Rifaximin may induce rifampicin resistance in coagulase-negative staphylococci.

Rifaximin is an oral, poorly absorbed antibiotic which is very effective in the prevention of hepatic encephalopathy (HE), a potentially severe complication of liver disease arising mainly in patients with advanced cirrhosis. HE is caused by insufficient clearance of neurotoxins such as ammonia which are taken up from the intestine where they are produced by bacteria.1 It has been assumed that the antibacterial properties of rifaximin are mainly responsible for the prevention of HE. Therefore, its capacity to prevent other complications caused by bacterial translocation from the intestine in patients with cirrhosis has been assessed, but with surprisingly limited success so far.1 In line, closer analysis of the impact of rifaximin on the gut microbiome has revealed that its effect is more subtle than might be expected from a broad-spectrum antibiotic. Modulation of bacterial metabolism by rifaximin might be more important than its antibacterial properties.2, 3 In most analyses, the fact that, in addition to being an antibiotic, rifaximin is an agonist to the pregnane X receptor (PXR) has scarcely been paid attention to. By the study of de Wit et al,4 this lack of knowledge has been overcome. The authors demonstrate that the interaction between rifaximin and PXR is important for intestinal ammonia detoxification, suggesting that the mechanism of action by rifaximin for the prevention of HE may be rather induction of cellular ammonia metabolism than intraluminal antibacterial properties. The prophylactic intake of an antibiotic, nevertheless, raises the issue of induction of antibacterial resistance that might hamper antibiotic treatment in case of systemic bacterial infection. So far, these general concerns have not been proven true for rifaximin. However, only classical multi-drug-resistant bacteria have been in the focus of investigation.5, 6 Given that rifaximin is structurally unrelated to the antibiotics commonly used in many clinical infections, it is unlikely that microbial resistance is induced. However, rifaximin is closely related to rifampicin, which is clinically mainly relevant for staphylococcal infections, so that induction of rifampicin resistant is conceivable. In experimental settings, induction of rifampicin resistance by treatment with rifaximin has been reported, but data from the real world are lacking so far.7, 8 To clarify this issue, we retrospectively investigated bacterial infections occurring in hospitalised patients with advanced liver cirrhosis and paid particular attention to rifampicin resistance in staphylococci. All patients with cirrhosis admitted to the hepatology ward of the University Hospital Bonn between 1 July 2017 and 30 June 2018 were analysed for the occurrence of infections and isolated bacteria. From clinical charts, standard clinical, laboratory and microbiological data were retrieved. Identification of bacteria and antibiotic resistance was carried out by routine standard procedures at the local microbiological institute. This retrospective analysis of clinical data was approved by the local ethics committee of the University Hospital Bonn without requiring patient consent (decision number 313/18) and carried out according to the standards of the declaration of Helsinki. Statistical analysis was performed with Mann–Whitney U test for quantitative and with Pearson's chi-squared test for qualitative data. Bonferroni correction was used to adjust for multiple testing. A p-value below 0.05 was considered statistically significant. Overall, n = 341 patients with n = 501 episodes of hospitalisation were evaluated. Mean age was 60 years, n = 202 (59%) of patients were male. Stage of cirrhosis was Child–Pugh A/B/C in 47%, 45% and 23% of patients. Viral hepatitis and metabolic liver disease were common as cause of cirrhosis (63% and 18%). Ninety-six patients had also diabetes, n = 242 and n = 185 patients were on diuretics and beta-blockers respectively. N = 122 infections occurred while patients were without and n = 79 infections in patients taking rifaximin. As expected for a drug used for secondary prevention of a common complication of cirrhosis, patients with rifaximin displayed more advanced liver dysfunction as assessed by a higher median MELD (13 vs. 11.5; p < .001) and Child–Pugh score (7 vs. 6; p < .001) compared to patients without rifaximin. Table 1 shows that the frequency of different infections was comparable between the groups. Concerning isolated bacteria (n = 154 without and n = 92 with rifaximin treatment; Table 1), apart from a lower rate of Staphylococcus aureus isolation under rifaximin, the pattern was also comparable. Rifampicin resistance was detected only in one S. aureus isolated from a patient taking rifaximin. However, in coagulase-negative staphylococci, 68% (n = 13/19) were rifampicin resistant in patients taking rifaximin versus only 11% (n = 3/28) in patients without rifaximin (p < .001) (Figure 1C). All these isolates were regarded as contamination by colonising flora and not as causative for the underlying infection by the treating physician. The origin of the coagulase-negative staphylococci is shown in Table 1: the vast majority of these bacteria was detected in blood cultures or ascites both in patients with (74%) and without rifaximin (93%). Under rifaximin, 15% of isolated bacteria were classified as multi-drug-resistant, versus 10% of bacteria isolated from patients without rifaximin (p = .24). The impact of administration of rifaximin on antibiotic resistance of bacterial infections in patients with cirrhosis in a real-world setting has incompletely been studied, although it is a clinically very relevant issue. In contrast to other potential indications for treatment with rifaximin, such as infectious enteritis, use of rifaximin for prevention of hepatic encephalopathy is a long-term treatment, making induction of antibacterial resistance more likely. Of note, in healthy volunteers, short-term administration of rifaximin was associated with induction of rifaximin-resistant bacteria in the stool, which, however, rapidly vanished after end of treatment.9 Large trials assessing prevalence of multi-drug-resistant bacteria did not identify treatment with rifaximin as an independent risk factor.5, 6 This fact is not surprising, because rifaximin is not structurally related to any antibiotic defining multi-resistance, such as penicillins, carbapenems or vancomycin. By contrast, rifaximin is closely related to rifampicin, an antibiotic mainly used to treat tuberculosis, bone infections and complicated staphylococcal infections. Therefore, the possibility that oral rifaximin, albeit poorly absorbed, leads to emergence of rifampicin resistance in colonising staphylococci is conceivable. In support of this hypothesis, administration of rifaximin for 7 days to healthy volunteers led to emergence of rifampicin-resistant staphylococci in 64% (7/11) of individuals.7 Similarly, 44% of 25 cirrhotic patients receiving rifaximin for prevention of HE developed rifampicin resistance in coagulase-negative staphylococci, mostly isolated from perianal skin.8 Comparable results were obtained from non-cirrhotic patients receiving rifaximin as prophylaxis before colorectal surgery, with emergence of staphylococcal rifampicin resistance in 43%.10 Data from real-world cirrhotic patients were missing so far. Our study indicates that induction of rifampicin resistance in cirrhotic patients receiving rifaximin may be observed also outside of a study setting. Fortunately, this effect was only detectable in coagulase-negative staphylococci not causing overt infection. Therefore, this potential side effect may in most cases not become clinically relevant. However, in patients with complicated staphylococcal infections, such as endocarditis or bone infection, resistance to rifampicin may limit treatment options. Our study is limited by being a monocentric study, including a medium-size cohort of patients and by being retrospective. However, with coagulase-negative staphylococci being often not analysed for reflecting rather contamination than cause of infection, our study offers the advantage of a broad analysis. In addition, our results are in line with data from experimental settings. While our findings may not have consequences for the majority of cirrhotic patients, they may become highly relevant in individual patients at risk of complicated staphylococcal infections, such as patients with prosthetic heart valve. Rifaximin constitutes an effective option for prevention of hepatic encephalopathy in patients with cirrhosis. The study by de Wit et al.4 expands our knowledge by suggesting that rifaximin may prevent HE rather by being a PXR agonist than by modulation of intestinal bacteria. However, the potential, but rare side effects of the antimicrobial properties of rifaximin must be taken into account. Open Access funding enabled and organized by Projekt DEAL. None of the authors has a conflict of interest regarding this article.