A phase 2 randomised controlled trial of serelaxin to lower portal pressure in cirrhosis (STOPP)

We enrolled male or female adult participants over the age of 18 years with cirrhosis and PH. Full eligibility criteria are listed in Additional file 1. The main inclusion criteria included: clinically diagnosed or biopsy-proven liver cirrhosis of any aetiology; evidence of PH either on imaging or on previous endoscopy; and suspected HVPG ≥ 10 mmHg at baseline (if the baseline HVPG was subsequently found to be < 10 mmHg, the participant was withdrawn from the study). The main exclusion criteria included: pregnancy or breast-feeding; women of child-bearing potential not using highly effective methods of contraception; severe liver failure; history of variceal bleed within the previous month; hepatocellular carcinoma or history of malignancy of any organ system (other than localised basal cell carcinoma of the skin); portal vein thrombosis; previous surgical shunt or TIPSS; current use of beta-blockers or nitrates, or any other drug therapy known to have an influence on portal pressure (diuretics were permitted provided patients had been on a stable dose for at least 30 days); history of active/recent drug or alcohol abuse; sitting systolic blood pressure < 110 mmHg at screening visit or within 10 min prior to starting study drug infusion; significant arrhythmias, including prolonged QT interval; documented hypersensitivity to i.v. contrast agents and/or iodine; severe renal impairment; significant structural heart disease (including cardiomyopathy, valvular disease); major neurologic events, including cerebrovascular events, within the previous month; clinical evidence of acute coronary syndrome currently or within the previous month; and pacemaker, cardiac resynchronisation device or implantable cardioverter-defibrillator in situ.

This was a single-site study, undertaken at the Royal Infirmary of Edinburgh, Edinburgh, UK between 19 October 2017 and 15 August 2018.

Participants attended the RIE Clinical Research Facility for screening (visit 1) consisting of a physical examination, blood tests (full blood count, coagulation and biochemistry), electrocardiogram (ECG), blood pressure measurement and written informed consent. Randomisation was performed once it was known that the participant had passed screening, prior to the study visit.

On the study day (visit 2; ≤ 7 days after the screening visit), eligible participants attended for baseline haemodynamic measurements, following an overnight fast and the avoidance of caffeine for > 8 h. After baseline evaluation and confirmation of HVPG ≥ 10 mmHg, participants received (in a double-blind fashion) either serelaxin or placebo. The haemodynamic measurements were repeated at specified time points. A peripheral blood sample was taken at baseline and after 2 h, processed and stored for potential future analysis. After the post-treatment assessments, participants were observed for a recovery period of 4 h which included repeat physical examination, blood pressure, ECG measurement and routine laboratory blood tests. There was no follow-up visit. Participants were contacted by a member of the research team via telephone at 24 h and again at 4 weeks after the study visit to collect information about potential AEs and concomitant medications.

Study data were collected and managed using REDCap electronic data capture tools hosted at The University of Edinburgh. REDCap [13] (Research Electronic Data Capture) is a secure, web-based application designed to support data capture for research studies, providing: an intuitive interface for validated data entry; audit trails for tracking data manipulation and export procedures; automated export procedures for seamless data downloads to common statistical packages; and procedures for importing data from external sources.

This study was a phase 2, double-blind, randomised controlled trial to investigate the effects of serelaxin on PH in patients with cirrhosis. Randomisation was carried out by an independent third party (Edinburgh Clinical Trials Unit) after it was confirmed that the participant had passed screening, prior to the study visit (visit 2). Random sequences of block sizes were generated by computer to achieve a 3:1 allocation ratio between serelaxin and placebo; there was no stratification to this allocation. Randomisation produced a four-digit integer matching a bottle number held by the pharmacy. The Investigators had no way of linking the drug allocation to the four-digit number. The person generating the randomisation list and allocation concealment was not involved in the later implementation of the sequence. The original randomisation list was held in a secure folder with restricted access to the Edinburgh Clinical Trials Unit Data Management team. The placebo was matched to serelaxin on appearance and administration protocol to create and maintain blinding.

The primary efficacy endpoint was the decrease in the fasting HVPG between baseline and 2 h post serelaxin treatment, targeting for a 20% reduction. The sample size calculation was based on a previous study in Edinburgh evaluating carvedilol [14] and the data from the previous Novartis-sponsored serelaxin phase 2 study (ClinicalTrials.​gov identifier: NCT01640964) [11]. Assuming a mean baseline HVPG of 16.37 (SD = 2.14) mmHg and a post-baseline HVPG of 13.1 (SD = 3.91) mmHg (20% decrease), the change from baseline in the HVPG was estimated to be 3.3 (SD = 4) mmHg. A sample size of 14 participants in the serelaxin group would provide 80% power to detect at least a 20% decrease from baseline in HVPG using a two-sided paired t test with an α level of 0.05. A small number of placebo-treated patients were included in order to preserve double-blindness, not as a comparison group. Therefore, it was proposed that a total of 20 patients (15 serelaxin and 5 placebo) would be randomised in a 3:1 ratio.

Recombinant human relaxin-2 (serelaxin; Novartis Pharmaceuticals, UK) or placebo (20 mM sodium acetate buffer solution at pH 5.0; Novartis Pharmaceuticals, UK) were administered via peripheral i.v. infusion at two different infusion rates: 80 μg/kg/day for 60 min followed by 30 μg/kg/day for at least 60 min (until completion of the final HVPG/ICG measurements). This was achieved by a single infusion bag, prepared by the clinical trials pharmacist, with a uniform change in the administration rate.

The primary endpoint was the change from baseline in the fasting HVPG after 2 h of serelaxin infusion. The HVPG was measured as previously described [14]. The procedure was performed after overnight fast and at the same time of day due to circadian variation in HVPG measurements. Prior to catheter insertion, participants were offered low-dose (≤ 0.02 mg/kg) midazolam to reduce any anxiety. A 7-F venous introducer was inserted into the right femoral vein using the Seldinger technique under ultrasound guidance. A balloon-tipped catheter was then advanced into a hepatic vein using fluoroscopy. The free hepatic venous pressure (FHVP) was measured with the balloon deflated and floating freely in the hepatic vein close to its junction with the inferior vena cava (IVC). The wedged hepatic venous pressure (WHVP) was measured with the balloon inflated until the branch of hepatic vein was completely occluded. The HVPG was obtained by subtracting the FHVP from the WHVP. All measurements were performed in triplicate and permanent tracings were printed, stored and read blindly at the end of the study prior to the opening of the randomisation codes. The HVPG was measured at baseline, and then repeated after 60 and 120 min of the IMP (either serelaxin or placebo). Baseline HVPG ≥ 10 mmHg confirmed the presence of CSPH. If HVPG < 10 mmHg was obtained, the study participant was withdrawn. The IVC pressure (IVCP) was measured at baseline and after the final HVPG measurement.

Secondary endpoints included: the change from baseline in the fasting HVPG after 1 h of serelaxin infusion; the change from baseline in the fasting hepatic blood flow (HBF) after 2 h of serelaxin infusion (measured from the concentration of indocyanine green (ICG) in the hepatic venous blood vs peripheral venous blood using the Fick Principle); the change from baseline in the IVCP after 2 h of serelaxin infusion; the change from baseline in the cardiac index (CI) after 2 h of serelaxin infusion; the change from baseline in the systemic vascular resistance index (SVRI) after 2 h of serelaxin infusion; the change from baseline in the aortic pulse wave velocity after 2 h of serelaxin infusion; safety and tolerability of serelaxin infusion (as assessed throughout the study by monitoring AEs, clinical laboratory blood tests, heart rate, blood pressure and ECG); and the change from baseline in exploratory blood biomarker measurements after 2 h of serelaxin infusion (if a demonstrable effect on the HVPG was observed).

The total HBF was calculated using the ICG constant infusion method [15, 16] and derived from measurements of ICG clearance and extraction. Baseline serum samples were taken prior to each ICG infusion. Thereafter, 10 mg of ICG (10 ml) was given as a slow i.v. bolus via a peripheral cannula, followed by an infusion of 0.2 mg/min (0.2 ml/min or 12 ml/h) by accurate infusion pump (Alaris Asena, Becton Dickinson, USA). After an equilibrium period of at least 40 min, samples were taken simultaneously from the right hepatic vein (via the catheter tip) and the femoral vein (via the side port of the introducer). Paired samples (hepatic and femoral) were taken in triplicate, 2 min apart, in order to confirm equilibrium. The HBF was measured in this way both at baseline and after 120 min of IMP infusion.

A Cardioscreen 1000 (Medis, Germany) was used for non-invasive measurement of cardiac output (CO (L/min) = heart rate × stroke volume / 1000), CI (L/min/m²) and SVRI (dyne·s·cm⁵·m²) by the bio-impedence technique before and after 120 min of IMP infusion. Similarly, arterial function was measured using an Arteriograph device (TensioMed, Hungary). Arterial stiffness and central haemodynamics were assessed by the application of an inflatable cuff to the upper arm. The aortic pulse wave velocity (APWV, m/s) was calculated as the distance the pulse wave travels in the aorta (as measured from the suprasternal notch to pubic bone) divided by the measured transit time.

Participants were monitored for 4 h after IMP discontinuation and removal of the catheter and introducer from the femoral vein. Vital signs were recorded every 30 min throughout the infusion and recovery periods, with regular ECG monitoring. Adverse events (AEs) were collected during visit 2 and by follow-up telephone calls at 24 h and 4 weeks. The severity, expectedness and causality of AEs in relation to the study medication were noted by the study team. The Investigators and the co-sponsors had the right at any time to terminate the study for clinical or administrative reasons.

Analysis was per protocol. The number of participants who were withdrawn by the investigator during the course of the study is presented broken down by treatment allocation and presented with reasons for withdrawal where available. Missing data as a result of patients not having the post-baseline measurement were not imputed. Patients with missing post-baseline data were excluded from the analysis at that time point.

Summary statistics (n, mean, SD, median, minimum, maximum, Q1 and Q3) were generated over time for the baseline, post-baseline and change from baseline measurements for the primary endpoint in the serelaxin and placebo group. The geometric mean was calculated for the baseline value, the post-baseline values and the ratio to the baseline values. Confidence intervals were calculated for both the arithmetic and geometric means. Paired t tests were used to test the mean change from baseline measurements. The secondary endpoints were subjected to the same analysis as the HVPG. For the placebo control group, the change from baseline to 2 h was analysed in the same way as the primary outcome, although, as this was not powered for, no direct statistical comparison was made between serelaxin and placebo.

Participant disposition is shown in the CONSORT diagram (Fig. 1). A total of 17 participants were recruited. Of these, two participants had a screening failure and did not proceed to randomisation. Fifteen patients were randomised and 11 completed the trial (n = 9 serelaxin, n = 2 placebo). Four participants were withdrawn by the investigators following randomisation, did not receive any study drug and are not included in any primary or secondary analyses. Reasons for withdrawal were baseline HVPG < 10 mmHg (n = 2) and HVPG technical failure (n = 2). The trial ended early due to manufacturer discontinuation of the study drug.

Participant characteristics are summarised in Table 1. The median age of participants was 56 (range 43–69) years and 73% were male. Cirrhosis aetiologies were alcohol-related liver disease (n = 10), non-alcoholic fatty liver disease (n = 2), chronic hepatitis C (n = 2) and chronic hepatitis B (n = 1). Participants were Child–Pugh class A (60%) and class B (40%) with a median MELD score of 10 (range 6–14). The mean baseline HVPG was 16.3 (range 10.3–21.7) mmHg.

In those allocated to the serelaxin arm (n = 11), two participants were withdrawn. The arithmetic mean ± SD HVPG at baseline was 15.9 ± 3.3 mmHg compared to 15.6 ± 4.3 mmHg after 2 h of serelaxin infusion. Although individual responses were variable (Fig. 2), there was no evidence of a significant change in the fasting HVPG between the baseline and 2 h time point (arithmetic mean of difference ± SD was 0.4 ± 3.5 mmHg (95% CI –2.3, 3.1; p = 0.76). In the two placebo-treated patients, the HVPG decreased over the 2-h observation period, but no further analysis or conclusions can be made in such a small sample (Table 2).

For all of the other secondary endpoints, which were measured after 2 h of serelaxin infusion (including HBF, IVCP, CI, SVRI and APWV), no substantial changes were observed (data summarised in Table 2).

Treatment with serelaxin was well-tolerated. Overall, 12 AEs were reported in 7 participants treated with serelaxin (Table 3). None were serious or considered related to the IMP. There were no serious adverse events (SAEs) in this study. No pregnancies were reported. There were no striking changes in laboratory blood tests monitored during the course of the study (Table S1 in Additional file 2); in particular, there were no newly occurring liver enzyme abnormalities observed. According to the product label, hypotension is a potential side effect of serelaxin. Patients with cirrhosis and PH often have lower baseline blood pressure levels, predominantly due to severe splanchnic vasodilatation. Following serelaxin, we observed a statistically significant increase in the heart rate (baseline 65 ± 8 bpm vs 2-h post serelaxin 72 ± 8 bpm, p = 0.02) and decrease in the mean arterial pressure (baseline 93 ± 7 mmHg vs 2-h post serelaxin 88 ± 5, p = 0.02) due to a reduction in diastolic rather than systolic blood pressure (Table S2 in Additional file 2). However, changes were not clinically significant and there were no discontinuations due to tachycardia or hypotension.

The main limitation is that the study was terminated before the recruitment target was met due to slow enrolment and, ultimately, a global drug supply issue (Novartis stopped manufacturing serelaxin and there was none available with a shelf-life beyond 31 August 2018). Therefore, based on the sample size calculation, the study is underpowered to detect the primary endpoint. The study was double-blind and placebo-controlled, which would have addressed potential sources of bias. A formal dose-ranging study of serelaxin in cirrhosis patients has not yet been undertaken. We used the same infusion regimen that had previously shown encouraging haemodynamic effects [23] and had achieved similar steady-state serum concentrations to that observed in our 72-h rat cirrhosis models [10] and in human heart failure following 48 h of i.v. serelaxin infusion [24]. However, the biological effects of relaxin are known to follow a U-shaped dose–response curve [25] and we do not know whether serelaxin might have induced more pronounced effects on the HVPG or secondary haemodynamic endpoints at higher (or lower) doses. Any future work on relaxin effects in cirrhosis should address dose–response relationships.