176
0
JournalofHepotology, 1992; 14: 176-182 1992Elsevier SciencePublishersB.V. All rights reserved. 0168.8278/92/$05.00
HEPAT 00960
Selenium in chronic liver
P. J.
Thuluvath and
D
. R. Triger
Department of hfedicine, Universi@of Sheffield Medical School, Sheffield, United Kingdom
(Received 7 August 1990)
In order to assess the role of selenium (Se) in chronic liver disease, we have measured serum, urinary and hepatic selenium in a range of liver diseases and correlated them with nutritional status and conventional liver biochemistry. Serum Se levels &g/l + SD.) were significantly lower in both alcoholic (63.6 f 18.2,~ C 0.0001) and non-alcoholic liver disease (NALD) (60.6 + 13.6, p c 0.0001) compared to healthy controls (87.8 + 21.2) and non-malignant ‘disease controls’ (80.3 + 19.1). Hepatic Se levels (j@g of dry weight) were also reduced in both ALD (0.568 + 0.647, p C 0.005) and NALD (0.863 + 0.308, p c 0.005) compared to controls (1.227 f 0.296), 24-h urinary Se excretion bg) in ALD (24.6 f 10.7) and NALD (29.0 f 14.3) was similar to controls (30.3 f 8.7). Serum Se showed a highly significant positive correlation with albumin 6.1c 0.001) in both ALD and NALD. Serum levels were also significantly correlated with anthropometric measurements.
Dietary assessment of patients with primary biliary cirrhosis and low serum Se levels did not show a reduced dietary intake. Our data show that Se levels are low in liver disease irrespective of aetiology and suggest that these low levels are more likely to be related to overall nutritional status than to dietary intake.
In 1957 Schwarz and Foltz reported that necrotic liver degeneration in rats could be prevented by oral administration of various selenium compounds (1). The biochemical basis for the importance of selenium was established in 1973when it was identified as a component of the active site of glutathione peroxidase (2). Reactive oxygen species can react with membrane lipids, proteins and DNA leading to their eventual disruption when anti-oxidant systems are overwhelmed (3). Selenium-dependent glutathione peroxidase, located in the cytosol and the mitochondrial matrix, interrupts the free radical peroxidative chain reaction by catalysing the reduction of hydrogen peroxide and organic hydroperoxides to water and hydroxyacids at the expense of reduced glutathione (GSH), which is oxidised to the disulfide (GSSG) (4,5). Selenium deficiency has been implicated in the pathogenesis of malignant disease (6-10) and ischaemic heart disease (1.1). A number of investigators have observed low serum selenium levels in alcoholic cirrhosis (12-16), but data on selenium status of patients with non-alcoholic cirrhosis are
insufficient and conflicting (12,13,17). The aims of this study were to establish the selenium status in a range of liver diseases and to determine the relationship between selenium and the nutritional status and also severity of liver damage.
Subjects and Methods
The study involved 132 consecutive unselected patients attending the liver clinic with biopsy proven chronic liver disease and 56 individuals, judged healthy on the basis of medical history, serving as controls. Subjects consuming excessive alcohol (> 40 g daily) or taking selenium or vitamin E supplementation were excluded from the control group. The control subjects and patients were non-vegetarian with the exception of four controls. A ‘disease control’ group of 46 patients with other clinical conditions was also studied; 17 had carcinoma (nine colon, five stomach, two pancreas and one breast), 17 inflammatory
Correspondence:Dr. DavidR. Triger, Departmentof Medicineand Pharmacology,M Floor, RoyalHallamshireHospital,Sheffield,SlO2JF, U.K.
SELENIUM IN CHRONIC LIVER DISEASE
177
bowel disease (ten ulcerative colitis and seven Crohns) and 12 other conditions (six benign colonic polyps, three gastric dysplasia, two chronic pancreatitis and one multiple sclerosis). Terminally ill patients were excluded. The age and sex of both controls and patients are given ir? Table 1, together with liver biochemistry, creatinine and haemoglobin values. Of the 98 patients with cirrhosis, 61 patients were Grade A, 31 Grade I3 and six Grade C, using the Child-Pugh classification (18). The respective figures for alcoholic cirrhosis were 25,18 and three; 34 patients with liver disease were taking regular diuretics (mostly spironolactone). Sixteen of nineteen patients with auto-immune CAH were receiving immunosuppressive drugs (prednisolone, azathioprine or combination). Ten patients with primary biliary cirrhosis were taking penicillamine . Informed consent was obtained from all subjects and the protocol was approved by the local ethical committee. Nutritional assessment Nutritional status was assessed by anthropometry (19-21). Right mid-arm circumference (MAC) was measured to the nearest cm with a tape at mid-point of the acromion and olecranon process. Triceps skinfold thickness (TSF) was measured (to the nearest mm) at the same point with a Harpenden caliper. Arm circumference was measured once; triceps skinfold thickness was measured three times and the mean of three readings was taken. All measurements were done by one person (P.J.T.). Mid-arm muscle circumference (MA&E), arm muscle area (AMA) and arm fat area (AFA) were calculated from TSF thickness and MAC. The standards of Frisancho (21) were used to compare the nutritional status; those who were below the 5th percentile were defined as undernourished. Blood was also taken at the same time for
the analysis of selenium, zinc, creatinine, liver function tests and blood count. Liver biochemistry and serum cream tinine were measured by a multi-channel auto-analyser (Technicon SRA 2000). Serum for the analysis of selenium was stored at -20 OCand tested in batches under code. Serum and urine selenium measurement Selenium was measured by hydride generation atomic absorption spectrophotometry (HG-AAS) using SP 9 series atomic absorption spectrophotometer (Pye Unicam) and PU 9060 continuous flow vapour system (Pye Unicam). Preliminary experiments had shown that the accuracy and precision of our technique was comparable to that described elsewhere (22). All specimens were analysed in duplicate. Urine selenium was measured by the same technique on an aliquot from a 24-h urine collection. Serial serum measurements Serial measurements (2-9 years) were performed on stored sera from nine patients with primary biliary cirrhosis and four with alcoholic cirrhosis. Hepatic selenium determination Liver selenium was measured on 12 cirrhotics with nonalcoholic liver disease (primary biliary cirrhosis four, sclerosing cholangitis two, chronic active hepatitis two, cryptogenic cirrhosis two and one each with secondary biliary cirrhosis and Wilson’s disease) and 10 subjects with alcoholic cirrhosis. In 15 cases liver selenium was measured on post-mortem specimens; the remaining seven were either surgical or needle biopsy specimens. Autopsy liver specimens from 20 subjects who died suddenly and who had no histological evidence of liver disease were used as controls. me specimens were weighed and then freeze-dried until a steady weight was attained. The specimens were
TABLE 1 Demographic data and laboratory values of subjects studied Number
Age (yr)
Sex (M)
Albumin (35-53) gn
Bil (2-20) poln
ALT (7-45) U/l
AlkPhos (3.5-105) U/l
Creat (60-120) pal;?
2;16)
Controls Liver disease (132) Alcoholic liver disease Primary biliary cirrhosis Cryptogenic cirrhosis
56
46.6 + 15.5
23133
51 45 8
47.1; + 10.8 58.6 f 9.5 49.6 + 20.3
34/17 2143 315
38.2 f 7.4 35.8 f 6.0 32.6 i 7.4
31 f 58 32 + 40 44 + 45
46 + 36 84 + 59 76 ? 40
131 f 70 490 +- 324 307 f 213
87 + 26 83k31 75 f 19
13.2 i 1.9 12.5 f 1.4 11.7 * 1.8
Chronic active hepatitis Auto-immune Viral
19 9
53.5 f 14.7 39.4 + 10.0
2117 8/I
39.8 f 3.9 41.4 + 5.7
1128 29 + 52
53+45 108 + 57
108 f 42 107 ? 69
812 16 84 & 9
13.1 t 1.9 14.8 f 2.3
17 29
68.8 f 9.2 51.0 f 18.2
10/7 18111
34.7 f 6.4 39.7 f 5.5
572164 9+4
47 -+ 85 37+39
184 f 305 130f 134
112 + 42 111+47
12.0 f 3.0 13.6 f 1.8
Disease controls (46) Carcinoma Miscellaneous
Normal range values for the Iabordtory are expressed in parentheses. All values expressedas mean + S.D.
P.J. THULUVATH
178 and stored in moisture-free COntainerS prior to analysis. Selenium was expressed as the concentration per g dry liver. then
hOmOgeniSed
Serum zinc estimation This was measured by flame atomic absorption using a modification of the method described by Dawson and Walker (23). For this determination, samples were tdken into Vacutainer Trace Element tubes (Becton Dickinson, Oxford, U.K.). Portalhypertension Clinical evidence of portal hypertension and porta-systemic shunting was obtained by endoscopic assessment for oesophageal varices, together with documented evidence of variceal haemorrhage. Data were available on 113 patients. Forty-eight had no evidence of oesophageal varices, 31 had varices which had not bled at the time of study, and in 34 variceal haemorrhage had occurred at some time prior to study. No patient had undergone decompression shunt surgery. Dietary assessment Daily dietary selenium intake was assessed in five female control subjects and 11 patients with stable primary biliary cirrhosis. This was conducted by a single dietitian. The diet taken over a period of one week (preceding the interview) was assessed and this was analysed using a database (Technicon SMAC II System). The assessment of selenium content of the food was based on Thorn et al. (24). Statiktics The data were expressed as mean + S.D. Paired and unpaired t-tests, Chi-square test, Mann-Whitney test, Pearson correlation coefficients and analysis of co-variante were used for statistical analysis as appropriate. TABLE
AND
D.R.
TRIGER
Results Serum selenium levels The precision of the technique was calculated by replicate analysis of pooled serum (No. of analyses 10, mean concentration 83 + 4.3 ng/ml, range 79.5-90 @ml, RSD 5.26%) and the accuracy was determined by recovery of added selenium (96% recovery, range 95-105%). The mean serum selenium levels in the control group was 87.8 f 21.2&l (range 51-134&l). Although the levels in female subjects were marginally lower than male subjects the difference was not significant either in controls (males Se 94.3 + 21.3 vs. females 83.4 _+20.3, t value = 1.91) or in alcoholic cirrhosis (males Se 65.0 + 19.3 vs. females 60.4 + 15.5, t value = 0.82). The trend towards reduction in selenium with increasing age (r = -0.11) was also not significant. The mean serum selenium levels in smokers (67.4 pg/l, n = 64) was not significantkj different from non-smokers (71.5&l, n = 131, tvalue 1.31,~ = 0.19). Serum levels were significantly lower in both alcoholic and non-alcoholic liver disease compared to the controls (Table 2). Patients with liver disease as a group had significantly lower serum levels than non-malignant disease controls (63.9 f 17.2 vs. 79.5 + 18.8,~ < 0.001). The levels in alcoholic (63.6 f 18.2 pg/l) and non-alcoholic liver disease (64.1 + 16.6 &l, t value 0.17) were similar. The serum levels, when all patients were considered, showed a modest correlation with albumin (n = 170, r = 0.401, p < 0.001, Fig. l), and while the correlation with alkaline phosphatase was very weak (r = -0.190, n = 160, p < 0.01). The correlations with bilirubin (r = -0.061), AST (r = -0.219) and ALT (r = 0.002) were not significant. Analysis of variance showed that a significant rt!ationship existed only between selenium and albumin {entire group, p C 0.001; liver disease, p < 0.001) but not with other biochemical parameters. Selenium levels did not correlate with the severity of
2
Serumseleniumlevelsin disease compared with controls Group
Selenium @g/l)
t value
2-taila probability
mean (+ SD.) Controls Liver disease Alcoholic cirrhosis Primary biliary cirrhosis Cryptogenic cirrhosis Chronic active hepatitis Auto-immune Viral Disease controls r3arcinoma Yiscellaneous a p values (vs. controls).
87.8 + 21.2 27-103 27-93 39-80
63.6 5 18.2 60.6 f 13.6 59.3 + 13.3
7.43 6.17 3.69
0
JournalofHepotology, 1992; 14: 176-182 1992Elsevier SciencePublishersB.V. All rights reserved. 0168.8278/92/$05.00
HEPAT 00960
Selenium in chronic liver
P. J.
Thuluvath and
D
. R. Triger
Department of hfedicine, Universi@of Sheffield Medical School, Sheffield, United Kingdom
(Received 7 August 1990)
In order to assess the role of selenium (Se) in chronic liver disease, we have measured serum, urinary and hepatic selenium in a range of liver diseases and correlated them with nutritional status and conventional liver biochemistry. Serum Se levels &g/l + SD.) were significantly lower in both alcoholic (63.6 f 18.2,~ C 0.0001) and non-alcoholic liver disease (NALD) (60.6 + 13.6, p c 0.0001) compared to healthy controls (87.8 + 21.2) and non-malignant ‘disease controls’ (80.3 + 19.1). Hepatic Se levels (j@g of dry weight) were also reduced in both ALD (0.568 + 0.647, p C 0.005) and NALD (0.863 + 0.308, p c 0.005) compared to controls (1.227 f 0.296), 24-h urinary Se excretion bg) in ALD (24.6 f 10.7) and NALD (29.0 f 14.3) was similar to controls (30.3 f 8.7). Serum Se showed a highly significant positive correlation with albumin 6.1c 0.001) in both ALD and NALD. Serum levels were also significantly correlated with anthropometric measurements.
Dietary assessment of patients with primary biliary cirrhosis and low serum Se levels did not show a reduced dietary intake. Our data show that Se levels are low in liver disease irrespective of aetiology and suggest that these low levels are more likely to be related to overall nutritional status than to dietary intake.
In 1957 Schwarz and Foltz reported that necrotic liver degeneration in rats could be prevented by oral administration of various selenium compounds (1). The biochemical basis for the importance of selenium was established in 1973when it was identified as a component of the active site of glutathione peroxidase (2). Reactive oxygen species can react with membrane lipids, proteins and DNA leading to their eventual disruption when anti-oxidant systems are overwhelmed (3). Selenium-dependent glutathione peroxidase, located in the cytosol and the mitochondrial matrix, interrupts the free radical peroxidative chain reaction by catalysing the reduction of hydrogen peroxide and organic hydroperoxides to water and hydroxyacids at the expense of reduced glutathione (GSH), which is oxidised to the disulfide (GSSG) (4,5). Selenium deficiency has been implicated in the pathogenesis of malignant disease (6-10) and ischaemic heart disease (1.1). A number of investigators have observed low serum selenium levels in alcoholic cirrhosis (12-16), but data on selenium status of patients with non-alcoholic cirrhosis are
insufficient and conflicting (12,13,17). The aims of this study were to establish the selenium status in a range of liver diseases and to determine the relationship between selenium and the nutritional status and also severity of liver damage.
Subjects and Methods
The study involved 132 consecutive unselected patients attending the liver clinic with biopsy proven chronic liver disease and 56 individuals, judged healthy on the basis of medical history, serving as controls. Subjects consuming excessive alcohol (> 40 g daily) or taking selenium or vitamin E supplementation were excluded from the control group. The control subjects and patients were non-vegetarian with the exception of four controls. A ‘disease control’ group of 46 patients with other clinical conditions was also studied; 17 had carcinoma (nine colon, five stomach, two pancreas and one breast), 17 inflammatory
Correspondence:Dr. DavidR. Triger, Departmentof Medicineand Pharmacology,M Floor, RoyalHallamshireHospital,Sheffield,SlO2JF, U.K.
SELENIUM IN CHRONIC LIVER DISEASE
177
bowel disease (ten ulcerative colitis and seven Crohns) and 12 other conditions (six benign colonic polyps, three gastric dysplasia, two chronic pancreatitis and one multiple sclerosis). Terminally ill patients were excluded. The age and sex of both controls and patients are given ir? Table 1, together with liver biochemistry, creatinine and haemoglobin values. Of the 98 patients with cirrhosis, 61 patients were Grade A, 31 Grade I3 and six Grade C, using the Child-Pugh classification (18). The respective figures for alcoholic cirrhosis were 25,18 and three; 34 patients with liver disease were taking regular diuretics (mostly spironolactone). Sixteen of nineteen patients with auto-immune CAH were receiving immunosuppressive drugs (prednisolone, azathioprine or combination). Ten patients with primary biliary cirrhosis were taking penicillamine . Informed consent was obtained from all subjects and the protocol was approved by the local ethical committee. Nutritional assessment Nutritional status was assessed by anthropometry (19-21). Right mid-arm circumference (MAC) was measured to the nearest cm with a tape at mid-point of the acromion and olecranon process. Triceps skinfold thickness (TSF) was measured (to the nearest mm) at the same point with a Harpenden caliper. Arm circumference was measured once; triceps skinfold thickness was measured three times and the mean of three readings was taken. All measurements were done by one person (P.J.T.). Mid-arm muscle circumference (MA&E), arm muscle area (AMA) and arm fat area (AFA) were calculated from TSF thickness and MAC. The standards of Frisancho (21) were used to compare the nutritional status; those who were below the 5th percentile were defined as undernourished. Blood was also taken at the same time for
the analysis of selenium, zinc, creatinine, liver function tests and blood count. Liver biochemistry and serum cream tinine were measured by a multi-channel auto-analyser (Technicon SRA 2000). Serum for the analysis of selenium was stored at -20 OCand tested in batches under code. Serum and urine selenium measurement Selenium was measured by hydride generation atomic absorption spectrophotometry (HG-AAS) using SP 9 series atomic absorption spectrophotometer (Pye Unicam) and PU 9060 continuous flow vapour system (Pye Unicam). Preliminary experiments had shown that the accuracy and precision of our technique was comparable to that described elsewhere (22). All specimens were analysed in duplicate. Urine selenium was measured by the same technique on an aliquot from a 24-h urine collection. Serial serum measurements Serial measurements (2-9 years) were performed on stored sera from nine patients with primary biliary cirrhosis and four with alcoholic cirrhosis. Hepatic selenium determination Liver selenium was measured on 12 cirrhotics with nonalcoholic liver disease (primary biliary cirrhosis four, sclerosing cholangitis two, chronic active hepatitis two, cryptogenic cirrhosis two and one each with secondary biliary cirrhosis and Wilson’s disease) and 10 subjects with alcoholic cirrhosis. In 15 cases liver selenium was measured on post-mortem specimens; the remaining seven were either surgical or needle biopsy specimens. Autopsy liver specimens from 20 subjects who died suddenly and who had no histological evidence of liver disease were used as controls. me specimens were weighed and then freeze-dried until a steady weight was attained. The specimens were
TABLE 1 Demographic data and laboratory values of subjects studied Number
Age (yr)
Sex (M)
Albumin (35-53) gn
Bil (2-20) poln
ALT (7-45) U/l
AlkPhos (3.5-105) U/l
Creat (60-120) pal;?
2;16)
Controls Liver disease (132) Alcoholic liver disease Primary biliary cirrhosis Cryptogenic cirrhosis
56
46.6 + 15.5
23133
51 45 8
47.1; + 10.8 58.6 f 9.5 49.6 + 20.3
34/17 2143 315
38.2 f 7.4 35.8 f 6.0 32.6 i 7.4
31 f 58 32 + 40 44 + 45
46 + 36 84 + 59 76 ? 40
131 f 70 490 +- 324 307 f 213
87 + 26 83k31 75 f 19
13.2 i 1.9 12.5 f 1.4 11.7 * 1.8
Chronic active hepatitis Auto-immune Viral
19 9
53.5 f 14.7 39.4 + 10.0
2117 8/I
39.8 f 3.9 41.4 + 5.7
1128 29 + 52
53+45 108 + 57
108 f 42 107 ? 69
812 16 84 & 9
13.1 t 1.9 14.8 f 2.3
17 29
68.8 f 9.2 51.0 f 18.2
10/7 18111
34.7 f 6.4 39.7 f 5.5
572164 9+4
47 -+ 85 37+39
184 f 305 130f 134
112 + 42 111+47
12.0 f 3.0 13.6 f 1.8
Disease controls (46) Carcinoma Miscellaneous
Normal range values for the Iabordtory are expressed in parentheses. All values expressedas mean + S.D.
P.J. THULUVATH
178 and stored in moisture-free COntainerS prior to analysis. Selenium was expressed as the concentration per g dry liver. then
hOmOgeniSed
Serum zinc estimation This was measured by flame atomic absorption using a modification of the method described by Dawson and Walker (23). For this determination, samples were tdken into Vacutainer Trace Element tubes (Becton Dickinson, Oxford, U.K.). Portalhypertension Clinical evidence of portal hypertension and porta-systemic shunting was obtained by endoscopic assessment for oesophageal varices, together with documented evidence of variceal haemorrhage. Data were available on 113 patients. Forty-eight had no evidence of oesophageal varices, 31 had varices which had not bled at the time of study, and in 34 variceal haemorrhage had occurred at some time prior to study. No patient had undergone decompression shunt surgery. Dietary assessment Daily dietary selenium intake was assessed in five female control subjects and 11 patients with stable primary biliary cirrhosis. This was conducted by a single dietitian. The diet taken over a period of one week (preceding the interview) was assessed and this was analysed using a database (Technicon SMAC II System). The assessment of selenium content of the food was based on Thorn et al. (24). Statiktics The data were expressed as mean + S.D. Paired and unpaired t-tests, Chi-square test, Mann-Whitney test, Pearson correlation coefficients and analysis of co-variante were used for statistical analysis as appropriate. TABLE
AND
D.R.
TRIGER
Results Serum selenium levels The precision of the technique was calculated by replicate analysis of pooled serum (No. of analyses 10, mean concentration 83 + 4.3 ng/ml, range 79.5-90 @ml, RSD 5.26%) and the accuracy was determined by recovery of added selenium (96% recovery, range 95-105%). The mean serum selenium levels in the control group was 87.8 f 21.2&l (range 51-134&l). Although the levels in female subjects were marginally lower than male subjects the difference was not significant either in controls (males Se 94.3 + 21.3 vs. females 83.4 _+20.3, t value = 1.91) or in alcoholic cirrhosis (males Se 65.0 + 19.3 vs. females 60.4 + 15.5, t value = 0.82). The trend towards reduction in selenium with increasing age (r = -0.11) was also not significant. The mean serum selenium levels in smokers (67.4 pg/l, n = 64) was not significantkj different from non-smokers (71.5&l, n = 131, tvalue 1.31,~ = 0.19). Serum levels were significantly lower in both alcoholic and non-alcoholic liver disease compared to the controls (Table 2). Patients with liver disease as a group had significantly lower serum levels than non-malignant disease controls (63.9 f 17.2 vs. 79.5 + 18.8,~ < 0.001). The levels in alcoholic (63.6 f 18.2 pg/l) and non-alcoholic liver disease (64.1 + 16.6 &l, t value 0.17) were similar. The serum levels, when all patients were considered, showed a modest correlation with albumin (n = 170, r = 0.401, p < 0.001, Fig. l), and while the correlation with alkaline phosphatase was very weak (r = -0.190, n = 160, p < 0.01). The correlations with bilirubin (r = -0.061), AST (r = -0.219) and ALT (r = 0.002) were not significant. Analysis of variance showed that a significant rt!ationship existed only between selenium and albumin {entire group, p C 0.001; liver disease, p < 0.001) but not with other biochemical parameters. Selenium levels did not correlate with the severity of
2
Serumseleniumlevelsin disease compared with controls Group
Selenium @g/l)
t value
2-taila probability
mean (+ SD.) Controls Liver disease Alcoholic cirrhosis Primary biliary cirrhosis Cryptogenic cirrhosis Chronic active hepatitis Auto-immune Viral Disease controls r3arcinoma Yiscellaneous a p values (vs. controls).
87.8 + 21.2 27-103 27-93 39-80
63.6 5 18.2 60.6 f 13.6 59.3 + 13.3
7.43 6.17 3.69
