What is already known on this topic
The intestinal microbiota is implicated in the metabolism of many medical drugs and contributes to the interpersonal variation in drug efficacy and safety. However, how this happens is still unclear.
What this research adds
This study proposes a new quantitative approach both experimental and computational to understand the real contribution of the microbiome in the metabolism of some medical drugs and to clearly distinguish it from the contribution of the host himself.
Knowing and quantifying the relationship between the microbiome and the host in the metabolism of medical drugs could help to adjust their response and safety. The proposed method could be applied to other active ingredients, nutrients or endogenous metabolites.
Several metabolic reactions are activated at various levels when a drug is taken, both by the host and the microorganisms colonizing it, especially intestinal bacteria. This increases the variability of inter-individual response to drugs in terms of efficacy and safety. But how is it possible to distinguish and quantify the contribution of the bacterial microbiome from that of the host, especially when the metabolic reactions performed are the same? Answering this question could help to influence the efficacy and safety of drugs with a personalized approach.
Michael Zimmermann and colleagues at Yale University in the United States proposed a novel approach in a study published in Science. The aim of their research was to develop an experimental and computational method to quantify the actual contribution of the bacteria to the metabolism of some drugs: brivudine (antiviral), sorivudine (antiviral) and clonazepam (benzodiazepine). To do this, appropriate mouse models, conventional (CV), germ-free (GF) and gnotobiotics (GN) were developed, through which it was possible to follow the pharmacokinetics of the selected active ingredients with the results shown below.
Brivudine metabolism: host vs microbiota
Brivudine (BRV) is an antiviral drug metabolised in the liver in bromoviniluracil (BVU) by both the host and the bacteria that colonize it.
To determine how much the contribution of the microbiota, compared to that of the host, really weighs on the metabolism of the drug, the researchers administered it to the CV and GF models, then comparing the serum levels and intestinal and hepatic concentrations of both BRV and BVU with the following observations:
- serum levels of the BVU metabolite in conventional models were five times higher than those of the counterpart without a parallel decrease in the BRV form, which suggests a substantial contribution of the intestinal microbiota
- BRV kinetics in the duodenum was similar in both models
- GF models recorded the highest BRV values in the gastrointestinal tract and faeces; on the other hand, intestinal levels of BVU were higher in CV than in GF models
- the increased serum concentration of BVU in the CV compared to the GF models is accompanied by an increase in the hepatic level
Some studies have shown that BVU interferes with the hepatic metabolism of pyrimidines by binding to the enzyme dihydropyrimidine dehydrogenase (DPD) with potentially lethal consequences especially for patients undergoing therapy with pyrimidine-like drugs (5-fluorouracil for example). It has been observed that, following administration of BRV, the CV models accumulate more endogenous DPD substrates than the GF ones, confirming once again the contribution of the microbiota.
Identification of the species involved
After confirming the role of the microbiome, researchers went on to identify the species directly involved in the process by selecting the eight ones usually more expressed in the intestine of mammals and tracing their ability to convert BRV into BVU.
Bacteroidetes thetaiotamicron and B. ovatus have shown the best metabolic activity and genomic mapping analyses have identified bt4554, a gene coding for the enzyme purine nucleoside phosphatase 2, as responsible for this activity.
The research focused on better understanding the role of the identified gene by administering BRV after colonizing gnotobiotic (GN) mouse models with the Bacteroidetes thetaiotamicron wild-type (WT) and with the (MUT) strain, mutated not to express the gene, respectively. The capacity for intestinal growth and colonization of the two variants proved to be comparable:
- serum BRV levels were similar in the two groups (GN-WT vs GN-MUT) suggesting that the bacterial metabolic activity does not affect the bioavailability of the drug or its elimination
- on the other hand, the GN-WT group recorded serum values of the BVU metabolite significantly higher than those of the counterpart. Similar results were also found at the hepatic level
- as for the CV and GF models, the increase in systemic BVU in GN-WT is accompanied by a significant intestinal metabolism of BRV
- the Bacteroidetes thetaiotamicron WT strain completely converts BRV into the cecum. The resulting BVU is then absorbed at the level of the cecum and colon. In the GN-MUT models, given its fecal concentrations, intestinal absorption of the metabolite was reduced
Developing a pharmacokinetic model
From the more experimental approach, the researchers switched to a computational one and developed a pharmacokinetic model to quantify the contribution of the host and of the microbiota (GN-WT vs GN-MUT) respectively to the systemic metabolism of the drug.
The parameters taken into account were: the passage of BRV from the intestine to the bloodstream; the elimination of BRV (location and speed); the conversion from BRV to BVU mediated by the host; the elimination of BVU; the intestinal transit.
The model that was generated showed:
- to predict the kinetics of BRV in the GN-WT serum as well as the host and bacterial contribution to the serum levels of BVU with reasonable accuracy
- that although the microbial activity is the most responsible for the serum concentration of BVU, the host variability greatly affects the bioavailability
- to be applicable also in conventional mouse models
To validate the method, the researchers tested it by administering to CV and GF models other drugs, namely sorivudine (SRV), a BRV analogue for structure and activity, and clonazepam. The drugs’ levels and related metabolites were then monitored over time in both intestinal tissues and serum.
In both cases, the pharmacokinetic model allowed to distinguish between the host and the bacterial contribution to the metabolism of the drug, even if the chemical processes were identical.
In conclusion, in the metabolism of a drug the bacterial component offers an important and interconnected contribution to the activity of the host. Understanding and quantifying the dynamics of this contribution to predict effectiveness and safety represents a great advantage in the field of personalized medicine.
The pharmacokinetic approach proposed in this work could be applied to the study of the transformation of nutrients, xenobiotics or endogenous metabolites.
Translated from italian by MicrobiomePost’s editorial staff