Dr T. Sauerbruch, Department of Internal Medicine I, University of Bonn, Sigmund Freud Strasse 25, D-53105 Bonn, Germany. E-mail: firstname.lastname@example.org
Background : Long-acting somatostatin analogues have been suggested as an alternative to propranolol for the prevention of variceal rebleeding.
Aim : To compare the effectiveness of lanreotide SR, a new depot formulation injected once-weekly, and propranolol in reducing circadian portal blood flow (PVF) and meal-stimulated hepatic venous pressure gradient (HVPG) in patients with liver cirrhosis.
Methods : Patients were randomized to receive either lanreotide SR intramuscularly (30 mg once weekly, n = 12) or propranolol (n = 12) orally. Hemodynamic measurements were performed on day 0 and on day 21 after a 3-week period of drug administration, while patients received three standard oral liquid test meals. On each study day 27 PVF measurements were performed over 24 h and eight measurements of HVPG during the first postprandial period.
Results : Propranolol was more effective than lanreotide SR in reducing baseline HVPG (−21.9 vs. −13.6%, P = 0.04) and meal-stimulated HVPG (−16.6 vs. −3.8%, P = 0.04). Propranolol reduced circadian PVF significantly by 9.3% (P = 0.03) but not lanreotide SR.
Conclusions : Long-term treatment with propranolol reduced baseline and postprandial HVPG and circadian PVF, while lanreotide SR did not. The results of our study do not encourage clinical testing of lanreotide SR 30 mg for the prevention of variceal haemorrhage.
Bleeding from oesophageal varices is a fatal complication in patients with liver cirrhosis. Non-selective beta-blockers are recommended as standard therapy for the prevention of recurrent variceal haemorrhage. However, at least one third of patients do not respond, have contraindications or cannot tolerate the propranolol therapy.1 Long-acting somatostatin analogues have been proposed as an alternative.2 A new depot formulation of the long-acting somatostatin analogue lanreotide (lanreotide SR), requiring only one intramuscular injection per week, reduced portal pressure and increased splanchnic vascular resistance in portal vein stenosed rats.3 Corresponding data in humans are not available.
Haemodynamic response to drug therapy is defined as decrease in baseline hepatic venous pressure gradient (HVPG) by at least 20%, preferably below 12 mmHg.4–6 A circadian study design, including three meals in order to mimic daily life, has been suggested for evaluation of chronic treatment effects on splanchnic haemodynamics.7 Such a study design offers a more detailed view on drugs' haemodynamic activity. Therefore, we investigated the effects of a three-week lanreotide SR therapy and of propranolol on circadian meal-stimulated portal blood flow and on meal-stimulated portal pressure.
Materials and methods
This study included 24 patients with liver cirrhosis and portal hypertension. Table 1 shows the patient characteristics. The diagnosis of cirrhosis and portal hypertension was based on clinical, biochemical, radiological and histological findings. All patients had signs of portal hypertension, such as recent variceal haemorrhage (n = 16), oesophageal varices (n = 21) or portal hypertensive gastropathy (n = 17). They were in a stable clinical condition at the time of study entry and had not taken vasoactive drugs such as somatostatin, vasopressin (or any analogues) or beta-blockers for at least 1 week prior to the study onset. Four out of nine patients with alcoholic liver disease in the propranolol group went on drinking regularly between day 0 and day 21 examination. All other patients with alcoholic liver disease had been abstinent for at least 6 months.
Table 1. Patient characteristics
Lanreotide n = 12
Propranolol n = 12
51 ± 11
57 ± 6
Etiology (alcoholic/posthepatitic/ cryptogenic)
Child–Pugh score (A/B/C)
Oesophageal varices (I °/II °/III °/none)
Hypertensive gastropathy (yes/no)
Previous variceal bleeding (yes/no)
Ascites present (yes/no)
Previous ascites (yes/no)
Patients with contraindications to beta-blockers, a known hepatocellular carcinoma, primary biliary cirrhosis or a therapeutic portosystemic shunt were not included into the study. Pregnant or breast-feeding women and active drug abusers were also excluded.
The study protocol was in accordance with the ethical guidelines of the 1975 Declaration of Helsinki and approved by the Local Ethics Committee. Written informed consent was obtained from each patient.
In a prospective, operator-blinded study, patients were randomly assigned to receive either propranolol or lanreotide SR according to an allocation list. Randomization was performed externally by IPSEN Ltd, Slough, UK. The operator who performed the haemodynamic measurements (L.K.) was unaware of the allocation.
Measurements of circadian meal stimulated portal haemodynamics were performed on day 0 and repeated on day 21. The patients received either lanreotide SR (Somatuline™ LA, IPSEN Biotech, Paris Cedex, France) every 7 days in the form of an injection of 30 mg intramuscularly each over a period of 3 weeks.
The propranolol dose was titrated until the reduction in heart rate was 25% but not lower than 50 counts/min. Dose titration for patients receiving propranolol at 3 day intervals began at a dose of 40 mg twice daily, increasing to 80 mg b.d. and 160 b.d. if necessary. When the optimal dose regime was achieved the patients continued to receive propranolol for a further 18 days (21 days in total). The mean propranolol dose was 93 ± 13 mg.
On every study day, 27 Doppler measurements of the portal vein were performed (08:30, 09:00, 09:15, 09:30, 09:45, 10:00, 10:30, 11:00, 13:00, 13:15, 13:30, 13:45, 14:00, 14:30, 15:00, 17:00, 19:00, 19:15, 19:30, 19:45, 20:00, 20:30, 21:00, 23:00, 01:00, 03:00, 08:30). An oral standard liquid meal of 500 mL Ensure Plus (Abbott, Wiesbaden, Germany) was given during a period of 5 min after measurements at 09:00, 13:00 and 19:00 had been performed. We also measured the hepatic venous pressure gradient (HVPG, mmHg) heart rate (HR, counts/min) and mean arterial pressure (MAP, mmHg) at 08:30, 09:00, 09:15, 09:30, 09:45, 10:00, 10:30, 11:00. Meal-stimulated splanchnic and systemic haemodynamics were recorded at 09:15, 09:30, 09:45, 10:00, 10:30 and 11:00.
Patients were examined after an overnight fast. For noninvasive Doppler measurements we used the Duplex ultrasound device, Toshiba SSH 140-A with a 2.5-MHz transducer (Toshiba Medical Systems, Neuss, Germany). All haemodynamic measurements except HR and MAP were carried out by one of the authors (L.K.). We recorded HR and MAP using a Sirecust life-surveillance monitor (Siemens, Germany).
Patients were in a supine position for at least 15 min prior to the measurements. All vessels were examined in a predetermined order: blood flow velocity (PVV, cm/s) and the cross-sectional area (PVA, mm2) were assessed in the portal vein at the crossing of the common hepatic artery in the hepatoduodenal ligament. A Doppler angle of less the 60° was obligatory to assure reliability of Doppler measurements. Each result for PVV and PVA was calculated as a mean from three subsequent measurements. From these we calculated the portal vein blood flow volume (PVF, mL/min = PVV × PVA).
For measurement of the HVPG we catheterized a main right hepatic vein with a balloon catheter (5F catheter, 8.5 mm balloon diameter, Meditech; Boston Scientific Company, Watertown, MA) via the right jugular vein under fluoroscopic control. The accurate free position of the catheter was documented. We calculated HVPG as a mean of three measurements with inflated and deflated balloon (‘wedged’ minus ‘free’ hepatic venous pressure). After the final measurement, the catheter was removed.
Serum lanreotide levels.
Blood samples for biochemistry were withdrawn 7 days before and 7 days after the treatment period. In addition, lanreotide serum levels were determined on day 7 at 09:00 and on day 22 at 09:00 after completing the haemodynamic experiments. Samples were centrifuged, and the extracted serum was stored at −20 °C. After completion of the study, lanreotide serum levels at on day 0 and day 21 were determined by LASA Laboratories, Barcelona, Spain.
For statistical analysis, we used the SPSS package (SPSS Inc., Chicago, IL). Using the trapezoid rule, we calculated the area under the curve (AUC) for postprandial measurements (AUCpp) and for circadian measurements (AUC24h). The Mann–Whitney U-test, Friedman test and Wilcoxon signed-rank test were used as appropriate. P-Values less than 0.05 (two-sided) were considered statistically significant. Results are given as mean ± standard deviation (s.d.) except where otherwise declared.
One patient in the lanreotide SR group refused from the second measurement.
Neither lanreotide SR nor propranolol showed a significant effect on biochemical parameters of liver function (see Table 2).
Table 2. Biochemical parameters of liver function before and after a 3-week treatment (day 0 and day 21) with either lanreotide SR (n = 11) or propranolol (n = 12)
P, Wilcoxon signed-rank test, mean ± s.d.
Comparison between both groups (Mann–Whitney U-test) showed no significant difference for any parameter, either before or after drug administration.
Total bilirubin (mg/dL)
1.41 ± 0.72
1.27 ± 0.78
1.62 ± 0.88
1.24 ± 0.58
3.78 ± 0.59
3.82 ± 0.80
3.49 ± 0.50
3.27 ± 0.36
Prothrombin ratio (%)
90.3 ± 21.8
85.3 ± 17.6
75.4 ± 18.4
79.2 ± 14.9
23.5 ± 11.3
19.3 ± 7.6
28.0 ± 18.3
24.4 ± 15.8
Propranolol and lanreotide SR reduced fasting HVPG, but the effect of propranolol was significantly more pronounced (−21.9 ± 24.0% vs. −13.6 ± 20.5%, P = 0.04; Table 3). Meal-stimulated HVPG was also more reduced after propranolol than after lanreotide SR (−16.6 ± 12.7% vs. −3.8 ± 13.8%, P = 0.04). Figure 1 shows that the effect of propranolol on AUCpp of HVPG was mainly caused by a reduction in baseline HVPG. However, when AUCpp of HVPG was calculated as percentage of baseline values propranolol still caused a −15.5 ± 13.2% reduction (P = 0.001), while lanreotide did not [−2.0 ± 15.0%, not significant (NS)]. In addition, propranolol reduced both fasting and meal-stimulated MAP and HR, while lanreotide SR did not (Table 4).
Table 3. Fasting, baseline splanchnic and systemic haemodynamics before and after a 3-week treatment (day 0 and day 21) with either lanreotide SR (n = 11) or propranolol (n = 12)
Table 4. Meal-stimulated splanchnic and systemic haemodynamics before and after a 3-week treatment (day 0 and day 21) with either lanreotide SR (n = 11) or propranolol (n = 12). Area under the postprandial time curve (AUCpp, 2 h)
Significant effects on fasting and on meal-stimulated PVF were not observed, either with propranolol or with lanreotide SR (Tables 3 and 4). AUC24h of PVF was significantly reduced in the propranolol group (−9.3 ± 17.8%) due to a notable decrease in PVV (−11.1 ± 8.1%), while PVA did not change (−0.1 ± 18.4%). Lanreotide SR showed no effect on AUC24h of PVF (−0.8 ± 22.4%; see also Table 5 and Figure 2).
Table 5. Circadian splanchnic haemodynamics before and after a three-week treatment (day 0 and day 21) with either lanreotide SR (n = 11) or propranolol (n = 12). Area under the circadian time curve (AUC24h).
Baseline HVPG was insignificantly lower in the lanreotide SR group compared to the propranolol group (13.8 ± 5.8 mmHg vs. 17.8 ± 5.4 mmHg, P = 0.101). However, changes in baseline HVPG between day 0 and day 21 did not correlate with the altitude of baseline HVPG either in the lanreotide group (r = 0.007, NS) or in the propranolol group (r = − 0.29, NS). Likewise, in both groups the effect of drug administration on AUCpp of HVPG did not correlate with the initial AUCpp HVPG (lanreotide group: r = 0.14, NS; propranolol group: r = − 0.46, NS) and the effect on AUC24h PVF did not correlate with the AUC24h before drug administration (lanreotide group: r = − 0.82, NS; r = − 0.53, NS).
Ten patients achieved a reduction of 20% or more in baseline HVPG (responders): seven patients with propranolol and three patients with lanreotide. Changes in baseline Doppler parameters could not predict this HVPG response. (responder vs. nonresponder: PVF, −3.1% vs. −0.8%, NS; PVV, −5.7% vs. −4.7%, NS). In addition, changes in AUCpp HVPG between day 0 and day 21 did not correlate with changes in AUCpp PVF (r = − 0.16, NS), AUCpp PVV (r = − 0.02, NS) and AUCpp PVA (r = 0.001, NS).
Before drug administration on day 7, no lanreotide was detectable. On day 22, mean serum lanreotide levels ranged from 2.043 to 6.912 ng/mL (3.343 ± 1.94 ng/mL). There was no correlation between lanreotide serum levels and changes in AUC24h between day 0 and day 21 (r = 0.34, NS).
This study demonstrated that a 3-week treatment with propranolol reduced postprandial portal pressure and circadian portal blood flow, whereas lanreotide SR had no significant effect.
In accordance with the definition of ‘clinically significant portal hypertension’, patients were included with endoscopically proven signs of portal hypertension such as oesophageal varices or portal hypertensive gastropathy and/or ascites. Observing these entry criteria the lanreotide SR group showed a slightly lower baseline HVPG. However, the haemodynamic response to octreotide infusions in patients with liver cirrhosis is obviously not dependent on the degree of portal hypertension, e.g. healthy volunteers showed similar changes as patients with portal hypertension.8 Thus the ability of lanreotide to reduce postprandial portal hyperemia was first demonstrated in healthy volunteers.9 Analysis of our own data found no correlation between the haemodynamic response and baseline HVPG levels. Likewise, the effect of both lanreotide and propranolol on postprandial increase in HVPG and on circadian PVF were not related to pre-treatment values. Therefore, an effect of lanreotide SR on postprandial portal hyperemia should have been demonstrable even in healthy volunteers, and thus it is not likely that the efficacy of lanreotide was missed due to the slight imbalance in both treatment arms.
Somatostatin analogues reduce portal venous inflow and portal-pressure secondary to splanchnic vasoconstriction.10,11 They are under current evaluation as an alternative for chronic treatment, since Jenkins et al. showed a reduction of rebleeding and an improvement in survival after 6 months of treatment with octreotide in cirrhotic patients following variceal haemorrhage.2 Lanreotide, a somatostatin analogue with a particular ligand affinity for somatostatin receptor subtypes SSTR-2 and SSTR-5, is now available as a depot formulation which requires only weekly intramuscular injections.12 This new depot formulation, lanreotide SR, was able to reduce portal pressure in portal hypertensive rats.3 We expected the lanreotide serum levels to be in a steady state when we performed the haemodynamic measurements 7 days after the third intramuscular injection. Accordingly, in every patient we measured lanreotide levels of above 2 ng/mL that were similar to those shown to reduce the postprandial area under the PVF curve in an acute experiment in healthy volunteers.9 Nonetheless, neither postprandial portal pressure nor circadian PVF was reduced, despite adequate lanreotide levels. Blunting of postprandial portal hyperemia is an accepted haemodynamic parameter for the efficacy of somatostatin analogues.13–16
Even when administered intravenously in an acute setting it is still a matter of debate whether somatostatin and its long-acting analogues have a prolonged haemodynamic activity. Studies which observed a prolonged reduction of portal pressure11,17 and of fasting11,18,19 and meal-stimulated portal blood20 flow were contradicted by others who found no effect.21 Few studies exist which examined chronic treatment with somatostatin analogues. These found octreotide to be still effective on postprandial splanchnic hyperemia22 and on portal pressure17 in patients who were pretreated with up to three subcutaneous injections a day. Likewise, in the clinical trial reported by Jenkins et al. octreotide was administered subcutaneously twice daily.2 From our data one may speculate that more constant plasma concentrations in the steady state after application of the slow-release depot formulation of lanreotide SR are less effective than the more oscillating concentrations occurring after subcutaneous injections. The somatostatin-agonist-dependent desensitization and internalization of the receptor SST223,24 may play a role in the attenuation of depot lanreotide's haemodynamic efficacy. From our data, however, we cannot exclude that higher doses of lanreotide SR might have an effect on postprandial and circadian portal hyperemia.
Propranolol is recommended as first-line treatment for primary and secondary prophylaxis of variceal haemorrhage.25,26 Our findings of a reduction in meal-stimulated HVPG and PVF, mainly caused by a reduction of the absolute values instead of a postprandial blunting, are well in line with previously published results.27,28 However, this is the first study to show the effect of a long-term treatment with propranolol on circadian meal-stimulated portal flow. The reduction of the circadian portal blood flow was almost entirely caused by a reduction in the portal blood flow velocity similar to that described previously.27
Until now, baseline HVPG is considered to be the gold standard for the evaluation of drug effects on the portal vascular system.25 While the acute HVPG response to propranolol can be accurately predicted by Doppler flowmetry,29–31 information on their agreement in the assessment of chronic treatment effects are sparse. In accordance with data from Merkel et al.32 the present study shows that portal flowmetry cannot distinguish long-term responders form nonresponders. This might be due to the fact that the relation between portal pressure and flow is determined by the portal vascular outflow resistance.11 The portal vascular outflow resistance is determined by the presence of spontaneous portosystemic shunts and by the hepatic vascular resistance. Both differ between patients. Consequently, Doppler sonographic measurement of portal blood flow is not useful to predict chronic long-term treatment response on portal pressure. In contrast to HVPG measurements the noninvasive Doppler ultrasound is particularly useful for serial and repeated measurements. This potency of Doppler ultrasound is demonstrated by the 27 measurements over a 24 h period performed twice in all patients.
The chronorisk of variceal haemorrhage33–35 was explained by a circadian variability of portal pressure36 and flow,37 and has since led to a circadian approach in the proof of drug efficacy on the portal vascular system.7,38 Observed drug effects on circadian portal blood flow contain both the abolition of circadian variability38 and the decrease of the 24 h AUC.7 The present data of lanreotide SR showed a 10% reduction in baseline HVPG but no effect on the circadian PVF. The circadian study design, including three oral meals, gives a more detailed impression of a drug's haemodynamic efficacy than a single measurement. These observations encourage further studies validating the predictive value of circadian portal Doppler flowmetry for clinical endpoints.
In conclusion, for the first time we applied circadian portal Doppler flowmetry including three oral meals to compare chronic treatment of propranolol and lanreotide SR. A 3-week treatment with propranolol showed a significant effect on portal haemodynamics, while lanreotide SR did not. The results of our study do not encourage clinical testing of lanreotide SR 30 mg for the prevention of variceal haemorrhage. On the other hand, the relatively high proportion of side-effects and of nonresponders to standard therapy with propranolol warrants the search for alternative treatments. Thus, further haemodynamic evaluations after higher doses of lanreotide SR may be performed.
We thank Prof Dr J. Drewe, Department of Clinical Pharmacology, Kantonsspital Basel, Switzerland for statistical support and Dr Juan Carlos Garcia-Pagan, Hepatic Hemodynamics Laboratory, Hospital Clinic i Provincial, University of Barcelona, Spain, for his constructive criticism of the manuscript. Mrs D. Bammer, Department of Internal Medicine I, University of Bonn, is acknowledged in particular for her kind and expert technical assistance. This study was supported by a grant from IPSEN Ltd, Slough, UK.