- Email: [email protected]
Plasma catecholamine level and portal venous pressure as guides to prognosis in patients with cirrhosis
Journal of Hepatology, 1988; 6:350-358 Elsevier
350 HEP 00403
Plasma catecholamine level and portal venous pressure as guides to prognosis in patients with cirrhosis
U. Tage-Jensen, J.H. Henriksen, E. Christensen, A. Widding, H. Ring-Larsen and N. Juel Christensen Departments of Clinical Physiology and Hepatology, Hvidovre Hospital, and Department of lnternal Medicine and Endocrinology, Herlev Hospital, Department of Medicine F, Gentofte Hospital, Universityof Copenhagen, Copenhagen (Denmark) (Received 11 August 1987) (Accepted 8 December 1987)
Summary Circulating noradrenaline is increased in patients with cirrhosis, especially in decompensated patients with ascites. Eighty-one patients with alcoholic cirrhosis were followed for up to 8 years in order to establish a possible relationship between plasma catecholamines, haemodynamics, and routine clinical and biochemical variables and survival. Forty-seven (58%) of the patients died during the follow-up period. Univariate analysis showed that plasma noradrenaline and adrenaline concentrations, portal pressure, indocyanine green clearance, serum sodium, bilirubin, and albumin concentrations, and the presence of ascites or cardiovascular disease were of significant prognostic value. In a multivariate analysis (Cox regression model), plasma noradrenaline concentration, portal pressure, serum bilirubin concentration, and the presence of ascites and cardiovascular disease remained significant independent predictors of survival. The results suggest that determination of the circulating level of noradrenaline and portal pressure may add to the prognostic information on survival obtained from routine tests. Thus, the activity of the sympathetic nervous system may indicate the severity of cirrhosis with respect to survival.
Introduction In medically treated cirrhosis, use of the ChildTurcotte criteria is a simple way of assessing prognosis [1]. However, it has been clearly shown that there are other variables, not significantly correlated to the
Child-Turcotte criteria, but with important additional prognostic information [1]. Thus, Arroyo et al. [2] and Genoud et al. [3] found that plasma renin activity is a prognostic indicator in patients with cirrhosis and ascites. Another powerful homeostatic factor, the sympathetic nervous system, has been the subject of
Correspondence: Jens H. Henriksen, M.D., Dept. of Clinical Physiology,239, Hvidovre Hospital, DK-2650 Hvidovre, Denmark. 0168-8278/88/$03.50 © 1988 Elsevier Science Publishers B.V. (Biomedical Division)
NORADRENALINE AND SURVIVAL IN CIRRHOSIS
investigation [4-7]. Recently, Liach et al. [8] demonstrated that plasma noradrenaline (NA) is a predictor of survival in ascitic patients, and in patients with congestive heart failure Cohn et al. [9] found that plasma NA was highly predictive of subsequent mortality. The objective of the present study was, therefore, to analyse whether the determination of plasma levels of catecholamines and haemodynamic variables might add any further long-term prognostic information in patients with cirrhosis. We studied compensated as well as decompensated patients.
351
Study population and Methods Patients
In a 4 year period, patients who were referred to the Department of Clinical Physiology for diagnostic liver vein catheterization also had blood samples taken for determination of plasma catecholamines. In 81 patients cirrhosis was demonstrated histologically, and these patients comprise the study population of the present study. Diagnostic liver vein catheterization was successful in 74 patients and clinical as well as biochemical data were collected at the time of en-
TABLE 1 VARIABLES SELECTED AT ENTRY IN 81 CIRRHOTIC PATIENTS AND CORRELATED TO SURVIVAL USING UNIVARIATE ANALYSIS (LOG-RANK TEST)
Variables
No. of patients
Positive for:
Median (range)
P-value
n
(%)
81 81
63 47
(78)
Clinical alcoholism cardiac disease myocardial infarction ascites former bleeding episodes
81 81 81 81 81
76 7 3 59 24
(94) (8) (4) (73) (30)
Biochemical albumin O~mol/I, 540-800) a bilirubin (~mol/I, 2-17) aspartate aminotransferase (U/I, 10-40) alkaline phosphatase (U/I, 50-275) coagulation factor 2,7,10 (index, 0.7-1.3) s-creatinine (,umol/I, 49-121 ) s-sodium (mmol/l, 136-146) s-potassium (mmol/I, 3.5-5.0) p-noradrenaline (nmol/I, 0.5-2.5) p-adrenaline nmol/l, 0.0-0.6)
79 79 79 78 78 76 78 79 79 79
426 q233-714) 26 (6-350) 59~ 16-558) 365,132-2300) 0.49,0.22-1.21) 79142-292) 138,115-145) 3.8,3.0-4.9) 2.510.6-11.1) 0.38 0.05-2.0)
ICG clearance (ml/min, 300-700)
74 74 67 59 55
25 (4-62) 16 (2-44) 84 (60-105) 1.37 (0.5-2.3) 1.82 (51-998)
0.002 0.0007 0.55 0.47 0.02
galactose elimination capacity (mmol/min: males <1.7; females < 1.4)
57
1.54 (0.89-2.49)
0.90
General age (years) males duration ofhistory (months) deaths
Physiological wedged hepatic vein pressure (mm Hg, ~<15) hepatic venous pressure gradient (ram Hg, ~<5) mean arterial blood pressure (ram Hg) hepatic blood flow (I/min, 0.5-2.3)
Reference interval.
54 (30-73) 12 (1-99)
0.02 0.53 -
(58)
0.49 <0.001 0.01 0.004 0.72 0.005 0.04 0.10 0.09 0.26 0.12 0.02 0.83 <0.0003 0.007
352 try. Nine patients were in Child-Turcotte group A, 25 in group B, and 43 in group C. Four patients could not be classified due to unsuccessful biochemical tests (see Table 1). Twenty-two patients were without fluid retention, and 59 had ascites. Serum creatinine concentration was normal in 69 patients, whereas increased values (median 148 ~mol/1, range 122-292) were present in 12 patients. Twenty-five patients were without diuretic treatment, 17 received spironolactone 100 rag/day, and 37 took, in addition to spironolactone, bumetamide 1-4 mg/day. Three patients had ischaemic heart disease (former myocardial infarction) and four patients had signs of alcoholic cardiomyopathy (i.e., cardiac decompensation or arrhythmia with echocardiographic signs of diffuse myocardial dysfunction without valvular disease or signs of myocardial ischaemia). None of the patients had arterial hypertension, episodes of hypoglycaemia, thyroid disorders or organic renal disease. All patients had been hospitalized for several days and were considered to be in a stable stage of their disease. All patients were abstaining from alcohol and were without withdrawal symptoms at the time of the catheterization. None of the patients had experienced recent gastrointestinal bleeding or had encephalopathy. Patients consented to participate in the investigations and the study was approved by the Hospital Ethics Committee. No complications or side-effects were seen. After entrance the patients were treated medically according to general clinical practice. No patient had surgical porta-caval anastomosis performed. Twenty-six of the patients attended a controlled randomized trial testing testosterone treatment vs placebo. The outcome of this study was that testosterone treatment had no significant influence on any of the variables tested, including survival [10]. Twenty-four patients were enrolled in a randomized study of peritoneovenous shunting vs. diuretic treatment of ascites. The long-term survival rates of these two treatments were identical. The study was concluded in 1986 and survivors have been followed-up for at least 3.5 years and up to 8 years. Autopsy was performed in 72% of cases in which a
U. TAGEJENSEN et al. TABLE 2 PRINCIPAL CAUSES OF DEATH AMONG 81 PATIENTS WITH CIRRHOSIS FOLLOWED FOR A MEDIAN OF 38 MONTHS (RANGE 1-97) Survivors were followed for a median of 42 months (range 36-97). n Hepatic" hepatic failure hepatic failure and Gl-bleeding GI-bleeding Non-hepatic cardiovascular respiratory insufficiency infections other h Total No. of dcaths
(°/~)
20 11 6
(43) (23) (13)
2 2 2 4
(4) (4) (4) (9)
47
(100)
" Hepatocellular carcinoma demonstrated in six patients. b Found dead at home (suicide. intoxication).
patient died, and 91% of the deaths occurred in a hospital, from which the recorded main cause of death was directly available (Table 2). For the remaining four patients, the cause was derived from the death certificates from the Danish Central Death Registry.
Blood sampling Blood sampling for determination of the plasma concentration of N A and adrenaline was performed a f t e r a n overnight fast with the patient in the supine position for at least 1 h. Arterial samples were obtained from a small indwelling polyethylene catheter in the brachial or femoral artery or as arterialized hand venous blood (45 °C heated hand, blood oxygen saturation above 92%), in order to avoid possible artefacts due to local catecholamine extraction [11]. Any vasodilatory drugs were withdrawn at least 48 h before blood sampling. Digoxin or diuretics were withheld on the day of the study. Plasma N A and adrenaline concentrations were determined by an enzymatic isotope derivative technique as described previously [ 12]. The serum concentrations of albumin, bilirubin, aspartate aminotransferase, alkaline phosphatase,
NORADRENALINE AND SURVIVAL IN CIRRHOSIS
353
coagulation factor 2,7,10 ('prothrombin index'), creatinine, sodium and potassium were determined by routine methods in an autoanalyser (SMAC, Technicon).
prising 55 of the patients was drawn. Using only the data of these patients the final Cox model was recalculated. The remaining 26 patients were divided into two groups according to their survival estimated from the recalculated model. The average estimated survival function in each of the two groups was compared with the corresponding observed survival curve using the one-sample log-rank test [21]. In addition, the product-moment correlation coefficient was used. P < 0.05 was considered significant.
Catheterization All patients were studied in the morning after an overnight fast and after at least 1 h in the supine position. Liver vein catheterization was performed under local anaesthesia in the resting supine patient under fluoroscopic control as described earlier [13]. Pressures were measured in the wedged and free hepatic vein position, in the right atrium, and in the femoral artery by a capitance transducer (Simonsen & Weel, Copenhagen) with the zero-pressure reference level at the mid-axillary line. Hepatic blood flow was determined by the indocyanine green (ICG) constant infusion technique [14]. ICG-clearance was measured as infusion rate divided by the arterial plasma concentration of ICG [14]. The functional liver cell mass was estimated by galactose elimination capacity (0.5 g galactose per kg body weight), as described by Tygstrup [15]. Statistical evaluation The relationship between variables and survival was analysed using the log-rank test [16]. For continuous variables three or four strata were defined according to round cut-off levels, dividing the patients in subgroups of reasonable size. Variables showing a significant association with survival were analysed further in a Cox multiple regression model [17] in order to determine which variables had independent associations with survival, using the same procedure as previously described in detail [18]. The assumption of proportional hazards was tested as described elsewhere [18,19]. In the multivariate analysis a few missing values (3.7%) were replaced by neutral estimates (i.e., the mean value of the variable in question) [20]. A Cox analysis including only patients with complete data gave regression coefficients which were not significantly different from those in the presented final Cox model. The Cox model obtained was validated using a split sample testing technique: a random 2/3 sample corn-
Results
Table 1 shows the association of the variables recorded with the subsequent survival in univariate analysis using the log-rank test. It appears that the following variables showed a significant association with a decreased survival: high age, the presence of ascites or cardiovascular disease, low serum albumin, high serum bilirubin, low serum sodium, high plasma NA, high plasma adrenaline concentration, high portal pressure, and low ICG-clearance. It should be noted that most patients with cardiovascular disease had a hepatic cause of death (Table 2). Figs. 1 and 2 show the highly significant relationship between survival and the circulating level of NA and portal pressure (determined as the hepatic venous pressure gra-
CUI'ItJLATIVE
SURV[gAL
l . O ~: o.e
q."! : .....
LI
,x:.
: : .......
0.4
: ......
]
: ....
L,
: -
a
.....
............
: .............
". . . . . . . . . . . . . . . . . . . . C i
, o.e
(NA)
.....
e.
0.6
PROBABILITY
[_.
- . . . . . . . .
' "
:.b . . . . . . . . . . .
d
0.0 0
1'
2'
i
;
6'
7' 8' YEARS
Fig. 1. Cumulative survival probability at different levels of circulating N A in patients with cirrhosis: a, N A ~ 1.5 nmol/l (n =
17): b, 1.5 < NA ~ 2.5 nmol/I (n = 29); c, 2.5 < NA ~< 5.0 nrnol/l (n = 24); d, NA above 5.0 nmol/l (n = 9). (~-test: P = 0.002 for heterogeneity, P = 0.0002 for trend.)
354
U. TAGE-JENSEN et al
CUHULAT]VE
1
.
0
0.8
:.:..
0.6
L, , , , "* .......... .
0.4
SURVIVAL
PROBABILITY
(HVPG)
~
"
"
~
I L.-:'-
1 "'"
"
G
..........................
b
: ........................
0.2
0.0 YEARS
Fig. 2. Cumulative survival probability at different levels of hepatic venous pressure gradient (HVPG) in patients with cirrhosis: a, HVPG ~< 12 mm Hg (n = 26); b, 12 < HVPG ~< 20 mm Hg (n = 31); c, HVPG > 20 mm Hg (n = 17). (Z2-test: P = 0.002 for heterogeneity, P = 0.0007 for trend.)
dient), respectively. Hepatic venous pressure gradient and plasma N A concentration were positively correlated (r = 0.47, P < 0.001). The results of the Cox regression analysis are shown in Table 3. The scorings presented have been
TABLE 3 REGRESSION COEFFICIENTS OF SIGNIFICANT PROGNOSTIC VARIABLES IN THE COX REGRESSION MODEL IN 81 PATIENTS WITH CIRRHOSIS Variable
Scoring
Regression coefficient
Standard Perror value"
Plasma NA concentration
Loglo (value in (nmol/l)
1.64
0.74
0.02
Hepatic venous pressure gradient
mm Hg
0.14
0.05
0.005
Wedged hepatic venous pressure
mmHg
-0.096
0.039
0.01
Serum bilirubin concentration
logm (value in /tmol/l)
1.16
0.46
0.01
used to fulfil the assumption of proportionality of the Cox model. It appears that the presence of cardiovascular disease or ascites, high serum bilirubin concentration, high plasma N A concentration, and a high hepatic venous pressure gradient are i n d e p e n d e n t indicators of decreased survival. The hepatic venous pressure gradient was significant only if the wedged hepatic pressure was also included in the model. Thus, the significance (and hence the predictive power) of the hepatic venous pressure gradient is increased if the level of the wedged hepatic venous pressure is adjusted for by being included in the model. (The negative sign of the coefficient for the wedged hepatic venous pressure is a consequence of a significant positive correlation of that variable with the hepatic venous pressure gradient [22].) The ChildTurcotte criteria did not show a significant contribution in the Cox regression analysis and were therefore not included in the final Cox model. For a single patient the information in the six significant variables in the Cox model can be c o m b i n e d to give a prognostic index (PI), being defined as the sum of the patient's variables, each being multiplied by its regression coefficient [18], A patient may, for example, have the following variables: plasma N A concentration 4.1 nmol/l (log N A = 0.61) hepatic venous pressure gradient = 20 mm Hg, wedged hepatic venous pressure = 28 mm Hg, serum bilirubin concentration = 9 ~mol/l (log bilirubin = 0.95), ascites
CUNULRTIVE
SURVIVRL
1.0
b~.'qT--"~-'--',
O. 8
L b, %
""~"............ : % . . . . 'L...,
~
0.45
0.01
Cardiovascular disease
present: 1 1.32 absent: 0
0.47
0.005
a Overall P-value of Cox regression model: 9 × 10 -7.
"'\
"":" - :..... : . . . . . . . .2...3. . . .
:
,,.__
I
L.
11" - k
0.2 __~_
present: 1 1.16 absent: 0
~3
. . . . . . . . . ~:__ 0.4
Ascites
(PI)
PROBRBILITY
o. o
0
%-L
~.
~.
5.3
,
.-'--_
l
2
3.3 % ..........
3
L--..~___
'
:"
4--3
,
,
," - - - - 7 -
4
5
6
- "-
7 YEflRS
Fig. 3. Estimated cumulative survival probability corresponding to different PI based on the obtained Cox regression model (see text and Table 3).
NORADRENALINE AND SURVIVAL IN CIRRHOSIS CU~LRTIVE SURVIVRL PROBRBILITY l.o ~ -'--c=v-&. . 0.8
~'~___~ l
" ~
I
o.~
O4
.
.
-%
0.2
.....
"t L
2 "L
0.0
L _
i
o
.
~_ . . . . .
'L
i YERRS
Fig. 4. Observed ( ) and estimated (.... ) survival functions of 26 patients divided according to estimated PI value: group 1, PI<:3.7, n=13, O = 6 , E = 3.9; group 2, Pl>~3.7, n=13, O= 10, E = 12.9. O and E are the observed and expected numbers dying in each group (~ = 1.8, d.f. = 2, P = 0.4). The estimated ~urvival functions and the PI values were obtained using a recalculated final Cox model based only on the data of the other 55 patients in the study. present and cardiovascular disease absent, PI = 1.64 × 0.61 + 0.14 × 20 - 0.096 x 28 + 1.16 × 0.95 + 1.16 x 1 + 1.32 x 0 = 3.3. In the present 81 patients the PI varied between 0.7 and 5.7, the median being 3.3. Fig. 3 illustrates the estimated survival curves for different values of the PI. Curve 3.3 illustrates the estimated cumulative survival probability which approximates the value of the patient presented above. For such a patient it appears that the estimated probability of surviving 1, 3 and 6 years is about 68%, 50% and 37%, respectively. The result of the validation analysis using the splitsample testing technique (see 'Statistical evaluation') is shown in Fig. 4. It appears that there is no significant difference between the observed and expected survival functions (P = 0.4).
Discussion
Before dealing with the present results, selection of the study population will be considered. The present study design was transversal with a longitudinal follow-up, using death as endpoint. Frorri Table 2, it can be seen that the cause of death was hepatic in at least 80% of cases. Therefore, death was applied as
355 the endpoint irrespective of the cause. Patients enrolled were those submitted to diagnostic liver vein catheterization. This procedure is routinely offered to all patients with suspected cirrhosis in order to determine the degree of portal hypertension, The catheterization was performed as an elective procedure in non-comatose patients without present or recent gastrointestinal bleeding. It is well-established that variceal bleeding is associated with a high mortality rate, only about 50% of patients being alive 3 months after the first bleeding [23,24]. However, the mortality decreases progressively with increasing time after the bleeding. As liver vein catheterization was performed in patients without recent bleeding, this may explain the missing relationship between former variceal bleeding and prognosis. Prothrombin index also had no statistically significant prognostic value in this study. This is probably due to the fact that most patients in the present study had relatively high and stable values of this test variable. A major finding in the present study was the highly significant prediction of survival by circulating N A and adrenaline. N A leaks from the synaptic cleft of sympathetic postganglionic neurons into plasma, where its concentration may reflect the integral neurotransmitter function and thereby the sympathetic nervous activity [25,26]. It is well-established that plasma N A is elevated in patients with decompensated heart disease [26,27]. Patients with acute myocardial infarction have very high values of plasma N A [28], and it has been speculated that a high plasma NA may reflect the severity of other diseases [26] e.g., chronic pulmonary disease [29]. Recently, several reports have shown that cirrhosis may be associated with enhanced sympathetic nervous activity [4-7], and it is well established that compensated patients have normal or moderately increased plasma NA, whereas decompensated patients with fluid retention and high portal pressure have raised plasma NA [7,30,31]. This is not due to decreased hepatic or overall clearance of N A in cirrhosis, but reflects an enhanced sympathetic nervous activity [13,32-34]. Different authors have found a positive correlation between portal pressure and circulating N A [13,31], and recently, Willett et al. [35] presented evidence
356
U. TAGE-JENSEN et al.
that sympathetic tone may modulate portal pressure in alcoholic cirrhosis. As plasma N A is high in patients with ascites [7,13,30], it might be expected that NA contains prognostic information, as shown by the log-rank analysis in the present study (Fig. 1). The multivariate analysis confirmed that the circulating level of N A is a prognostic variable containing predictive information on survival, independently of ascites and portal pressure. Thus, determination of plasma NA may add significantly to the prognostic information obtained by clinical and h a e m o d y n a m i c variables. The mechanism behind a link between survival and low plasma N A concentration is not clear, but it is conceivable that the sympathetic nervous tone may reflect the severity of the disease not only by being of importance to portal pressure, renal sodium retention, decreased kidney function and blood flow [7, 30-34], but also by, for example, indicating the degree of overall systemic h a e m o d y n a m i c disarrangement [26]. In fact, cirrhotic patients with increased plasma N A show evidence of impaired cardiovascular function [36]. Moreover, H6rtnagl et al. [37] and Lenz et al. 138] found evidence of a relationship between the peripheral vasodilatation in advanced disease and a compensatory sympathetic overactivity in patients with cirrhosis as earlier considered by Shaldon et al. [39]. Some of our patients received diuretics. Diuretic medication in itself only changes plasma NA and adrenaline slightly in supine patients with
cirrhosis [40]. H o w e v e r , intensive diuretic treatment may induce hypovolaemia, which may evoke renal failure. H y p o v o l a e m i a , a powerful stimulus of the sympathetic nervous system, may elevate plasma NA and thereby reflect the reduced survival in these patients. The present study demonstrates a significant prognostic value of portal venous pressure (as d e t e r m i n e d by the hepatic venous pressure gradient) independently of ascites, catecholamines and o t h e r biochemical variables. This is at variance with some earlier studies, but in accordance with others [8,41-44]. In contrast to several previous reports, the present study contained patients with a large range of portal pressure values ( 2 - 4 4 mm Hg). As stated by Conn [45], this may highly influence the analysis of portal pressure as a prognostic indicator in cirrhosis. The relationship between increased portal pressure and plasma N A has been considered above. The results suggest that d e t e r m i n a t i o n of the circulating level of N A and portal pressure may add to the prognostic information on survival o b t a i n e d from routine tests. Thus, the activity of the sympathetic nervous system may indicate the severity of cirrhosis with respect to survival. N A might be a candidate for refining prognostic estimates used in the timing [46] of complex therapeutic p r o c e d u r e s such as, for example, liver transplantation [47].
References
Sympathetic nervous activity and renal and systemic hemodynamics in cirrhosis: plasma norepinephrine concentration, hepatic extraction and renal release. Hepatology 1982; 2: 304-310. 6 Keller U, Gerber PPG, Buhler FR, Stauffacher W. Role of the splanchnic bed in extracting circulating adrenaline and noradrenaline in normal subjects and in patients with cirrhosis of the liver. Clin Sci 1984; 67: 45-49. 7 Henriksen JH, Ring-Larsen H, Christensen NJ. Sympathetic nervous activity in cirrhosis. A survey of plasma catecholamine studies. J Hepatol 1985; 1: 55-65. 8 Llach H, Gin6s P. Tito LI, et al. Prognostic factors in cirrhotics with ascites. J Hepatol 1986; 3 (suppl 1): 56. 9 Cohn JN, Levine TB, Olivari MT, et al. Plasma norepinephrine as a guide to prognosis in patients with chronic congestive heart failure. N Engl J Med 1984; 311: 819-823. 10 Gluud C., Henriksen JH, Copenhagen Study Group for Liver Disease. Liver haemodynamics and function in alcoholic cirrhosis. Relation to testosterone treatment and
1 Christensen E, Schlichting P, Faurholdt L, et al. Prognostic value of Child-Turcotte criteria in medically treated cirrhosis. Hepatology 1984; 3: 430-435. 2 Arroyo V, Bosch J, Gaya-Beltran J, et al. Plasma renin activity and urinary sodium excretion as prognostic indicators in nonazotemic cirrhosis with ascites. Ann Int Med 1981; 94: 198-201. 3 Genoud E. Gonvers J-J, Schaller M-D, Essinger P, Hailer E, Brunner H-R. Valeur pronostique du syst~me rdnincangiotensine dans la rdponse fi la restriction sodique et le pronostic de I'ascitc cirrhotiquc d'origine alcoolique. Schweiz Med Wochenschr 1986; 116: 463-469. 4 Bichet DG, Van Putten VJ, Schrier RW. Potential role of increased sympathetic activity in impaired sodium and water excretion in cirrhosis. N Engl J Med 1982: 307: 1552-1557. 5 Ring-Larsen H, Hesse B, Henriksen JH, Christensen NJ.
N O R A D R E N A L I N E AND S U R V I V A L IN CIRRHOSIS ethanol consumption. J Hepatol 1987; 4: 168-173. 11 Hilsted J, Christensen NJ, Madsbad S. Whole body clearance of norepinephrine. The significance of arterial sampling and surgical stress. J Clin Invest 1983; 71: 500-505. 12 Christensen N J, Vestergaard P, Sorensen T, Rafaelsen O. Cerebrospinal fluid adrenaline and noradrenaline in depressed patients. Acta Psychiatr Scand 1980; 61 : 178-182. 13 Henriksen JH, Ring-Larsen H, Kanstrup l-L, Christensen NJ. Splanchnic and renal elimination and release of catecholamines in cirrhosis. Evidence of enhanced sympathetic nervous activity in patients with cirrhosis. Gut 1984; 25: 1034-1043. 14 Henriksen JH, Winkler K. Hepatic blood flow determination. J Hepatol 1987; 4: 166-17(I. 15 Tygstrup N. Determination of the hepatic galactose elimination capacity after a single intravenous injection in man. Acta Physiol Scand 1963; 58: 162-172. 16 Peto R, Pike MC, Armitage P. Design and analysis of randomized clinical trials requiring prolonged observation of each patient. II. Analysis and examples. Br J Cancer 1977; 35: 1-39. 17 Cox DR. Regression models and life tables (with discussion). J R Stat Soc B 1972: 34: 187-2211. 18 Christensen E, Neubergcr J. Crowc J. Bcncficial effect of azathioprine and predicnon of prognosis in primary biliary cirrhosis. Final results of an intcrnational trial. Gastrocnterology 1985: 89: 1084-11191. 19 Christensen E. Multivariate survival analysis using Cox's rcgrcssion model. Hepatology 1987: 7: 1346-1358. 2/) Rosenmann RH, Brand RJ, Sholter RI, et al. Multivariate prediction of coronary heart disease during 8.5 year followup in the Western Collaborative group study. Am J Cardiol 1976; 37: 903-910. 21 Brcslow NE. Analysis of survival data under the proportional hazards model, lnt Stat Rev 1975; 43: 45-58. 22 Gordon RA. Issues ill multiple regression. Am .1 Sociol 1968: 73: 592-616. 23 Christensen E, Fauerholdt L, Schlichting P, et al. Aspects of the natural history of gastrointestinal bleeding in cirrhosis and the effect of prednisone. Gastroenterology 1981: 81: 944-952. 24 The Copenhagen Esophageal Varices Sclerotherapy Projcct. Sclerotherapy after first variceal hemorrhage in cirrhosis. N Engl J Med 1984; 311 : 1594-1600. 25 Silverberg AB, Shah SD, Haymond MW, Cryer PE. Norepinephrine, hormone and neurotransmitter in man. Am J Physiol 1978: 234: 252-256. 26 Christensen NJ, Henriksen O, Lassen NA. The Sympathoadrenal System. Physiology and Pathophysiology. Copenhagen: Munksgaard, 1986. 27 Hansen JF, Hesse B, Christensen NJ. Enhanced sympathetic nervous activity after intravenous propranolol in ischaemic heart disease. Plasma noradrenalinc, splanchnic blood flow and mixed venous oxygen saturation at rest and during exercise. Eur J Clin Invest 1978; 8:31-36. 28 Videbzek J, Christensen N J, Stendorff B. Serial determination of plasma catecholamines in myocardial infarctkm. Circulation 1972: 46: 846-855. 29 Henriksen JH, Christensen N J, Kok-Jensen A, Christiansen I. Enhanced sympathetic nervous activity in patients
357
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
with chronic obstructive lung disease. Scand J Clin Lab Invest 1980; 40: 419-427. Burghardt W, Wernze H, Schaffrath I. Changes of circulatory noradrenaline and adrenaline in hepatic cirrhosis. Relation to stage of disease, liver and renal function. Acta Endocrinol 1982; 99 (suppl 246): 100-101. Arroyo V. Phmas R, Gaya I, et al. Sympathetic nervous activity, renin-angiotcnsin system and renal excretion of prostaglandin E, in cirrhosis. Relation to functional renal failure and sodium and water cxcrction. Eur J Clin Invest 1983: 13: 271-278. Nicholls KM, Shapiro MD, Van Putten VJ, et al. Elevated plasma norepinephrine concentrations in dccompcnsatcd cirrhosis. Associations with increased secretion rates, normal clearance rates, and suppressibility by central blood volume expansion. Circ Res 1985; 56: 457-461. Wilier I, Esler M, Burke F, Leonard P. Dudlcy F. Total and renal sympathetic nervous system activity in alcoholic cirrhosis. J Hepatol 1985: 1: 639-648. Hcnrikscn JH, Ring-karscn H. Chrtstcnscn NJ. Hcpaticintcstinal uptake and release of cutccholammes in alcoholic cirrhosis. Evidence of enhanced hcpatic-intcstinal sympathetic nervous activity. Gut 1987: 28: 1637- 1642. Willett IR, Eslcr M, Jcnnings G, Dudley FJ. Sympathetic tone modulates portal venous pressure in alcoholic cirrhosis. Lancet 1986; ii: 939-942. Henriksen JH, Ring-Larscn H, Christcnscn NJ. Circulating noradrenalinc and central hacmodvnamics in patients with cirrhosis. Scand J Gastrocntcrol 1985: 211:1185-11911. H/Srtnagl H, Singer EA, Lenz K. Substancc P is markedly increased in plasma of patients with hepatic coma. Lancet 1984: i: 480-483. Lenz K, H6rtnagl H, Magometschnigg D, Klcinbcrger G, Druml W, Laggner A. Function of the autonomic nervous system in patients with hepatic encephalopathy. Hcpatology 1985: 5: 831-836. Shaldon C, Peacock JH. Walker RM. Portal venous content of adrenaline and noradrenaline in portal hypertension. Lancet 1961; i: 957-961. Ring-Larscn H, Henrikscn JH, Wilken C, Clausen I, Pals H, Christenscn NJ. Diuretic treatment in decompensated cirrhosis and congestive heart failure: effect of posture. Br Mcd J 1986: 292: 1351-1353. Lebrec D. Dc Fleury P. Rucff B, Nahum H. Benhamou J-P. Portal hypertcnsion, size of esophageal varices and risk of gastrointestinal bleeding in alcoholic cirrhosis. Gastroenterology 1980; 79:1139-1144. Vinel JP, Cassigneul I, Louis A, Levade M, Pascal JH. Clinical and prognostic significance of porto-hepatic gradient in patients with cirrhosis. Surg Gynecot Obstet 1982: 155: 347-352. Polio J, Groszmann RJ. Hcmodynamic factors involved in the development and rupture of esophageal varices: a pathophysiologic approach to treatment. Semin Liver Dis 1986; 6: 318-331. Gluud C, Hcnrikscn JH. Nielsen G. Copenhagen Study Group for Liver Disease~,. Prognostic indicators in alcoholic cirrhotic men. Hepatology 1988: 8: (in prcssL Corm HO. The varix volcano connection. Gastroenterology 19811; 79: 1333-1337.
358 46 Christensen E, Schlichting P, Kragh Andersen P, et al. Updating prognosis and therapeutic effect evaluation in cirrhosis with Cox's multiple regression model for time-dependent variables. Scand .1 Gastroenterol 1986; 21: 163-174.
U. TAGE-JENSEN et al. 47 Neuberger J, Altman DG, Christensen E, et al. Use of a prognostic index in assessment and outcome of liver transplantation in patients with primary biliary cirrhosis. Transplantation 1986; 41: 713-716.
350 HEP 00403
Plasma catecholamine level and portal venous pressure as guides to prognosis in patients with cirrhosis
U. Tage-Jensen, J.H. Henriksen, E. Christensen, A. Widding, H. Ring-Larsen and N. Juel Christensen Departments of Clinical Physiology and Hepatology, Hvidovre Hospital, and Department of lnternal Medicine and Endocrinology, Herlev Hospital, Department of Medicine F, Gentofte Hospital, Universityof Copenhagen, Copenhagen (Denmark) (Received 11 August 1987) (Accepted 8 December 1987)
Summary Circulating noradrenaline is increased in patients with cirrhosis, especially in decompensated patients with ascites. Eighty-one patients with alcoholic cirrhosis were followed for up to 8 years in order to establish a possible relationship between plasma catecholamines, haemodynamics, and routine clinical and biochemical variables and survival. Forty-seven (58%) of the patients died during the follow-up period. Univariate analysis showed that plasma noradrenaline and adrenaline concentrations, portal pressure, indocyanine green clearance, serum sodium, bilirubin, and albumin concentrations, and the presence of ascites or cardiovascular disease were of significant prognostic value. In a multivariate analysis (Cox regression model), plasma noradrenaline concentration, portal pressure, serum bilirubin concentration, and the presence of ascites and cardiovascular disease remained significant independent predictors of survival. The results suggest that determination of the circulating level of noradrenaline and portal pressure may add to the prognostic information on survival obtained from routine tests. Thus, the activity of the sympathetic nervous system may indicate the severity of cirrhosis with respect to survival.
Introduction In medically treated cirrhosis, use of the ChildTurcotte criteria is a simple way of assessing prognosis [1]. However, it has been clearly shown that there are other variables, not significantly correlated to the
Child-Turcotte criteria, but with important additional prognostic information [1]. Thus, Arroyo et al. [2] and Genoud et al. [3] found that plasma renin activity is a prognostic indicator in patients with cirrhosis and ascites. Another powerful homeostatic factor, the sympathetic nervous system, has been the subject of
Correspondence: Jens H. Henriksen, M.D., Dept. of Clinical Physiology,239, Hvidovre Hospital, DK-2650 Hvidovre, Denmark. 0168-8278/88/$03.50 © 1988 Elsevier Science Publishers B.V. (Biomedical Division)
NORADRENALINE AND SURVIVAL IN CIRRHOSIS
investigation [4-7]. Recently, Liach et al. [8] demonstrated that plasma noradrenaline (NA) is a predictor of survival in ascitic patients, and in patients with congestive heart failure Cohn et al. [9] found that plasma NA was highly predictive of subsequent mortality. The objective of the present study was, therefore, to analyse whether the determination of plasma levels of catecholamines and haemodynamic variables might add any further long-term prognostic information in patients with cirrhosis. We studied compensated as well as decompensated patients.
351
Study population and Methods Patients
In a 4 year period, patients who were referred to the Department of Clinical Physiology for diagnostic liver vein catheterization also had blood samples taken for determination of plasma catecholamines. In 81 patients cirrhosis was demonstrated histologically, and these patients comprise the study population of the present study. Diagnostic liver vein catheterization was successful in 74 patients and clinical as well as biochemical data were collected at the time of en-
TABLE 1 VARIABLES SELECTED AT ENTRY IN 81 CIRRHOTIC PATIENTS AND CORRELATED TO SURVIVAL USING UNIVARIATE ANALYSIS (LOG-RANK TEST)
Variables
No. of patients
Positive for:
Median (range)
P-value
n
(%)
81 81
63 47
(78)
Clinical alcoholism cardiac disease myocardial infarction ascites former bleeding episodes
81 81 81 81 81
76 7 3 59 24
(94) (8) (4) (73) (30)
Biochemical albumin O~mol/I, 540-800) a bilirubin (~mol/I, 2-17) aspartate aminotransferase (U/I, 10-40) alkaline phosphatase (U/I, 50-275) coagulation factor 2,7,10 (index, 0.7-1.3) s-creatinine (,umol/I, 49-121 ) s-sodium (mmol/l, 136-146) s-potassium (mmol/I, 3.5-5.0) p-noradrenaline (nmol/I, 0.5-2.5) p-adrenaline nmol/l, 0.0-0.6)
79 79 79 78 78 76 78 79 79 79
426 q233-714) 26 (6-350) 59~ 16-558) 365,132-2300) 0.49,0.22-1.21) 79142-292) 138,115-145) 3.8,3.0-4.9) 2.510.6-11.1) 0.38 0.05-2.0)
ICG clearance (ml/min, 300-700)
74 74 67 59 55
25 (4-62) 16 (2-44) 84 (60-105) 1.37 (0.5-2.3) 1.82 (51-998)
0.002 0.0007 0.55 0.47 0.02
galactose elimination capacity (mmol/min: males <1.7; females < 1.4)
57
1.54 (0.89-2.49)
0.90
General age (years) males duration ofhistory (months) deaths
Physiological wedged hepatic vein pressure (mm Hg, ~<15) hepatic venous pressure gradient (ram Hg, ~<5) mean arterial blood pressure (ram Hg) hepatic blood flow (I/min, 0.5-2.3)
Reference interval.
54 (30-73) 12 (1-99)
0.02 0.53 -
(58)
0.49 <0.001 0.01 0.004 0.72 0.005 0.04 0.10 0.09 0.26 0.12 0.02 0.83 <0.0003 0.007
352 try. Nine patients were in Child-Turcotte group A, 25 in group B, and 43 in group C. Four patients could not be classified due to unsuccessful biochemical tests (see Table 1). Twenty-two patients were without fluid retention, and 59 had ascites. Serum creatinine concentration was normal in 69 patients, whereas increased values (median 148 ~mol/1, range 122-292) were present in 12 patients. Twenty-five patients were without diuretic treatment, 17 received spironolactone 100 rag/day, and 37 took, in addition to spironolactone, bumetamide 1-4 mg/day. Three patients had ischaemic heart disease (former myocardial infarction) and four patients had signs of alcoholic cardiomyopathy (i.e., cardiac decompensation or arrhythmia with echocardiographic signs of diffuse myocardial dysfunction without valvular disease or signs of myocardial ischaemia). None of the patients had arterial hypertension, episodes of hypoglycaemia, thyroid disorders or organic renal disease. All patients had been hospitalized for several days and were considered to be in a stable stage of their disease. All patients were abstaining from alcohol and were without withdrawal symptoms at the time of the catheterization. None of the patients had experienced recent gastrointestinal bleeding or had encephalopathy. Patients consented to participate in the investigations and the study was approved by the Hospital Ethics Committee. No complications or side-effects were seen. After entrance the patients were treated medically according to general clinical practice. No patient had surgical porta-caval anastomosis performed. Twenty-six of the patients attended a controlled randomized trial testing testosterone treatment vs placebo. The outcome of this study was that testosterone treatment had no significant influence on any of the variables tested, including survival [10]. Twenty-four patients were enrolled in a randomized study of peritoneovenous shunting vs. diuretic treatment of ascites. The long-term survival rates of these two treatments were identical. The study was concluded in 1986 and survivors have been followed-up for at least 3.5 years and up to 8 years. Autopsy was performed in 72% of cases in which a
U. TAGEJENSEN et al. TABLE 2 PRINCIPAL CAUSES OF DEATH AMONG 81 PATIENTS WITH CIRRHOSIS FOLLOWED FOR A MEDIAN OF 38 MONTHS (RANGE 1-97) Survivors were followed for a median of 42 months (range 36-97). n Hepatic" hepatic failure hepatic failure and Gl-bleeding GI-bleeding Non-hepatic cardiovascular respiratory insufficiency infections other h Total No. of dcaths
(°/~)
20 11 6
(43) (23) (13)
2 2 2 4
(4) (4) (4) (9)
47
(100)
" Hepatocellular carcinoma demonstrated in six patients. b Found dead at home (suicide. intoxication).
patient died, and 91% of the deaths occurred in a hospital, from which the recorded main cause of death was directly available (Table 2). For the remaining four patients, the cause was derived from the death certificates from the Danish Central Death Registry.
Blood sampling Blood sampling for determination of the plasma concentration of N A and adrenaline was performed a f t e r a n overnight fast with the patient in the supine position for at least 1 h. Arterial samples were obtained from a small indwelling polyethylene catheter in the brachial or femoral artery or as arterialized hand venous blood (45 °C heated hand, blood oxygen saturation above 92%), in order to avoid possible artefacts due to local catecholamine extraction [11]. Any vasodilatory drugs were withdrawn at least 48 h before blood sampling. Digoxin or diuretics were withheld on the day of the study. Plasma N A and adrenaline concentrations were determined by an enzymatic isotope derivative technique as described previously [ 12]. The serum concentrations of albumin, bilirubin, aspartate aminotransferase, alkaline phosphatase,
NORADRENALINE AND SURVIVAL IN CIRRHOSIS
353
coagulation factor 2,7,10 ('prothrombin index'), creatinine, sodium and potassium were determined by routine methods in an autoanalyser (SMAC, Technicon).
prising 55 of the patients was drawn. Using only the data of these patients the final Cox model was recalculated. The remaining 26 patients were divided into two groups according to their survival estimated from the recalculated model. The average estimated survival function in each of the two groups was compared with the corresponding observed survival curve using the one-sample log-rank test [21]. In addition, the product-moment correlation coefficient was used. P < 0.05 was considered significant.
Catheterization All patients were studied in the morning after an overnight fast and after at least 1 h in the supine position. Liver vein catheterization was performed under local anaesthesia in the resting supine patient under fluoroscopic control as described earlier [13]. Pressures were measured in the wedged and free hepatic vein position, in the right atrium, and in the femoral artery by a capitance transducer (Simonsen & Weel, Copenhagen) with the zero-pressure reference level at the mid-axillary line. Hepatic blood flow was determined by the indocyanine green (ICG) constant infusion technique [14]. ICG-clearance was measured as infusion rate divided by the arterial plasma concentration of ICG [14]. The functional liver cell mass was estimated by galactose elimination capacity (0.5 g galactose per kg body weight), as described by Tygstrup [15]. Statistical evaluation The relationship between variables and survival was analysed using the log-rank test [16]. For continuous variables three or four strata were defined according to round cut-off levels, dividing the patients in subgroups of reasonable size. Variables showing a significant association with survival were analysed further in a Cox multiple regression model [17] in order to determine which variables had independent associations with survival, using the same procedure as previously described in detail [18]. The assumption of proportional hazards was tested as described elsewhere [18,19]. In the multivariate analysis a few missing values (3.7%) were replaced by neutral estimates (i.e., the mean value of the variable in question) [20]. A Cox analysis including only patients with complete data gave regression coefficients which were not significantly different from those in the presented final Cox model. The Cox model obtained was validated using a split sample testing technique: a random 2/3 sample corn-
Results
Table 1 shows the association of the variables recorded with the subsequent survival in univariate analysis using the log-rank test. It appears that the following variables showed a significant association with a decreased survival: high age, the presence of ascites or cardiovascular disease, low serum albumin, high serum bilirubin, low serum sodium, high plasma NA, high plasma adrenaline concentration, high portal pressure, and low ICG-clearance. It should be noted that most patients with cardiovascular disease had a hepatic cause of death (Table 2). Figs. 1 and 2 show the highly significant relationship between survival and the circulating level of NA and portal pressure (determined as the hepatic venous pressure gra-
CUI'ItJLATIVE
SURV[gAL
l . O ~: o.e
q."! : .....
LI
,x:.
: : .......
0.4
: ......
]
: ....
L,
: -
a
.....
............
: .............
". . . . . . . . . . . . . . . . . . . . C i
, o.e
(NA)
.....
e.
0.6
PROBABILITY
[_.
- . . . . . . . .
' "
:.b . . . . . . . . . . .
d
0.0 0
1'
2'
i
;
6'
7' 8' YEARS
Fig. 1. Cumulative survival probability at different levels of circulating N A in patients with cirrhosis: a, N A ~ 1.5 nmol/l (n =
17): b, 1.5 < NA ~ 2.5 nmol/I (n = 29); c, 2.5 < NA ~< 5.0 nrnol/l (n = 24); d, NA above 5.0 nmol/l (n = 9). (~-test: P = 0.002 for heterogeneity, P = 0.0002 for trend.)
354
U. TAGE-JENSEN et al
CUHULAT]VE
1
.
0
0.8
:.:..
0.6
L, , , , "* .......... .
0.4
SURVIVAL
PROBABILITY
(HVPG)
~
"
"
~
I L.-:'-
1 "'"
"
G
..........................
b
: ........................
0.2
0.0 YEARS
Fig. 2. Cumulative survival probability at different levels of hepatic venous pressure gradient (HVPG) in patients with cirrhosis: a, HVPG ~< 12 mm Hg (n = 26); b, 12 < HVPG ~< 20 mm Hg (n = 31); c, HVPG > 20 mm Hg (n = 17). (Z2-test: P = 0.002 for heterogeneity, P = 0.0007 for trend.)
dient), respectively. Hepatic venous pressure gradient and plasma N A concentration were positively correlated (r = 0.47, P < 0.001). The results of the Cox regression analysis are shown in Table 3. The scorings presented have been
TABLE 3 REGRESSION COEFFICIENTS OF SIGNIFICANT PROGNOSTIC VARIABLES IN THE COX REGRESSION MODEL IN 81 PATIENTS WITH CIRRHOSIS Variable
Scoring
Regression coefficient
Standard Perror value"
Plasma NA concentration
Loglo (value in (nmol/l)
1.64
0.74
0.02
Hepatic venous pressure gradient
mm Hg
0.14
0.05
0.005
Wedged hepatic venous pressure
mmHg
-0.096
0.039
0.01
Serum bilirubin concentration
logm (value in /tmol/l)
1.16
0.46
0.01
used to fulfil the assumption of proportionality of the Cox model. It appears that the presence of cardiovascular disease or ascites, high serum bilirubin concentration, high plasma N A concentration, and a high hepatic venous pressure gradient are i n d e p e n d e n t indicators of decreased survival. The hepatic venous pressure gradient was significant only if the wedged hepatic pressure was also included in the model. Thus, the significance (and hence the predictive power) of the hepatic venous pressure gradient is increased if the level of the wedged hepatic venous pressure is adjusted for by being included in the model. (The negative sign of the coefficient for the wedged hepatic venous pressure is a consequence of a significant positive correlation of that variable with the hepatic venous pressure gradient [22].) The ChildTurcotte criteria did not show a significant contribution in the Cox regression analysis and were therefore not included in the final Cox model. For a single patient the information in the six significant variables in the Cox model can be c o m b i n e d to give a prognostic index (PI), being defined as the sum of the patient's variables, each being multiplied by its regression coefficient [18], A patient may, for example, have the following variables: plasma N A concentration 4.1 nmol/l (log N A = 0.61) hepatic venous pressure gradient = 20 mm Hg, wedged hepatic venous pressure = 28 mm Hg, serum bilirubin concentration = 9 ~mol/l (log bilirubin = 0.95), ascites
CUNULRTIVE
SURVIVRL
1.0
b~.'qT--"~-'--',
O. 8
L b, %
""~"............ : % . . . . 'L...,
~
0.45
0.01
Cardiovascular disease
present: 1 1.32 absent: 0
0.47
0.005
a Overall P-value of Cox regression model: 9 × 10 -7.
"'\
"":" - :..... : . . . . . . . .2...3. . . .
:
,,.__
I
L.
11" - k
0.2 __~_
present: 1 1.16 absent: 0
~3
. . . . . . . . . ~:__ 0.4
Ascites
(PI)
PROBRBILITY
o. o
0
%-L
~.
~.
5.3
,
.-'--_
l
2
3.3 % ..........
3
L--..~___
'
:"
4--3
,
,
," - - - - 7 -
4
5
6
- "-
7 YEflRS
Fig. 3. Estimated cumulative survival probability corresponding to different PI based on the obtained Cox regression model (see text and Table 3).
NORADRENALINE AND SURVIVAL IN CIRRHOSIS CU~LRTIVE SURVIVRL PROBRBILITY l.o ~ -'--c=v-&. . 0.8
~'~___~ l
" ~
I
o.~
O4
.
.
-%
0.2
.....
"t L
2 "L
0.0
L _
i
o
.
~_ . . . . .
'L
i YERRS
Fig. 4. Observed ( ) and estimated (.... ) survival functions of 26 patients divided according to estimated PI value: group 1, PI<:3.7, n=13, O = 6 , E = 3.9; group 2, Pl>~3.7, n=13, O= 10, E = 12.9. O and E are the observed and expected numbers dying in each group (~ = 1.8, d.f. = 2, P = 0.4). The estimated ~urvival functions and the PI values were obtained using a recalculated final Cox model based only on the data of the other 55 patients in the study. present and cardiovascular disease absent, PI = 1.64 × 0.61 + 0.14 × 20 - 0.096 x 28 + 1.16 × 0.95 + 1.16 x 1 + 1.32 x 0 = 3.3. In the present 81 patients the PI varied between 0.7 and 5.7, the median being 3.3. Fig. 3 illustrates the estimated survival curves for different values of the PI. Curve 3.3 illustrates the estimated cumulative survival probability which approximates the value of the patient presented above. For such a patient it appears that the estimated probability of surviving 1, 3 and 6 years is about 68%, 50% and 37%, respectively. The result of the validation analysis using the splitsample testing technique (see 'Statistical evaluation') is shown in Fig. 4. It appears that there is no significant difference between the observed and expected survival functions (P = 0.4).
Discussion
Before dealing with the present results, selection of the study population will be considered. The present study design was transversal with a longitudinal follow-up, using death as endpoint. Frorri Table 2, it can be seen that the cause of death was hepatic in at least 80% of cases. Therefore, death was applied as
355 the endpoint irrespective of the cause. Patients enrolled were those submitted to diagnostic liver vein catheterization. This procedure is routinely offered to all patients with suspected cirrhosis in order to determine the degree of portal hypertension, The catheterization was performed as an elective procedure in non-comatose patients without present or recent gastrointestinal bleeding. It is well-established that variceal bleeding is associated with a high mortality rate, only about 50% of patients being alive 3 months after the first bleeding [23,24]. However, the mortality decreases progressively with increasing time after the bleeding. As liver vein catheterization was performed in patients without recent bleeding, this may explain the missing relationship between former variceal bleeding and prognosis. Prothrombin index also had no statistically significant prognostic value in this study. This is probably due to the fact that most patients in the present study had relatively high and stable values of this test variable. A major finding in the present study was the highly significant prediction of survival by circulating N A and adrenaline. N A leaks from the synaptic cleft of sympathetic postganglionic neurons into plasma, where its concentration may reflect the integral neurotransmitter function and thereby the sympathetic nervous activity [25,26]. It is well-established that plasma N A is elevated in patients with decompensated heart disease [26,27]. Patients with acute myocardial infarction have very high values of plasma N A [28], and it has been speculated that a high plasma NA may reflect the severity of other diseases [26] e.g., chronic pulmonary disease [29]. Recently, several reports have shown that cirrhosis may be associated with enhanced sympathetic nervous activity [4-7], and it is well established that compensated patients have normal or moderately increased plasma NA, whereas decompensated patients with fluid retention and high portal pressure have raised plasma NA [7,30,31]. This is not due to decreased hepatic or overall clearance of N A in cirrhosis, but reflects an enhanced sympathetic nervous activity [13,32-34]. Different authors have found a positive correlation between portal pressure and circulating N A [13,31], and recently, Willett et al. [35] presented evidence
356
U. TAGE-JENSEN et al.
that sympathetic tone may modulate portal pressure in alcoholic cirrhosis. As plasma N A is high in patients with ascites [7,13,30], it might be expected that NA contains prognostic information, as shown by the log-rank analysis in the present study (Fig. 1). The multivariate analysis confirmed that the circulating level of N A is a prognostic variable containing predictive information on survival, independently of ascites and portal pressure. Thus, determination of plasma NA may add significantly to the prognostic information obtained by clinical and h a e m o d y n a m i c variables. The mechanism behind a link between survival and low plasma N A concentration is not clear, but it is conceivable that the sympathetic nervous tone may reflect the severity of the disease not only by being of importance to portal pressure, renal sodium retention, decreased kidney function and blood flow [7, 30-34], but also by, for example, indicating the degree of overall systemic h a e m o d y n a m i c disarrangement [26]. In fact, cirrhotic patients with increased plasma N A show evidence of impaired cardiovascular function [36]. Moreover, H6rtnagl et al. [37] and Lenz et al. 138] found evidence of a relationship between the peripheral vasodilatation in advanced disease and a compensatory sympathetic overactivity in patients with cirrhosis as earlier considered by Shaldon et al. [39]. Some of our patients received diuretics. Diuretic medication in itself only changes plasma NA and adrenaline slightly in supine patients with
cirrhosis [40]. H o w e v e r , intensive diuretic treatment may induce hypovolaemia, which may evoke renal failure. H y p o v o l a e m i a , a powerful stimulus of the sympathetic nervous system, may elevate plasma NA and thereby reflect the reduced survival in these patients. The present study demonstrates a significant prognostic value of portal venous pressure (as d e t e r m i n e d by the hepatic venous pressure gradient) independently of ascites, catecholamines and o t h e r biochemical variables. This is at variance with some earlier studies, but in accordance with others [8,41-44]. In contrast to several previous reports, the present study contained patients with a large range of portal pressure values ( 2 - 4 4 mm Hg). As stated by Conn [45], this may highly influence the analysis of portal pressure as a prognostic indicator in cirrhosis. The relationship between increased portal pressure and plasma N A has been considered above. The results suggest that d e t e r m i n a t i o n of the circulating level of N A and portal pressure may add to the prognostic information on survival o b t a i n e d from routine tests. Thus, the activity of the sympathetic nervous system may indicate the severity of cirrhosis with respect to survival. N A might be a candidate for refining prognostic estimates used in the timing [46] of complex therapeutic p r o c e d u r e s such as, for example, liver transplantation [47].
References
Sympathetic nervous activity and renal and systemic hemodynamics in cirrhosis: plasma norepinephrine concentration, hepatic extraction and renal release. Hepatology 1982; 2: 304-310. 6 Keller U, Gerber PPG, Buhler FR, Stauffacher W. Role of the splanchnic bed in extracting circulating adrenaline and noradrenaline in normal subjects and in patients with cirrhosis of the liver. Clin Sci 1984; 67: 45-49. 7 Henriksen JH, Ring-Larsen H, Christensen NJ. Sympathetic nervous activity in cirrhosis. A survey of plasma catecholamine studies. J Hepatol 1985; 1: 55-65. 8 Llach H, Gin6s P. Tito LI, et al. Prognostic factors in cirrhotics with ascites. J Hepatol 1986; 3 (suppl 1): 56. 9 Cohn JN, Levine TB, Olivari MT, et al. Plasma norepinephrine as a guide to prognosis in patients with chronic congestive heart failure. N Engl J Med 1984; 311: 819-823. 10 Gluud C., Henriksen JH, Copenhagen Study Group for Liver Disease. Liver haemodynamics and function in alcoholic cirrhosis. Relation to testosterone treatment and
1 Christensen E, Schlichting P, Faurholdt L, et al. Prognostic value of Child-Turcotte criteria in medically treated cirrhosis. Hepatology 1984; 3: 430-435. 2 Arroyo V, Bosch J, Gaya-Beltran J, et al. Plasma renin activity and urinary sodium excretion as prognostic indicators in nonazotemic cirrhosis with ascites. Ann Int Med 1981; 94: 198-201. 3 Genoud E. Gonvers J-J, Schaller M-D, Essinger P, Hailer E, Brunner H-R. Valeur pronostique du syst~me rdnincangiotensine dans la rdponse fi la restriction sodique et le pronostic de I'ascitc cirrhotiquc d'origine alcoolique. Schweiz Med Wochenschr 1986; 116: 463-469. 4 Bichet DG, Van Putten VJ, Schrier RW. Potential role of increased sympathetic activity in impaired sodium and water excretion in cirrhosis. N Engl J Med 1982: 307: 1552-1557. 5 Ring-Larsen H, Hesse B, Henriksen JH, Christensen NJ.
N O R A D R E N A L I N E AND S U R V I V A L IN CIRRHOSIS ethanol consumption. J Hepatol 1987; 4: 168-173. 11 Hilsted J, Christensen NJ, Madsbad S. Whole body clearance of norepinephrine. The significance of arterial sampling and surgical stress. J Clin Invest 1983; 71: 500-505. 12 Christensen N J, Vestergaard P, Sorensen T, Rafaelsen O. Cerebrospinal fluid adrenaline and noradrenaline in depressed patients. Acta Psychiatr Scand 1980; 61 : 178-182. 13 Henriksen JH, Ring-Larsen H, Kanstrup l-L, Christensen NJ. Splanchnic and renal elimination and release of catecholamines in cirrhosis. Evidence of enhanced sympathetic nervous activity in patients with cirrhosis. Gut 1984; 25: 1034-1043. 14 Henriksen JH, Winkler K. Hepatic blood flow determination. J Hepatol 1987; 4: 166-17(I. 15 Tygstrup N. Determination of the hepatic galactose elimination capacity after a single intravenous injection in man. Acta Physiol Scand 1963; 58: 162-172. 16 Peto R, Pike MC, Armitage P. Design and analysis of randomized clinical trials requiring prolonged observation of each patient. II. Analysis and examples. Br J Cancer 1977; 35: 1-39. 17 Cox DR. Regression models and life tables (with discussion). J R Stat Soc B 1972: 34: 187-2211. 18 Christensen E, Neubergcr J. Crowc J. Bcncficial effect of azathioprine and predicnon of prognosis in primary biliary cirrhosis. Final results of an intcrnational trial. Gastrocnterology 1985: 89: 1084-11191. 19 Christensen E. Multivariate survival analysis using Cox's rcgrcssion model. Hepatology 1987: 7: 1346-1358. 2/) Rosenmann RH, Brand RJ, Sholter RI, et al. Multivariate prediction of coronary heart disease during 8.5 year followup in the Western Collaborative group study. Am J Cardiol 1976; 37: 903-910. 21 Brcslow NE. Analysis of survival data under the proportional hazards model, lnt Stat Rev 1975; 43: 45-58. 22 Gordon RA. Issues ill multiple regression. Am .1 Sociol 1968: 73: 592-616. 23 Christensen E, Fauerholdt L, Schlichting P, et al. Aspects of the natural history of gastrointestinal bleeding in cirrhosis and the effect of prednisone. Gastroenterology 1981: 81: 944-952. 24 The Copenhagen Esophageal Varices Sclerotherapy Projcct. Sclerotherapy after first variceal hemorrhage in cirrhosis. N Engl J Med 1984; 311 : 1594-1600. 25 Silverberg AB, Shah SD, Haymond MW, Cryer PE. Norepinephrine, hormone and neurotransmitter in man. Am J Physiol 1978: 234: 252-256. 26 Christensen NJ, Henriksen O, Lassen NA. The Sympathoadrenal System. Physiology and Pathophysiology. Copenhagen: Munksgaard, 1986. 27 Hansen JF, Hesse B, Christensen NJ. Enhanced sympathetic nervous activity after intravenous propranolol in ischaemic heart disease. Plasma noradrenalinc, splanchnic blood flow and mixed venous oxygen saturation at rest and during exercise. Eur J Clin Invest 1978; 8:31-36. 28 Videbzek J, Christensen N J, Stendorff B. Serial determination of plasma catecholamines in myocardial infarctkm. Circulation 1972: 46: 846-855. 29 Henriksen JH, Christensen N J, Kok-Jensen A, Christiansen I. Enhanced sympathetic nervous activity in patients
357
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
with chronic obstructive lung disease. Scand J Clin Lab Invest 1980; 40: 419-427. Burghardt W, Wernze H, Schaffrath I. Changes of circulatory noradrenaline and adrenaline in hepatic cirrhosis. Relation to stage of disease, liver and renal function. Acta Endocrinol 1982; 99 (suppl 246): 100-101. Arroyo V. Phmas R, Gaya I, et al. Sympathetic nervous activity, renin-angiotcnsin system and renal excretion of prostaglandin E, in cirrhosis. Relation to functional renal failure and sodium and water cxcrction. Eur J Clin Invest 1983: 13: 271-278. Nicholls KM, Shapiro MD, Van Putten VJ, et al. Elevated plasma norepinephrine concentrations in dccompcnsatcd cirrhosis. Associations with increased secretion rates, normal clearance rates, and suppressibility by central blood volume expansion. Circ Res 1985; 56: 457-461. Wilier I, Esler M, Burke F, Leonard P. Dudlcy F. Total and renal sympathetic nervous system activity in alcoholic cirrhosis. J Hepatol 1985: 1: 639-648. Hcnrikscn JH, Ring-karscn H. Chrtstcnscn NJ. Hcpaticintcstinal uptake and release of cutccholammes in alcoholic cirrhosis. Evidence of enhanced hcpatic-intcstinal sympathetic nervous activity. Gut 1987: 28: 1637- 1642. Willett IR, Eslcr M, Jcnnings G, Dudley FJ. Sympathetic tone modulates portal venous pressure in alcoholic cirrhosis. Lancet 1986; ii: 939-942. Henriksen JH, Ring-Larscn H, Christcnscn NJ. Circulating noradrenalinc and central hacmodvnamics in patients with cirrhosis. Scand J Gastrocntcrol 1985: 211:1185-11911. H/Srtnagl H, Singer EA, Lenz K. Substancc P is markedly increased in plasma of patients with hepatic coma. Lancet 1984: i: 480-483. Lenz K, H6rtnagl H, Magometschnigg D, Klcinbcrger G, Druml W, Laggner A. Function of the autonomic nervous system in patients with hepatic encephalopathy. Hcpatology 1985: 5: 831-836. Shaldon C, Peacock JH. Walker RM. Portal venous content of adrenaline and noradrenaline in portal hypertension. Lancet 1961; i: 957-961. Ring-Larscn H, Henrikscn JH, Wilken C, Clausen I, Pals H, Christenscn NJ. Diuretic treatment in decompensated cirrhosis and congestive heart failure: effect of posture. Br Mcd J 1986: 292: 1351-1353. Lebrec D. Dc Fleury P. Rucff B, Nahum H. Benhamou J-P. Portal hypertcnsion, size of esophageal varices and risk of gastrointestinal bleeding in alcoholic cirrhosis. Gastroenterology 1980; 79:1139-1144. Vinel JP, Cassigneul I, Louis A, Levade M, Pascal JH. Clinical and prognostic significance of porto-hepatic gradient in patients with cirrhosis. Surg Gynecot Obstet 1982: 155: 347-352. Polio J, Groszmann RJ. Hcmodynamic factors involved in the development and rupture of esophageal varices: a pathophysiologic approach to treatment. Semin Liver Dis 1986; 6: 318-331. Gluud C, Hcnrikscn JH. Nielsen G. Copenhagen Study Group for Liver Disease~,. Prognostic indicators in alcoholic cirrhotic men. Hepatology 1988: 8: (in prcssL Corm HO. The varix volcano connection. Gastroenterology 19811; 79: 1333-1337.
358 46 Christensen E, Schlichting P, Kragh Andersen P, et al. Updating prognosis and therapeutic effect evaluation in cirrhosis with Cox's multiple regression model for time-dependent variables. Scand .1 Gastroenterol 1986; 21: 163-174.
U. TAGE-JENSEN et al. 47 Neuberger J, Altman DG, Christensen E, et al. Use of a prognostic index in assessment and outcome of liver transplantation in patients with primary biliary cirrhosis. Transplantation 1986; 41: 713-716.