, Dukehart M, Narang A, Lathman PS. Functional and structural changes in parotid glands of alcoholic cirrhotic patients. Gastroenterology 1989; 96: 510-8. 17 Abelson DC, Mandel ID, Karmoil M. Salivary studies in alcoholic cirrhosis. Oral Surg Orai Med Oral Path01 1976; 41: 188-92. 18 Wolfe SJ, Summerskill WHJ, Davison CS. Parotid swelling, alcoholism and cirrhosis. N Engl J Med 1957; 256: 491-5. 19 Zylber-Katz E, Granit L, Levy M. Relationship between caffeine concentrations in plasma and saliva. Clin Pharmacol Ther 1984; 36: 133-7. 20 Bianchetti MG, Kraemer R, Passweg 3, Jost J, Preisig R. Use of salivarv levels to predict clearance of caffeine in patients with cystic ibrosis. J Ped Gastroenterol Nutr 1988; 7: 688-93. 21 Newton R, Broughton LJ, Lind MJ, Morrison PJ, Rogers HJ, Bradbrook ID. Plasma and salivary pharmacokinetics of caffeine in man. Eur J Clin Pharmacoll981; 21: 45-52. 22 Statland BE, Demas T, Danis M. Caffeiue accumulation associated with alcoholic liver disease. N Engl J Med 1976; 295: 110-I. 23 Stavric B. Methylxanthines: toxicity to humans. 2. caffeine. Food Chem Tox 1988; 26: 645-62. 24 Villeneuve JP, Infante-Rivard C, Ampelas M, Pomier-Layrargues G, Huet PM, Marleau D. Prognostic value of the aminopyrine breath test in cirrhotic patients. Hepatology 1986; 6: 928-31. 25 Merkel C, Bolognesi M, Finucci GF, et al. Indocyanine green intrinsic hepatic clearance as a prognostic index of survival in patients with cirrhosis. J Hepatology 1989; 9: 16-22. 26 Albers I, Hartmann H, Bircher J, Creutzfeldt W. Superiority of the Child-Pugh classification to quantitative liver function tests for assessing prognosis of liver cirrhosis. Stand J Gastroenterol 1989; 24: 269-76.