Randomised clinical trial: the safety and efficacy of long-acting octreotide in patients with portal hypertension

Authors


Correspondence to: Dr P. S. Kamath, Division of Gastroenterology and Hepatology, Mayo Clinic and Foundation, 200 First Street SW, Rochester, MN, 55905 USA.

E-mail: kamath.patrick@mayo.edu

Summary

Background

It remains unclear whether a long-acting preparation of octreotide (Sandostatin LAR) can be safely used for portal hypertension in patients with compensated cirrhosis.

Aim

To determine the safety and efficacy of LAR among patients with Child Pugh Class A or B cirrhosis and small oesophageal varices.

Methods

A randomised, double-blind, placebo-controlled study was conducted in 39 patients with cirrhosis and small oesophageal varices. Safety was based on frequency and severity of adverse events. Efficacy was determined by hepatic vein pressure gradient (HVPG) measured at baseline and day 84 following administration of LAR 10 mg (n = 15), 30 mg (n = 10) or saline (n = 14). Fasting and postprandial portal blood flow (PBF), superior mesenteric artery pulsatility index (SMA-PI), glucagon and octreotide levels were measured. An intention-to-treat analysis was performed.

Results

Four patients in the LAR 30 group (40%) withdrew from the study due to serious adverse events. No patient in the LAR 10 or control group had serious adverse events. There was no statistically significant decrease between HVPG at day 84 and baseline with LAR 30 mg (11.8 ± 2.3 mmHg vs. 14.1 ± 3.2), LAR 10 mg (15.3 ± 4.8 mmHg vs. 15.1 ± 3.8), or saline (13.3 ± 3.8 mmHg vs. 15.1 ± 4.3) (P = 0.26). Neither PBF, SMA-PI nor plasma glucagon levels were significantly decreased from baseline (P = 0.56).

Conclusions

The absence of significant haemodynamic benefit, as well as the high frequency of severe adverse events associated with use of LAR, do not support the use of this agent in the treatment of portal hypertension.

Introduction

Portal hypertension results from a combination of both an increased resistance to and increased flow of portal blood. Variceal haemorrhage is the most dramatic and serious presentation of portal hypertension. Pharmacological agents used to control as well as prevent variceal bleeding decrease portal pressure predominantly by decreasing portal blood flow. Practice guidelines recommend somatostatin analogues, such as octreotide as adjuvant therapy for the control of variceal hemorrhage, along with timely therapeutic endoscopy.[1-3] Octreotide acts by inhibiting glucagon and blunting the postprandial increase in hepatic vein pressure gradient (HVPG) and portal blood flow. Octreotide also influences splanchnic haemodynamics by modulating the postprandial increase in portal vein and superior mesenteric artery velocity, and the postprandial decrease in superior mesenteric artery pulsatility index (SMA-PI).[1, 4, 5] Although octreotide is mainly used to control variceal bleeding, there are some data to suggest that long-term use of octreotide may be beneficial in the prevention of variceal haemorrhage.[6, 7]

As octreotide has a half-life of less than 2 h in patients with cirrhosis, the therapeutic effects decline rapidly following subcutaneous injection or cessation of infusion.[8] Hence, there is clinical interest in the long-acting release formulation of octreotide [Sandostatin LAR, (LAR, in short); Novartis, Basel, Switzerland] given as an intramuscular injection. The safety profile, pharmacokinetics and pharmacodynamics of LAR have not been studied carefully in patients with cirrhosis, but are well described in patients with acromegaly.[9] In patients with acromegaly, blood levels peak within 1 h of administration, plateau on day 14 and remain stable until 14 to 42 days before declining. If LAR causes a sustained decrease in portal pressure, LAR given by intramuscular injection at regular intervals could possibly prevent initial or recurrent variceal haemorrhage in patients with cirrhosis. This hypothesis is supported by a small randomised controlled study in cirrhotic patients in which the HVPG was compared among a group of patients receiving LAR 20 mg intramuscular every month vs. placebo. After 3 months of therapy, the treatment group had a statistically significant reduction in HVPG, independent of liver function.[7] However, this trial only included 18 patients, and thus provides insufficient evidence of the safety profile of LAR in the cirrhotic population. Furthermore, many of the patients in that trial were on nonselective beta-blockers during the study period which may have impacted the HVPG measurements and determination of LAR related adverse events. Finally, a dose- response relationship between LAR and HVPG was not assessed in the study design to validate the relationship between LAR and an improvement in hepatic haemodynamics.

The primary objectives of the present multicenter study were to evaluate in patients with cirrhosis the safety and efficacy of LAR in reducing portal pressure as measured by the HVPG. The secondary objectives of this study were to assess the pharmacokinetics and pharmacodynamics of LAR, and mechanism of action in reducing portal pressure. Only patients with Child Pugh Class A and B cirrhosis and small oesophageal varices at low risk of bleeding and not requiring beta-blockers were studied. It was considered unethical to enter patients with large oesophageal varices or with small varices and Child Pugh Class C into the study as these patients require beta-blockers and cannot be entered into a study with a placebo control arm.

Materials and Methods

Patient population

Consecutive patients were considered for study enrollment if they fulfilled the following inclusion criteria: (i) age ≥18 years; (ii) presence of small oesophageal varices, defined as varices <5 mm in diameter, without red signs documented on endoscopy within 3 months of enrollment; (iii) Cirrhosis documented by biopsy, or suggested by characteristic features on abdominal imaging (nodular appearance, irregular contour) along with impaired synthetic liver function and thrombocytopaenia and Child Pugh Class A or B. Informed consent was obtained from all subjects and the study was independently approved by the various institutional review boards of each participating tertiary-care centre. The trial was registered at www.clinicaltrials.gov with registry number NCT01188733.

Patients were excluded from the study if they met any of the following criteria: (i) pregnant, lactating or of child-bearing potential and not practicing acceptable method of birth control; (ii) allergy to octreotide; (iii) high risk varices on endoscopy carried out within 3 months of assessment (large varices or red signs); (iv) Child Pugh Class C cirrhosis; (v) hepatocellular carcinoma; (v) evidence of ongoing alcohol or illicit drug abuse within 6 months of the study; (vii) serum creatinine greater than 2 mg/dL; (viii) platelet count below 50 000/μL; (ix) prothrombin time >4 s from control; (x) human immunodeficiency virus positive; (xi) symptomatic gallstones; (xii) previous history of upper gastrointestinal bleeding in the previous 3 months, defined as haematemesis and/or melena; (xiii) previous history of variceal bleeding; (xiv) history of congestive heart failure, unstable angina, sustained ventricular tachycardia, or ventricular fibrillation; (xv) use of any investigational drug within 1 month prior to screening; and (xvi) current use of beta blockers or long-acting nitrates, any other drug therapy known to have an influence on portal pressure (diuretics were allowed provided patients were on a stable dose for at least 30 days).

Study design

A multi-centre, randomised, double-blind, placebo-controlled trial of two doses of LAR was conducted at four academic institutions within the United States. Patients who met eligibility criteria and provided written informed consent were enrolled consecutively.

Upon completing a screening assessment, subjects underwent baseline testing including physical examination, complete blood counts with differential, blood chemistry, urinalysis, pregnancy test for females (serum beta human choriogonadotropin), electrocardiogram and ultrasound of the liver and gallbladder.

Patients then had baseline HVPG,[10] portal blood flow and SMA-PI[11] measured in the fasting, and postprandial state after consuming 700 kcal standardised meal (two cans of Ensure Plus, Abbott Nutrition). To obtain these measurements, subjects first received conscious sedation with midazolam intravenously, and a venous catheter introducer (USCI International Inc., Burlington, MA, USA) was placed in the right femoral vein by a Seldinger technique. A six French balloon-tipped catheter (MediTech, Cooper Scientific Corporation, Watertown, MA, USA) was then placed in the main right hepatic vein; contrast was injected to document correct positioning. Wedge hepatic vein pressure (WHVP) and free hepatic vein pressure (FHVP) were measured three times by inflating and deflating the balloon.[10] Portal pressure was determined from the HVPG as being the average of the three readings of the difference between WHVP and FHVP. All HVPG measurements were collected on a recorder tracing and reviewed by a single investigator (JB) at a site where no patients were studied. The baseline HVPG was measured followed by measurement of the portal blood flow and pulsatility index by duplex ultrasound.[11] The sampling position for measurement of the portal vein by ultrasound was set at the midpoint of the portal vein trunk (midpoint between the bifurcation of the right and left portal venous branches and the confluence of the splenic and superior mesenteric veins). PBF was calculated using the following equation[12]:

display math

where A and B are short and long axes of the cross-section of the vessel respectively; Vdmax is the velocity obtained from the Doppler spectrograms; and θ is the angle between the doppler beam and the blood vessel.

SMA-PI was determined by the following formula:

PI = VpeakVmin/V, where Vpeak is the peak velocity, Vmin is the minimum velocity and V is the mean velocity.[11]

The portal vascular resistance (PVR) was calculated using the following equation:

display math

where, Ppv is the portal venous pressure in mmHg, Phv is the hepatic venous pressure, and Qpv is the portal blood flow per body weight in mL/min/kg of body weight.

The meal was then administered and postprandial portal blood flow and SMA-PI (12) were measured 30 min and 60 min later. Postprandial HVPG was measured as soon as possible after portal blood flow and SMA-PI measurements. Thereafter, subjects received a subcutaneous test dose of Sandostatin 100 mcg. Blood samples were drawn for measurement of octreotide levels 30 min; and 2, 4, 6, 8 and 24 h after the test dose. Octreotide levels in plasma were determined by radioimmunoassay.

Patients who tolerated the test dose (all subjects) were then assigned by random number allocation to receive LAR 10 mg or 30 mg intramuscular, or saline intramuscular every 4 weeks, for a total of three injections. Blood samples were drawn for measurement of octreotide and plasma glucagon levels at baseline, 14 days after the first dose, immediately prior to second and third dose, and 28 days after the third dose (end of study). Portal blood flow and SMA-PI were measured in the fasting and non-fasting states at day 28, day 56 and day 84 visits 5, 6, 7 (Table 1); these measurements were performed within 24 h of octreotide blood sampling. Subjects returned 28 days after receiving the third dose of LAR or saline (at day 84) to undergo repeat HVPG measurements performed in fasting and non-fasting states.

Table 1. Schedule of evaluations
Visit 1Visit 2Visit 3Visit 4Visit 5Visit 6Visit 7
  1. HVPG, hepatic vein pressure gradient; LAR, long-acting release; PBF, portal blood flow; SMA-PI, superior mesenteric artery pulsatility index.

Day −14Day −1Day 0Day 14Day 28Day 56Day 84
ScreeningSandostatin 100 mg s.c.Blood sample collectionBlood sample collectionBlood sample collectionBlood sample collectionBlood sample collection
HVPGBlood samplesRandomisation to LAR or saline PBF and SMA-PI via ultrasoundPBF and SMA-PI via ultrasoundPBF and SMA-PI via ultrasound
PBF and SMA-PI via ultrasound First injection of sandostatin LAR 10 mg, 30 mg, or saline Second injection of sandostatin LAR 10 mg, 30 mg, or salineThird injection of sandostatin LAR 10 mg, 30 mg, or salineHVPG

Any patient who discontinued treatment prematurely completed an end of study visit evaluation, and any safety concerns were assessed. Patients who could not have scheduled evaluations completed within 5 weeks of the last dose of sandostatin LAR were considered lost to follow-up.

A comprehensive safety assessment examining adverse side effects, ultrasound for gallbladder stones, laboratory abnormalities (complete blood count, blood chemistry and urinalysis) and electrocardiogram abnormalities was conducted for each subject. Adverse events were characterised as serious if they were fatal or life-threatening, lead to prolonged hospitalisation, disabling or incapacitating, or required change in management.

The effects of treatment on portal hypertension were measured by the following: change from baseline in HVPG; change from baseline in portal blood flow; or change from baseline in SMA-PI.

A blinded evaluator design was used as matching placebo for sandostatin LAR was not available. Each study site designated one individual who was responsible for administration of all study drugs; this individual did not participate in study assessments and dispensed the medication according to the randomisation list. The dispenser maintained a log of medications administered, and the log was periodically monitored to verify accuracy in dispensing and compliance.

Concomitant medication(s) prescribed to subjects for any complication of liver disease were recorded. Unless patients required treatment for an adverse event, investigators discouraged patients from starting any medication during the study that they were not taking at baseline. Reasons for patient discontinuation from the study were recorded.

Statistical analysis

An intention-to-treat analysis was performed. For sample size calculations, the primary outcome measure was the development of a serious adverse event. The secondary outcome measures were the change in portal blood flow, HVPG and SMA-PI from baseline values. Clinical and demographic characteristics were presented for the LAR 30, LAR 10 and saline groups. Continuous variables were summarised by sample size, mean and 95% CIs; categorical variables were summarised as frequencies and percentages. Results are expressed as mean with standard deviations. Treatment groups were compared using two-sample tests and analysis of variance (anova).

The safety assessment was based on the frequency of adverse events. Adverse events were summarised by presenting the type and percentage of subjects having any adverse event. Descriptive statistics were calculated for all pharmacokinetic concentrations and for derived model-independent pharmacokinetic parameters by each dose of sandostatin LAR. Clearance and volume distribution were summarised and compared with data on file at Novartis from healthy subjects. Individual serum concentration and pharmacokinetic parameters were listed. Summary statistics (mean, 95% CI) were displayed for each treatment group (10 mg, 30 mg and Saline). The effect of treatments for different doses (10, 30 mg) on the pharmacokinetic profile (i.e., area under the curve days 0–84, AUC0–84, Cmax and tmax) was evaluated by using a standard anova model. The pharmacokinetic/pharmacodynamic relationship was evaluated for the two treatment groups (10, 30 mg) based on the PD variables at different time periods against PK concentrations at those time points. A Pearson correlation analysis was performed to assess the pharmacokinetic/pharmacodynamic relationship.

The primary sample size calculation was conducted for safety end points. Based on the assumption that there would be a 40 percentage point difference in the incidence of serious adverse events between the placebo and active drug groups, 12 subjects were needed in each group for an 80% power (two-tailed, alpha 0.05). We also performed a post hoc sample size calculation for the efficacy end point. The current sample size would be sufficient to detect differences of 5 mmHg or greater in the mean HVPG between groups.

Results

Thirty-nine patients were consecutively enrolled from four centres. These patients were randomised as follows: 15 in the LAR 10 mg group; 10 in the LAR 30 mg group; and 14 in saline group. Baseline clinic-demographic characteristics of the patients are depicted in Table 2. There were 26 male and 13 female subjects, the majority of whom were Caucasian. The age and race distributions were comparable across each group. Four patients, two in the control group and one each in the LAR 10 and LAR 30 groups had ascites, well controlled on spironolactone. One Child's C patient was inadvertently entered into the saline group and dropped from the analysis.

Table 2. Patient demographics
 Sandostatin LAR 10 mg (n = 15)Sandostatin LAR 30 mg (n = 10)Saline (n = 14)
  1. LAR, long-acting release.

Age (years)Mean (95% CI)  
 54.7 (50.5–58.9)61.6 (54.6–68.7)57.3 (53.3–61.3)
Gender
Male10 (67%)6 (60%)10 (71%)
Female5 (33%)4 (40%)4 (29%)
Race
Caucasiann = 14n = 812
Black0n = 21
Oriental000
Othersn = 101
Aetiology of cirrhosis
Alcohol245
Hepatitis C705
Alcohol/hepatitis C222
Others442

Safety

Serious adverse events

Four of the 10 patients in the LAR 30 mg group dropped out of the study because of side effects, mainly abdominal cramps, diarrhoea and hypoglycaemia (Table 3). The patients dropped out of the study following the second drug injection, that is, after day 28. In one patient in the LAR 30 mg group who had autoimmune hepatitis (she was in clinical and biochemical remission for about 2 years at the time of entry into the study), there was a flare of the autoimmune disease. The elevation in the serum AST and ALT was preceded by intractable abdominal cramps and diarrhoea. This patient was withdrawn from the study and required high doses of prednisone for control of the autoimmune hepatitis. She subsequently developed fatal pulmonary aspergillosis. It was adjudicated that the death was not as a result of the study drug.

Table 3. Adverse events
 Saline (n = 14)LAR 10 (n = 15)LAR 30 (n = 10)P
  1. a

    Requiring hospitalisation, fatal or life-threatening, significantly incapacitating, or requiring medical intervention.

  2. b

    Adjudicated to be due to flare of autoimmune hepatitis and not study drug.

  3. c

    One patient who was diabetic needed to discontinue insulin during the study.

Deaths001b 
Seriousa004P < 0.001
Ab abdominal cramps002 
Diarrhoea001 
Hypoglycaemic001 
Injection site pain3/148/157/10P = 0.01
Abdominal cramps and diarrhoea9/1411/158/10P = 0.2
Constipation1/1400 
Flatulence1/142/153/10P = 0.4
Cholelithiasis/biliary colic000 
Hypertension000 
Dizziness012/10 
Hypoglycaemia002/10c 

Other adverse events

Among adverse events not listed as severe, there were more reports of injection site pain lasting greater than 24 h in the LAR 30 mg (7 of 10 patients) and LAR 10 mg (8 of 15 patients) groups than in the saline controls (3 of 14) P = 0.05. Abdominal cramps and diarrhoea not requiring treatment were similar between the three groups (8/10 in the LAR 30 mg group; 11/15 in the LAR 10 mg group; and 9/14 in the saline control group) P = 0.2. There was also no statistically significant difference in the frequency of constipation, flatulence, dizziness and hypoglycaemia between the groups. No patient in the study developed gallstones, biliary colic, hypertension, or cardiac dysrhythmias. No change in diuretic requirement was required for any patient.

HVPG: baseline fasting and postprandial (day 14)

There was no statistical difference in baseline HVPG measurements between the three groups (Figure 1). The HVPG increased in the postprandial state in all three groups, but the increase was not statistically significantly different between groups (2.45 ± 1.23 mmHg LAR 10; 2.61 ± 0.98 mmHg LAR 30; 1.7 ± 1.4 mmHg saline: P > 0.5)

Figure 1.

Hepatic vein pressure gradient (HVPG) fasting at day 0 and day 84 in all three groups was not statistically different.

HVPG: fasting and postprandial at day 84

The results of day 84 haemodynamic studies were available in only the six patients in the LAR 30 group who completed both studies. There was no significant decrease (P = 0.26 by anova) in fasting HVPG at day 84 as compared with baseline with LAR 30 mg (11.8 ± 2.3 mmHg vs. 14.1 ± 3.2); LAR 10 mg (15.3 ± 4.8 mmHg vs. 15.1 ± 3.8); or saline (13.3 ± 3.8 mmHg vs. 15.1 ± 4.3). There was no attenuation of the postprandial HVPG increase, either with LAR 10 mg or LAR 30 mg as compared both with baseline studies and with saline placebo (Figures 1, 2).

Figure 2.

There was no significant attenuation of the postprandial HVPG response with both LAR 10 mg and LAR 30 mg doses as compared with saline control.

Portal blood flow and SMA-PI

Portal blood flow did not change significantly (−97 mL/min with LAR 10 mg; +27 mL/min with LAR 30 mg; +4.7 mL/min saline; P = 0.8) (Figure 3a), whereas SMA-PI was almost identical (0.6) (Figure 3b).

Figure 3.

The increase in postprandial blood flow (PBF) at day 0 and day 84 was not statistically significantly different between LAR 10 mg, LAR 30 mg doses and saline control (a). The postprandial superior mesenteric artery pulsatility index was similar between LAR 10 mg, LAR 30 mg, and saline control (b).

Pharmacokinetics

Following multiple intramuscular injections of LAR depot every 28 days, the octreotide Tmax and AUC0–84 days in patients receiving LAR 30 mg were proportionately increased compared with patients receiving LAR 10 mg.The systemic clearance of octreotide was reduced in patients with cirrhosis (AUC 15 757 pg*h/mL) as compared with healthy subjects (10 750 pg*h/mL) with prolongation of the elimination half-life (t½ was 3.65 h in cirrhotics vs. 1.7 h in healthy controls). In only two patients (both in the LAR 30 group) did octreotide levels rise above 5000 pg/mL, the level achieved during continuous infusions of octreotide (8, 13). These levels were achieved in subject 8 on Day 56 (7280 pg/mL) and in subject 4 on Day 28 (5970 pg/mL). The plasma octreotide levels at day 84 were significantly higher in the LAR 30 mg group (3170 pg/mL) than in the LAR 10 mg group (1020 pg/mL) P = 0.02.

Plasma glucagon

The plasma glucagon levels at each study time point in each group are given in Table 4. The fasting levels in all patients were significantly higher at baseline as compared with normal values (<80 pg/mL). There was no significant decrease in glucagon levels compared with baseline at any time point during the study either with LAR 30 mg or LAR 10 (P = 0.56).

Table 4. Plasma glucagon concentrations (pg/mL) following octreotide LAR (Mean ± S.E.M.)
 Day 0Day 14Day 28Day 56Day 84
  1. LAR, long-acting release.

  2. a

    Data on six patients who completed the study.

Placebo152 ± 112123 ± 88123 ± 109145 ± 136128 ± 144
LAR 10 mg121 ± 113108 ± 127131 ± 161129 ± 132107 ± 152
LAR 30 mg149 ± 82100 ± 13895 ± 6691 ± 108130 ± 161
LAR 30 mga138 ± 64112 ± 11898 ± 6891 ± 108130 ± 161

Discussion

This trial demonstrated that in patients with cirrhosis and small oesophageal varices, there was a significantly higher rate of serious adverse events (40%) with LAR 30 mg administered monthly by intramuscular injection which could limit its use as an agent in the treatment of portal hypertension. There was no statistically significant reduction in HVPG, portal blood flow and SMA-PI in patients who received LAR 30 mg as compared with those who received LAR 10 mg and placebo, although we cannot discount the possibility of a type II error. Even if the LAR 30 mg dose was effective in reducing HVPG, the safety profile is far inferior to nonselective beta blockers, thereby considerably reducing the attractiveness of LAR in the treatment of portal hypertension. The LAR 10 dose was adequately tolerated, but the drug at that dose was ineffective in improving portal haemodynamics. The limited haemodynamic effect may be because an effective therapeutic plasma level of LAR was not achieved. Despite seemingly adequate dose intervals of the drug in patients in both treatment groups, as well as reduced metabolism of the drug in cirrhotics when compared with healthy subjects, only sub-therapeutic plasma levels of octreotide were achieved, and the decrease in levels in plasma glucagon in treated subjects was not optimal.

Our study has a number of strengths that internally and externally validate the negative findings. First, this trial was sufficiently powered to fulfil the primary objective of assessing the safety of LAR. Second, HVPG (a measure of portal pressure), portal blood flow and SMA-PI (measures of the hyperdynamic circulation) values in both fasting and postprandial states were measured at the start and at the end of the study, allowing for pre- and post-intervention comparisons to be made. Third, LAR levels were measured using radioimmunoassay, and plasma glucagon levels were measured in all subjects before and after the study drug to determine if a therapeutic level of LAR was achieved, and to determine the mechanism of action of octreotide on the haemodynamic response. Moreover, patients on beta-blockers or other medications known to impact portal pressures were excluded from the study, removing potential confounding variables. The randomised and blinded design of the study further minimised the impact of unknown confounding variables.

LAR 30 mg could not be tolerated by 4 of the 10 patients who were administered the dose because of more significant gastrointestinal side effects and hypoglycaemia. This would argue against attempting further studies using a higher or more frequent dose of octreotide LAR. Minor adverse events including abdominal cramps and diarrhoea not requiring treatment were not significantly different between the different groups. Injection site pain was more common in the LAR group than the saline control group.

Although this trial was adequately powered to primarily assess the safety of LAR, meaningful data were obtained on haemodynamic effects of LAR. The study suggests that a dose of LAR lower than 30 mg (for example, 20 mg) might be better tolerated, but doses higher or more frequent than 30 mg monthly may be necessary to obtain a significant haemodynamic effect. Targeting a higher level of plasma octreotide by more frequent dosing of octreotide LAR, or using higher doses, may be effective in reducing portal pressure, but our study suggests that higher dosing regimens of LAR may be associated with unacceptably high rates of serious adverse events. On the other hand, it is possible that a more significant haemodynamic response may be achieved in patients with Child Pugh Class C disease, or in patients with large varices, but it would be ethically improper to study these patients without placing them on a nonselective beta-blocker as a means of primary prophylaxis against variceal bleeding. It remains unknown at this time whether serious adverse events would be more frequent in patients with more advanced cirrhosis.

The results of this study are in contrast with another published clinical trial by Spahr et al. showing a positive impact of LAR on portosystemic pressure.[7] It is important to note that many of those patients were on nonselective beta-blockers during the study. That is, the subjects were likely to have had large varices. In contrast, in our study, patients with large varices were excluded. Thus, a combination of nonselective beta-blockers and LAR as used in the trial by Spahr et al. may have resulted in a more significant reduction in HVPG. There does remain the possibility that the beneficial effect on HVPG with a combination of LAR and beta-blockers may have largely been related to the nonselective beta-blockers.

A major approach in the management of patients with cirrhosis is reduction of PBF; this is usually accomplished in clinical practice in patients with acute variceal bleeding by vasoactive drugs like octreotide.[14-16] Whereas octreotide as a continuous infusion has been shown to decrease HVPG and azygos blood flow, lacks significant side effects and improves the initial control of bleeding and rate of re-bleeding, this effect was not seen in the present study. The lack of haemodynamic effect may be related to the fact that continuous high doses of octreotide may result in desensitisation because of internalisation of receptors.[13] This phenomenon is likely to be similar to the tachyphylaxis that occurs with acute intravenous administration of octreotide.

The lack of demonstration of a haemodynamic benefit following administration of LAR even at the 30 mg dose could possibly be due to a type II error. The study was powered primarily for safety, but had sufficient power to detect HVPG differences of >5 mmHg between groups. That is, the study may be underpowered to demonstrate smaller, but still significant (>10%) differences in HVPG between groups. However, it is much more likely that in the doses used in this study, LAR does not result in a significant haemodynamic benefit. In support of this conclusion, therapeutic levels of octreotide were not maintained in plasma, even though there is probably not a good correlation between plasma octreotide levels and decrease in HVPG. The mean maximum plasma octreotide concentrations of 3302 ± 2076 pg/mL seen at Day 56 in the LAR 30 group was well below the approximately 5000 pg/mL levels maintained with continuous infusion.[8, 13, 15] These levels were low in spite of the longer elimination half-life of octreotide in cirrhotic patients. In turn, these sub-therapeutic levels of octreotide in plasma were reflected by the lack of significant decrease in plasma glucagon levels, even with the LAR 30 mg dose. That is, neither the plasma octreotide levels nor the plasma glucagon levels reached a level in this study where a benefit on hepatic haemodynamics would be expected. In contrast, other studies have demonstrated a decrease in the postprandial elevation in glucagon following repeated octreotide boluses or infusion; the levels of glucagon following LAR injections in patients with cirrhosis have not been previously studied.[16-18]

This study did have several limitations, the most important of which was the relatively small sample size which precludes multivariate analysis. Moreover, only patients with Child Pugh Class A and B were included in the study; it is possible that the effects might be different in patients with Class C cirrhosis, or with large varices. In addition, the main study endpoint was safety; a change in HVPG was another end-point rather than a clinical outcome, such as variceal haemorrhage. No patients in the study had any gastrointestinal bleeding or a drop in haemoglobin during the course of the study, however, they were at low risk of variceal bleeding based on our selection criteria. It is important to emphasise that the quality of the HVPG tracings across centres was variable, and in some cases, interpretation was difficult. The blinded evaluator made final decisions regarding the pressure values in such situations. It is possible that if this study was carried out at a single centre where HVPG measurements are routine, the study results may have been different.

In summary, this study demonstrates that the LAR 30 mg dose was associated with significant side effects limiting possible use in the therapy of portal hypertension. In addition, both doses of octreotide LAR (10, 30 mg) did not significantly improve hepatic haemodynamics in patients with cirrhosis and small varices. These results do not support a role for long-acting octreotide in the prevention of variceal bleeding.

Acknowledgements

Declaration of personal interests: Dr. Kowdley serves on a data safety monitoring committee for Novartis for which he is paid an honorarium. Professor Jaime Bosch has relevant financial relationship of commercial interest with CHIASMA (consultant, research support), MICROTECH (consultant, research support) and GORE (research support). Declaration of funding interests: This study was sponsored and supported by Novartis Pharmaceuticals, East Hanover, NJ. The statistical analysis of the entire data sets pertaining to efficacy (specifically primary and major secondary efficacy endpoints) and safety (specifically, serious adverse events as defined in federal guidelines) have been independently confirmed by a biostatistician who is not employed by the corporate entity; and the corresponding author had full access to all of the data and takes full responsibility for the veracity of the data and analysis.

Ancillary