Insertion of a transjugular intrahepatic porto-systemic shunt (TIPS) increases body cell mass (BCM) in patients with liver cirrhosis. The responsible mechanism is unidentified, but may involve changes in insulin sensitivity and glucose metabolism. Eleven patients with liver cirrhosis were examined before and 6 mo after a TIPS procedure with bioimpedance analyses, 2-h oral glucose tolerance tests, and two-step hyperinsulinemic euglycemic clamp with tracer-determined endogenous glucose production. After TIPS, BCM increased by 4.8 kg [confidence interval (CI): 2.7–7.3]. Fasting (f)-insulin increased from 123 ± 81 to 193 ± 124 pmol/l (P = 0.03), whereas f-glucose was unchanged (6.0 ± 0.8 vs. 6.2 ± 1.0 mmol/l). Glucose and insulin oral glucose tolerance test area under the curve increased by 14% (CI: 7–22%) and 53% (CI: 14–90%), respectively, P < 0.05. The C-peptide-to-insulin ratio decreased by 21% (CI: 8-35%, P = 0.01). Insulin sensitivity based on glucose infusion rate (4.69 ± 1.82 vs. 4.85 ± 2.37 mg·kg−1·min−1) and glucose tracer-based rate of disappearance were unchanged (5.01 ± 1.61 vs. 4.97 ± 2.13 mg·kg−1·min−1). Despite a further increase in peripheral hyperinsulinemia, f-endogenous glucose production did not change between study days (2.01 ± 0.42 vs. 2.42 ± 0.58 mg·kg−1·min−1) and was suppressed equally by insulin (1.1 ± 0.1 vs. 1.0 ± 0.1 mg·kg−1·min−1). Insulin clearance, growth hormone, cortisol, and glucagon levels were unchanged. BCM improvement did not correlate with the measured variables. After TIPS, BCM rose, despite enhanced hyperinsulinemia and aggravated glucose intolerance, but unchanged peripheral and hepatic insulin sensitivity. This apparent discrepancy may be ascribed to shunt-related decreased insulin exposure to the liver cells. However, the anabolic effect of TIPS seems not to be related to improvements in insulin sensitivity and remains mechanistically unexplained.
- hyperinsulinemic euglycemic clamp
- oral glucose tolerance test
- resting energy expenditure
- endogenous glucose production
malnutrition is common in patients with liver cirrhosis and strongly increases their morbidity and mortality (2, 6, 9). The mechanism is poorly understood, but may, among other factors, be related to reduced liver function, insufficient food intake, and hypermetabolism with increased resting energy expenditure (REE) (8, 29).
The majority of patients are glucose intolerant, and, in time, 30% develop diabetes (20, 48). Patients’ glucose metabolism is characterized by normal or slightly elevated fasting glucose, together with fasting hyperinsulinemia due to insulin resistance (23, 33, 35). These disturbances are often implicated in the development of hepatic malnutrition (18, 41).
Insertion of a transjugular intrahepatic porto-systemic shunt (TIPS) is reported to produce a remarkable gain in body cell mass (BCM) (4, 17, 38). We still do not know the mechanisms responsible for this favorable effect. It is not simply explained by increased food intake, and it is not due to anabolic changes in the growth hormone/insulin-like growth factor I system (17). Changes in glucose and insulin metabolism are likely to be involved, and previous studies have described immediate and sustained increases in peripheral insulin and glucagon after TIPS (19, 38, 40). Still, Stefankova et al. (50) detected no change in peripheral insulin sensitivity with a nontracer hyperinsulinemic euglycemic clamp 1 mo after TIPS.
TIPS is very effective in reducing portal pressure. Theoretically, the portal decompression may ultimately improve peripheral insulin sensitivity, as it prevents stressful events, such as variceal bleeding and peritonitis and their accompanying insulin resistance, and reduces low-grade inflammation by reducing bacterial translocation from the gastrointestinal tract (7, 27, 42, 52). The low-resistance passage shunts portal blood to the systemic circulation. Thus the normal first-pass metabolism of insulin is by-passed, and peripheral blood insulin levels increase (19).
We hypothesized that the improvement in nutritional status after TIPS in patients with liver cirrhosis is related to improvements in insulin sensitivity and glucose metabolism. This hypothesis was tested by means of oral glucose tolerance test (OGTT) and two-step hyperinsulinemic euglycemic clamp with tracer-determined endogenous glucose production (EGP). We related the findings to changes in body composition.
Subjects and ethics.
Twelve patients with liver cirrhosis set to undergo an elective TIPS procedure were consecutively included. During follow-up, one patient died from complications related to surgery (umbilical herniotomia). Of the remaining 11 patients who participated in the study, one could not complete the second glucose clamp examination due to severe back pain linked to lumbar spondylosis, and, as a result, clamp data comprise 10 subjects, and the remaining data all 11 patients. The diagnosis of cirrhosis was established by a combination of biochemical, clinical, and ultrasonographic findings. The indications for TIPS were intractable ascites in seven patients and a combination of repeated variceal bleeding, despite appropriate secondary prophylaxis and intractable ascites in four cases. All diagnoses were attributable to former alcohol abuse. Abstinence from alcohol is a prerequisite during the baseline work-up at our facility, and all participants adhered to that policy. All participants gave written, informed consent in accordance with the Helsinki declaration, and the study was approved by the Ethics Committee in Aarhus County.
Patients were examined at our research facility in the week before and again 6 mo after the TIPS procedure. Seven patients underwent paracentesis the day before bioimpedance measurements at the baseline examination. After an overnight fast, day 1 consisted of blood sampling, followed by a 2-h OGTT. On day 2, a hyperinsulinemic euglycemic clamp was conducted. Patients were instructed not to take their morning medication.
At time t = 0 min, patients drank 150 ml of standard oral glucose solution (0.5 g/ml), equal to 75 g of glucose. Blood samples were taken from an antecubital vein, starting at t = −5 and from t = 0 min every 30 min for 2 h. Based on the work by Abdul-Ghiani et al. (1), we calculated the glucose area under the curve (AUC) from t = 0 min to t = 30 min (AUC0–30) and the slope of the second half (60–120 min) of the glucose curve. The former depends on hepatic glucose production, whereas the latter reflects peripheral glucose uptake.
Two-step hyperinsulinemic euglycemic clamp and indirect calorimetry.
Glucose was clamped at 5 mmol/l by a variable glucose infusion, and insulin (Actrapid; NovoNordisk, Bagsværd, Denmark) infusion rates were 0.2 and 1.0 mU·kg−1·min−1. Samples taken during the basal steady-state period were used to estimate basal glucose clearance. The suppression of the average rate of tracer-determined EGP during the steady-state period of step 1 compared with the basal EGP was used to evaluate hepatic insulin sensitivity. The average amount of glucose infused to maintain euglycemia during step 2 steady-state period was used as a measure of the insulin effect (M-value). Insulin clearance during steady state was calculated as plasma insulin concentration divided by the insulin infusion rate.
A primed-continuous infusion of [3-3H]glucose (17 μCI prime + 17 μCI/min continuous; New England Nuclear Life Science Products, Boston, MA) was initiated at 0 min and continued throughout the clamp examination. Glucose flux rates were calculated using Steele's non-steady-state equation (49). During the clamp, EGP was calculated by subtracting the rate of exogenous glucose infusion from the appearance of [3-3H]glucose.
Indirect calorimetry was performed over a period of 30 min at time T = 120 min (basal), t = 270 min (step 1), and t = 420 min (step 2) (Open Hood Type, Deltatrac II, Datex, Helsinki, Finland). The simultaneous measurement of O2 uptake, CO2 production, and urine nitrogen provided the basis for the determination of energy production and the amount of oxidized glucose, protein, and fatty acids. Theoretical basis and equations are from the work by Ferrannini (15) and Frayn (16).
See Fig. 1 for a schematic overview of the protocol used.
Nutritional therapy and body composition.
Patients received standard nutritional education and therapy by a specialized dietary consultant, according to European Society for Clinical Nutrition and Metabolism guidelines during baseline work-up and follow-up (37). Paracentesis for moderate-to-tense ascites was performed the day before estimation of body composition by multifrequency bioimpedance analysis (Quadscan 4000, Bodystat, Isle of Man, UK). Body composition regression equations are developed by Kushner et al. (24) and Lautz et al. (26). The post-TIPS clinical examination did not reveal ascites in any of the participants. We have focused on BCM, since it constitutes metabolically active cells in the body and is relatively independent of the presence of ascites (32, 36).
Clinical status was assessed according to the model for end-stage liver disease (MELD) score (28). The galactose elimination capacity (GEC) was used to quantitatively measure liver function. The GEC was determined from blood concentration decay curves corrected for urinary excretion, as described by Tygstrup (53).
Plasma glucose concentrations were analyzed by the glucose oxidase method (Beckman Instruments, Palo Alto, CA). Serum insulin and C peptide were measured by ELISA assays (Dako, Glostrup, Denmark). Plasma glucagon was measured by an in-house radioimmunoassay modified from Orskov et al. (34). Serum-free fatty acid and fructosamine concentrations were determined by a colorimetric method using commercial kits [free fatty acid (FFA): Wako Chemicals, Neuss, Germany; fructosamine: HORIBA ABX, Montpellier, France]. Catecholamines were analyzed using a DECADE electrochemical detector (Antec, Leyden, Holland) on a Thermo BDS-Hypersil C18 3 μm 100 × 4.6 mm analytical column (Polygen Scandinavia Aps., Aarhus, Denmark) (10). Serum growth hormone and cortisol were measured with radioimmunoassays (DELFIA; Wallac Oy, Turku, Finland). All analyses were carried out in duplicate.
The TIPS was inserted using covered stents (Gore Viatorr Endoprosthesis; WL Gore and Associates, Flagstaff, AZ), as previously described (44).
Data are presented as means ± SD. Differences between baseline and post-TIPS values were explored by paired T-test and presented as point estimates with 95% confidence intervals (CI). Insulin and related variables (e.g., homeostatic model of assessment calculated insulin resistance) depended on logarithmic transformation before parametric methods could be applied. Epinephrine and growth hormone did not follow a normal distribution and were tested by the Wilcoxon matched-pairs signed-ranks test. Repeated measurements obtained during clamp were additionally explored by Hotelling's T-squared generalized means test. Correlations were explored by means of linear regression analyses. Two-tailed significance limit was set at 0.05.
The obtained baseline values for BCM, fasting glucose, fasting insulin, and fructosamine were compared with the results from 11 healthy controls (5 women and 6 men, mean age 55.5 yr, mean body mass index 23.9 kg/m2) recruited among hospital staff. They were matched for sex, age, and body mass index and examined in similar protocols using identical equipment at our research facility and have previously been described (17).
Patient characteristics, body composition, liver function, and REE.
Eleven patients completed the study (8 men and 3 women) with mean age of 58 ± 4 yr. None of the patients received antidiabetic drugs during follow-up. None of the patients needed paracentesis during follow-up, seven were able to discontinue diuretic treatment, and no gastrointestinal bleeding episodes were registered.
Body mass index rose by 2.2 kg/m2 (CI: 1.2-3.1) at 6 mo (Table 1). Both total body weight, lean body mass, and BCM increased after TIPS insertion, whereas fat percentage was unchanged. Ten of eleven patients gained in BCM (range: −1.5–10.9 kg), with a mean increase of 4.8 kg (CI: 2.7–7.3), equivalent to 15%. The BCM at baseline was reduced compared with age, weight, and sex-matched healthy subjects (TIPS baseline: 24.9 ± 8.1 vs. controls: 32.5 ± 8.2 kg, P = 0.04).
The MELD score and liver function expressed as GEC was unaltered after TIPS (Table 1). Levels of hemoglobin, creatinin, albumin, and bilirubin displayed no significant changes.
REE expressed per kilogram BCM was almost identical at the two examinations. Substrate oxidation, i.e., absolute amount and distribution, was unchanged: carbohydrates 1.28 ± 0.81 vs. 1.03 ± 0.60 mg·kg−1·min−1; lipid 0.99 ± 0.26 vs. 1.09 ± 0.26 mg·kg−1·min−1; protein 0.29 ± 0.13 vs. 0.27 ± 0.12 mg·kg−1·min−1.
Glucose tolerance-OGTT and glycemic control.
In accordance with the criteria of the American Diabetes Association (5), seven patients had a diabetic glucose tolerance, one an impaired glucose tolerance, and two a normal glucose tolerance at baseline. At the post-TIPS examination, eight were diabetic, one impaired, and two had a normal glucose profile, i.e., one patient moved from the impaired to diabetic glucose tolerance group. Fructosamine levels were unaffected by TIPS (326 ± 52 vs. 326 ± 81 μmol/l).
Fasting glucose measured before OGTT did not change, whereas fasting insulin tended to increase slightly after the TIPS procedure (P = 0.12). After TIPS glucose and insulin AUCs (AUC0–120) were increased by 14% (CI: 7–22%, P = 0.002) and 53% (CI: 14–90%, P = 0.012), respectively (Table 2 and Fig. 2). The C-peptide AUC0–120 tended to be larger after tips (CI: 0–39%, P = 0.05), and the C-peptide-to-insulin ratio decreased by 21% (CI: 8-35%, P = 0.01), indicating that part of the increase in plasma-insulin after TIPS is attributable to the shunting of portal blood. Neither fasting levels of glucagon, free fatty acids, cortisol, nor growth hormone changed from baseline to post-TIPS examination.
Glucose AUC0–30 showed a 24% increase (CI: 13–37%, P < 0.001), and the slope of the glucose curve from t = 60 to t = 120 changed from 0.01 to −0.03 mmol/min (P = 0.03) at post-TIPS examination.
Compared with matched controls, fructosamine levels were 28% higher at the baseline examination (CI: 10-47%, P = 0.008). Patients had slightly higher fasting glucose levels (patients = 6.0 ± 0.8 vs. controls 5.1 ± 0.5 mmol·l−1·min, P = 0.005) and nearly three times higher fasting insulin levels than controls (patients = 143 ± 93 pmol/l vs. controls = 56 ± 37 pmol/l, P = 0.009).
Whole body and hepatic insulin sensitivity-clamp.
Basal insulin levels, i.e., before insulin infusion, were 57% higher after TIPS (P = 0.03; Table 3 and Fig. 3). This difference disappeared after the start of insulin infusion. Blood glucose levels during clamp were unchanged from the baseline to the post-TIPS examination, reaching the designated level of 5.0 mmol/l during step 1. As shown in Table 3 and Fig. 4, peripheral insulin sensitivity (M-value) did not change after the TIPS procedure.
Tracer-determined rates of glucose appearance (Ra) and disappearance (Rd) followed the expected pattern of increasing values with increasing insulin levels (Table 3). EGP diminished in a reciprocal pattern. At step 1, EGP was suppressed to 50 and 59% of basal value at the baseline and post-TIPS examination, whereas, at step 2, the respective values were 12 and 2%. We did not observe changes in Rd, Ra, and EGP from baseline to post-TIPS examination (Table 3). Insulin clearance during steady state did not change. Basal EGP did not change, despite a higher insulin level. This is integrated in the validated index of hepatic insulin resistance (fasting insulin·basal EGP)(12), which tended to increase as the mean value rose by 36% (P = 0.08).
Glucose disposal (Rd) was unchanged from baseline to post-TIPS examination (Table 3). Nonoxidative glucose disposal, the equivalent to glycogen formation, did not change either.
Glucoregulatory hormones during clamp.
C-peptide, FFAs, and glucagon decreased in a stepwise manner from the basal period through step 1 to step 2 of the clamp (Table 4). There was no significant change in glucagon. Cortisol tended to decrease during each clamp (baseline by 9%, P = 0.02, and post-TIPS by 20%, P = 0.06), but was unchanged from the baseline to post-TIPS clamp examination (H0: curves coincide; P = 0.58). At both clamps, growth hormone tended to increase from basal to step 2 (baseline P = 0.03 and post-TIPS P = 0.10), but it did not differ between examinations. Epinephrine and norepinephrine were measured in the basal resting and fasting state with no change from the baseline to post-TIPS examination.
Predictors of BCM.
Neither baseline liver function, OGTT-derived AUC glucose, AUC insulin, fasting insulin, nor clamp-derived M-value statistically predicted the changes in BCM.
TIPS resulted in a gain in total body weight, and especially BCM, in all but one patient. Based on the OGTT, patients became more glucose intolerant, even though the glucose clamp examination detected no change in peripheral and hepatic insulin sensitivity, as expressed by the M-value and glucose Rd. The increase in BCM, despite worsened glucose intolerance and unchanged peripheral and hepatic insulin sensitivity, was unexpected and may be viewed in the light of the effects of the artificial shunt through the liver.
The small number of subjects under study entails a risk of a type 2 error. However, the paired design, almost complete follow-up, and the use of established techniques give us point estimates with narrow CIs, especially considering our primary findings, i.e., body composition and glucose metabolism.
The remarkable gain in BCM is in accordance with previous studies in which the TIPS procedure acts as an anabolic stimulus (4, 38). We hypothesized this to be related to improvements in insulin sensitivity and glucose intolerance, but our data refuted both expectations, and none of our variables regarding glucose metabolism predicted the gain in BCM.
Our results also concur with those of previous studies in that peripheral insulin levels rose due to the shunting of portal blood and decreased insulin clearance (19, 38). TIPS circumvents the first-pass metabolism of insulin that normally exceeds 50%. Likewise, switching the release of insulin from the portal vein to a peripheral vein immediately doubles peripheral levels and reduces hepatic sinusoidal insulin levels by 50%, as demonstrated in dogs (14). The metabolic consequences of this chronic hyperinsulinemia depend on the net effect on target cells and are likely to be tissue dependent, but studies exploring anabolism and body composition are not available.
TIPS has several effects that may improve food energy intake and reduce energy consumption. It eliminates portal hypertension, thereby preventing stressful catabolic events, such as variceal hemorrhage and spontaneous bacterial peritonitis (3, 21). Furthermore, resolution of ascites restores the effective stomach-volume and makes normal food intake possible (22). However, in a recent study, our laboratory could not demonstrate any change in food intake following TIPS (17). Portal hypertension is also implicated in the development of malabsorption and presence of bacterial products in the abdominal lymph nodes (30, 43, 46). Hence, elimination of portal hypertension may reduce energy-consuming low-grade inflammation (31). However, we detected no change in catecholamine levels or REE, which is in line with previous reports (4, 39). Improvement in liver function could not explain the BCM increase, as the quantitative metabolic liver function tests did not change.
The TIPS insertion also alters hepatic sinusoidal perfusion. After insertion, the blood flows in a retrograde manner from the liver sinusoids to the shunt so that liver sinusoidal perfusion comes to depend on the arterial blood supply (25, 45, 51). Insulin-rich portal blood now directly enters the systemic circulation via the shunt. The hepatocytes are exposed to relatively insulin-poor arterial blood, which reduces insulin clearance, as indicated by the decrease in C-peptide-to-insulin ratio. The available methods for assessing sinusoidal insulin concentration assume a normal distribution between the portal and arterial blood supply to the liver, which is violated by both cirrhosis and especially the TIPS. We are, therefore, not able to base our calculations of hepatic insulin sensitivity on estimates of prevailing hepatocyte insulin concentration.
This insulin imbalance and the ensuing blunted hepatic response (i.e., reduction of EGP) to increasing insulin secretion from the pancreas may explain several of our findings. The observed increase in the hepatic insulin resistance index may be attributed to TIPS-induced peripheral hyperinsulinemia and relative hepatic hypoinsulinemia, with the latter resulting in elevated and nonresponsive EGP. During step 1 and, especially, step 2, this relative lack of insulin was overruled by the exogenously administered high doses of insulin, eventually reaching the hepatocytes and suppressing EGP the normal way.
Moreover, this hepatocytic insulin deprivation could also explain the apparent discrepancy between the OGTT and glucose clamp with respect to estimated insulin sensitivity. The OGTT demonstrated that patients became more glucose intolerant after TIPS insertion. In patients with liver cirrhosis, glucose intolerance is attributed to peripheral insulin resistance (11, 47). But the clamp examination did not detect changes in peripheral or hepatic insulin sensitivity, since the M-value, Rd, and suppression of EGP were almost identical before and after TIPS. The glucose clamp step 2 with its high exogenous insulin supply overcome and mask the peripheral/liver tissue insulin imbalance. In contrast, the OGTT is a physiological examination exploring the effects of endogenous insulin supply. Therefore, EGP can be expected not to be suppressed during the post-TIPS OGTT. This difference between the two experimental methods might explain the marked increase in both insulin and glucose AUC during the OGTT, especially from 0 to 30 min, which depends mostly on the EGP (1). The same mechanism may explain why C-peptide-to-insulin ratio during OGTT decreased, although insulin clearance calculated during insulin infusion did not change after TIPS insertion.
Previous studies are conflicting regarding glycemic status after TIPS. One study reports an increase in the need for antidiabetic medicine, but it included only patients with medically treated diabetes already at the baseline examination (13). Two other studies found that glycemic control was maintained, as reflected in unaltered fasting glucose and stable fructosamine (19) or HbA1C (40). Our data suggest that glycemic control is mainly conserved after TIPS insertion with unchanged fructosamine and fasting glucose levels, whereas absorptive glucose levels were higher. However, we followed glucose levels for only 120 min during OGTT, and it is possible that longer recording would have yielded more equal AUCs and hence eliminated the signs of glucose intolerance after TIPS. Furthermore, the nonoxidative glucose disposal or FFAs did not change after TIPS, and there was no change in the glucoregulatory factors studied (glucagon, growth hormone, and cortisol).
In conclusion, the TIPS insertion leads to an anabolic state with a 15% increase in BCM in patients with liver cirrhosis. This was not due to improved insulin action, since both insulin sensitivity and glucose intolerance, as measured by the physiological OGTT method, deteriorated after TIPS. The clamp- and tracer-based techniques showed unchanged peripheral, as well as hepatic insulin sensitivity after TIPS. The apparent discrepancy may reflect a TIPS-induced relative reduction in hepatocyte insulin levels, causing defective suppression of EGP under physiological conditions. The anabolic effect of TIPS thus remains mechanistically unexplained from these studies.
This study received financial support from the Clinical Institute, Aarhus University; Aarhus University Research Fund; The NOVO Nordic foundation, and The A. P. Møller Foundation for the Advancement of Medical Science.
No conflicts of interest, financial or otherwise, are declared by the author(s).
The authors thank Karen Mathiassen for invaluable and expert laboratory assistance.
- Copyright © 2010 the American Physiological Society