Rifaximin reduces the incidence of spontaneous bacterial peritonitis, variceal bleeding and all‐cause admissions in patients on the liver transplant waiting list

Summary Background Rifaximin reduces the risk of overt hepatic encephalopathy (HE) and is associated with significant reductions in hospitalisations and 30‐day readmissions. Aim To examine the outcomes of patients listed for liver transplantation with a diagnosis of HE on rifaximin compared to those naïve to the drug. Methods Patient records of those listed for liver transplantation over a 2‐year period were retrospectively reviewed. Patients were included if they had at least two episodes of overt HE resulting in hospitalisation or were encephalopathic at the time of assessment. Results Of the 622 patients listed for transplantation, 101 had HE. Sixty‐six patients were treated with rifaximin and 35 were naïve at listing. The use of concurrent lactulose was not significantly different between groups. Median MELD score was similar (15 [14‐16)] rifaximin‐treated and 16 [14‐18] rifaximin‐naïve). Patients on the waiting list treated with rifaximin had reduced all‐cause admissions, episodes of spontaneous bacterial peritonitis and variceal bleeding. Mean length of stay was 9 days (95% CI 6‐12) in the rifaximin‐treated group vs 14 (95% CI 7‐21) in the rifaximin‐naïve group. Multivariate regression analysis demonstrated that rifaximin was independently associated with an increase in average days to readmission (adjusted effect estimate 71, 95% CI 3‐140 days) and reduced likelihood of requirement for prioritisation on the waiting list (odds ratio 0.29; 95% CI 0.89‐0.93). Conclusion Rifaximin prescribed for HE in patients listed for liver transplantation improved outcomes with significant reduction in admissions related to spontaneous bacterial peritonitis, ascites and variceal bleeding.


| Study design and setting
The patient records of 622 patients with confirmed cirrhosis (the diagnosis of cirrhosis was confirmed by a combination of a least 2 modalities: clinical, biochemical, radiological and histopathological) who were listed for liver transplantation at King's College Hospital NHS Foundation Trust over a 2-year period [1st January 2014 -31st January 2016] were retrospectively reviewed.

| Participants
Patients were included if they had at least two historic episodes of overt HE resulting in hospitalisation or were overtly encephalopathic at the time of assessment. Patients under the age of 18 were excluded from the study.

| Data collection
Information collected included patient demographics, aetiology of liver disease, Child Pugh Turcotte score, 12 Model for End-stage Liver Disease (MELD) score 13 at the time of assessments, United Kingdom End-stage Liver Disease (UKELD) score, maximum grade of HE (defined using the West Haven criteria), 14 blood ammonia concentration (venous), concurrent lactulose therapy, medical co-morbidities, emergency admission whilst on the waiting list including admissions to high dependency and intensive care beds, requirement for prioritisation (UKELD score ≥63), duration on the waiting list (days) and mortality on the waiting list. Elective admissions, such as for large volume paracentesis were excluded unless the paracentesis was complicated (defined as necessitating a hospital admission for greater than 24 hours) which may occur for example due to spontaneous bacterial peritonitis, acute kidney injury or electrolyte disturbance.

| Statistical methods
The primary outcome was defined as the number of days to all cause readmission on the transplant waiting list. Secondary outcomes evaluated included requirement for prioritisation, hospital admissions with sepsis, variceal bleeding, ascites and hepatic encephalopathy, length of hospital stay, intensive care admissions, length of intensive care stay and mortality on the waiting list.
Normality testing was undertaken on all continuous clinical and biochemical data using the D'Agnostino-Pearson omnibus normality test. Normally distributed data are presented as mean with 95% CIs and non-normally distributed data as median and interquartile range. Univariate statistical analysis of non-normally distributed unpaired data was completed using the Mann-Whitney test and analysis of normally distributed data using unpaired t tests with Welch's correction assuming unequal standard deviations.
Categorical variables are expressed as number and proportion and compared using the X 2 test or Fisher's exact test, as appropriate. Hospital admission data were presented as annualised rates. Multivariate linear regression analysis was performed on admission and complication data (related to sepsis, acute variceal hemorrhage, encephalopathy and complications of ascites) identified as the dependent variable. Binary logistic regression analysis was subsequently performed with requirement for prioritisation on the transplant waiting list (yes/no) as the dependent variable.
Forward selection was used to select independent variables within the regression models with P < 0.1; we also included variables considered by subject knowledge or literature (ie lactulose use and MELD score) to be associated with hospital readmission in endstage chronic liver disease. Univariate and regression analyses were performed using IBM spss Statistics 24 for Mac. A P < 0.05 was considered to be statistically significant.

| Patient demographics
Of the 622 adult patients listed for transplantation, 101 were listed with HE. Sixty-six patients were treated with rifaximin and 35 were naïve. There was a male preponderance in both groups and alcohol-related liver disease was the most common aetiology. The rifaximin-treated group was marginally older mean age 55 vs 49 (mean difference 5.85 years [95% CI 0.72-10.98]). Organ severity scores were similar; mean MELD score was 15 (95% CI [14][15][16] in the rifaximin cohort and 16 (95% CI [14][15][16][17][18]   acute overt HE did not significantly differ and neither was mortality on the waiting list impacted upon despite the reduced rate of variceal bleeding, complications of ascites and all-cause hospitalisation in the rifaximin-treated cohort. The overall waiting list mortality in this study was 13.86%. Clinical outcomes are summarised in Table 2 and Figure 1.

| Clinical outcomes
Multivariate linear regression analysis with days to readmission (related to complications of ascites, variceal bleeding, hepatic encephalopathy or sepsis) defined as the dependent variable demonstrated that rifaximin treatment was independently associated with increased length to all-cause readmission whilst on the liver transplant waiting list; adjusted effect estimate 71 (95% CI 3-140 days) when adjusting for age, sex, BMI, disease severity score and concomitant lactulose use. Similarly, rifaximin use was independently  and has been demonstrated to be safe and well tolerated in patients for long-term maintenance of remission from overt HE. 16  score of ≤10, two-thirds with a MELD score between 11 and 18 and <10% with a MELD between 19 and 24. We therefore set out to determine the impact of treatment with rifaximin in patients listed with overt HE whilst they were on the liver transplant waiting list at a large UK transplant centre.

| D ISCUSS I ON
One of the strengths of this study was that whilst a large proportion of the patients included in this study had alcohol-related cirrhosis because they had been abstinent for at least 6 months prior to listing, active alcohol intake was not a confounder in this study as it has been in other 'real-world' studies 10,11 where it was difficult to determine if abstinence from alcohol or rifaximin was the major driver in the improved outcomes in this cohort of patients.
Patients with advanced cirrhosis are susceptible to unplanned emergency hospitalisations for a variety of reasons, including susceptibility to infection, 17 overt HE, 5 acute kidney injury, acute variceal bleeding, electrolyte disturbance, large volume ascites, spontaneous bacterial peritonitis, falls, malnutrition and sarcopenia.
These patients frequently progress to requiring high dependency or intensive care support 2,18 and may at any time progress to developing ACLF. 19,20 Infection is the leading cause of death in patients with end-stage liver disease and confers a fourfold increased mortality compared   with rifaximin was associated with a reduction in length of hospital stay and provides good value for money in terms of health economics and resource utilisation. 10,11 In summary, rifaximin prescribed for the recurrence of overt HE in patients listed for liver transplantation improved outcomes on the waiting list with a significant reduction in hospital admissions related to decompensation, variceal bleeding and complications of ascites.
There was a reduced requirement for prioritisation of patients on the waiting list in patients treated with rifaximin and an increased time to hospital readmission. This study provides 'real-world' data that demonstrates the potential value of rifaximin in reducing hospital admissions and length of stay within the advanced cirrhotic population awaiting liver transplantation. At this point, however, the data should not be interpreted as a reason to change clinical practice but should act as a catalyst for further prospective studies in this patient group.