To the best of our knowledge, this is the first experimental study investigating 5-FU-induced microbiota alterations, as well as the potential of prebiotic fibre mixtures to counteract the manifestation of microbial dysbiosis during 5-FU administration, using a whole human-derived gut microbial consortium in the TIM-2 model. We observed that administration of 5-FU induced changes in gut microbiota composition, but not in overall microbial α-diversity. As a next step, we showed that different prebiotic fibre mixtures increased the abundance of potentially beneficial bacteria (e.g. Bifidobacterium, Lactobacillus, Anaerostipes) and inhibited the growth of potentially pathogenic bacteria (e.g. Klebsiella, Enterobacter) in the presence of 5-FU and could therefore be used to prevent 5-FU-induced microbial changes. In addition, we examined the effect of 5-FU with and without prebiotic fibre mixtures on cumulative levels of SCFA and BCFA and observed that acetate increased slightly due to 5-FU, but even more due to 5-FU + prebiotics, while propionate decreased. Compared to the control condition, BCFA were increased due to 5-FU and decreased by 5-FU + prebiotics.
First of all, we were interested in major 5-FU-induced shifts in overall gut microbiota composition. Therefore, we chose to focus on the abundance of the most abundant taxa on phylum and genus level. In contrast to other studies, which were performed in rats (34), mice (23), or with individual bacterial strains (19, 22), we investigated the effects on a whole human-derived microbial consortium. In line with previous results (19, 22, 23, 34), we identified a considerable 5-FU-induced shift in gut microbiota composition. On phylum level, we observed an altered Bacillota/Bacteroidota (Firmicutes/Bacteroidetes) ratio due to 5-FU administration, caused by a relative increase of Bacillota (Firmicutes) and a relative decrease of Bacteroidota (Bacteroidetes). This observation was contradictory to the results from Spanogiannopoulos et al. who reported no significant differences between phyla concerning 5-FU sensitivity (19). Due to their well-described pro-inflammatory character and association with microbial dysbiosis (58), we expected an increased abundance of Pseudomonadota (Proteobacteria) during 5-FU treatment, which was only found in one of the duplicate experiments. Possibly, Pseudomonadota (Proteobacteria) overgrowth might require a prolonged period of 5-FU administration and/or simulation of an inflammatory state (oxygen influx), which was not simulated in the system here and should be investigated further in future studies.
On genus level, 5-FU seemed to have an inhibitory effect on Prevotella and a more stimulating effect on Bacteroides, CAG-352 (a genus from the Ruminococcaceae family), Faecalibacterium, and Blautia. In line with this, Spanogiannopoulos et al. also found significant growth inhibition of a Prevotella strain (19), while Blautia strains seemed to be relatively insensitive to 5-FU in different studies (19, 22). In the same study, the growth of a Faecalibacterium strain was also significantly inhibited by 5-FU (19), which was not the case in our experiments. However, it should be noted that sensitivity towards 5-FU was very strain-specific (19, 22), which makes it difficult to compare previous results on individual strains with our data on genus level in a complex consortium. For instance, some Bacteroides strains showed complete growth inhibition at the maximum dose of 5-FU, while other Bacteroides strains could still grow (19). In contrast to an earlier study, showing reduced α-diversity during 5-FU administration in tumor-bearing mice (23), microbial diversity did not change due to 5-FU in our experiments.
Prebiotic fibres are commonly used in humans to promote a balanced gut microbiota and to counteract microbial dysbiosis. However, limited knowledge exists on whether cancer patients, receiving chemotherapy, would also benefit from prebiotic fibre administration. Therefore, we aimed to examine whether selected prebiotic fibre mixtures also stimulate potentially beneficial bacteria in the presence of 5-FU and could therefore be used to prevent overgrowth of potentially pathogenic bacteria and the manifestation of 5-FU-induced microbial dysbiosis in cancer patients. Our experiments provided several insights which will be of great value for the future design of targeted interventions.
Firstly, all prebiotic fibre mixtures under investigation stimulated Bifidobacterium, Lactobacillus, and Anaerostipes, which was in line with our expectations. While Bifidobacterium and Lactobacillus are well-known for their beneficial probiotic properties (59), Anaerostipes has also been linked to improved host health and production of SCFA (60). On the other hand, it should be noted that the preTA operon, involved in 5-FU metabolism, has also been identified in strains belonging to Anaerostipes and Lactobacillus (19), but it requires further investigation whether these genera actually play an active role in 5-FU metabolism. Besides, the genera Weissela, Olsenella, and Senegalimassilia were associated with prebiotic fibre mixture administration in our experiments. In particular, Weissela has been suggested to exert probiotic properties as well as antimicrobial effects against pathogens, such as E.coli and F. nucleatum (61). The exact physiological functions of Olsenella and Senegalimassilia, which both belong to the family of Coriobacteriaceae (which is part of the phylum of Actinomycetota, like Bifidobacterium), remain to be investigated.
The initial rise and later decline of the genera Blautia, Faecalibacterium, and CAG_352 (belonging to Clostridium) are potentially caused by an initial strong effect of 5-FU, which is counteracted by the prebiotic fibre mixtures effect at later time points. Based on this observation, it might be suggested that clinical interventions with prebiotic fibre mixtures in cancer patients should start before the initiation of 5-FU treatment.
Another interesting finding of our study was that the prebiotic fibre mixtures were able to keep the abundances of Klebsiella and Enterobacter at relatively low levels, in contrast to the rising levels in the conditions without prebiotic fibre mixtures. These genera are both members of the Enterobacteriaceae family with pathogenic properties. This bacterial family is of particular interest in the context of 5-FU treatment, since it might be associated with 5-FU efficacy as well as toxicity. As described in the introduction, E.coli, also an Enterobacteriaceae family member, has been shown to metabolize 5-FU, thereby decreasing its anti-cancer efficacy (16, 19). Consequently, it will be an important question for future research whether Klebsiella and Enterobacter are also involved in 5-FU metabolism and the reduction of treatment efficacy. Furthermore, overgrowth of Enterobacteriaceae is commonly observed in various diseases involving inflammation in the gut, most likely due to a growth advantage under inflammatory circumstances (62). As reviewed by Zeng et al., several mechanisms have been proposed to explain the bloom of Enterobacteriaceae in the inflamed gut, which is also thought to further exacerbate inflammation (62). In the context of anti-cancer treatment, 5-FU-induced dysbiosis might thus be accompanied by a bloom of Enterobacteriaceae, inducing a vicious cycle of increased inflammation, more severe gastrointestinal toxicity, and further Enterobacteriaceae overgrowth. Consequently, the simultaneous stimulation of potentially beneficial bacteria and inhibition of Enterobacteriaceae, as achieved by the prebiotic fibre mixtures under investigation, is considered to represent a promising strategy to counteract the manifestation of 5-FU-induced microbial dysbiosis.
Next to the composition and diversity of the gut microbiota, we also investigated whether 5-FU and prebiotic fibre mixtures had an impact on the production of bacterial metabolites. In this context, SCFA are of special interest, because they exert various beneficial metabolic, anti-carcinogenic as well as anti-inflammatory effects (29, 63, 64). In particular, their anti-inflammatory effects make SCFA interesting molecules that could potentially support the reduction of 5-FU-induced toxicity. BCFA are derived from branched-chain amino acids or ingested via the diet, but their physiological roles are currently not fully understood (30, 65). They are, however, often considered as a proxy for protein fermentation, which leads to the production of several toxic metabolites, including ammonia, p-cresol, indole, and phenol. Analysis of these putrefactive metabolites was not included in the current study, but these should be considered to be measured in subsequent experiments, to investigate the effect of 5-FU on the production of other (toxic) bacterial metabolites.
In our experiments, the administration of prebiotic fibre mixtures in combination with 5-FU considerably increased the production of acetate, which was not surprising because Lactobacillus and Bifidobacterium are known to be potent acetate producers (30, 66). Similarly, the decreased propionate levels were in line with the observed changes in taxa abundance, since propionate is mainly produced by the succinate pathway in Bacteroidota (Bacteroidetes), which were found to diminish during our experiments (67). Next to succinate, lactate is also a precursor of propionate (67). Therefore, it would be of interest to measure lactate levels in future studies, particularly because Bifidobacterium and Lactobacillus are also lactate producers (30).
On the other hand, it was surprising that the prebiotic fibre mixtures intervention did not significantly affect butyrate levels, although M1 and M3 seemed to slightly stimulate butyrate production. This was also unexpected because the stimulated Anaerostipes is known to be a butyrate producer (67). However, it should be noted that butyrate can also be produced from acetate (66, 67). Therefore, it might be possible that the increased acetate levels would also translate into increased butyrate levels if the intervention period would be prolonged. Furthermore, it might be possible that other prominent butyrate-producing bacteria (e.g. Eubacterium, Roseburia, Coprococcus) could not grow well under the conditions in the TIM-2 model.
Both BCFA were increased by 5-FU and decreased by prebiotic fibre mixtures, which is of notable interest in view of our recent observation that iso-butyrate was significantly lower in the faeces of CRC patients who showed partial response, compared to patients with stable or progressive disease during capecitabine treatment (68). Thus, reduction of amino acid degradation and subsequent BCFA production, by administering adequate amounts of prebiotic fibre mixtures, might positively influence tumour response during 5-FU-based treatment and requires further investigation.
This study has its limitations, which arise mainly due to the in vitro design, which mimics but still does not entirely matches a human colon. During the intervention period, 5-FU was injected into the lumen of the model. Consequently, the microbiota were directly exposed to the chemotherapeutic agent, while it reaches the colon more gradually in patients. We took this physiological difference partly into account by choosing a lower dose than the concentration of 5-FU or its metabolites that would be anticipated to be present in vivo (Supplementary Table S1). In this context, it should also be noted that oral capecitabine would be expected to lead to higher chronic colonic exposure, while intravenous 5-FU is expected to induce temporary increases of 5-FU and its metabolites in the colon. In addition, the microbiota used in this study was a pooled microbiota derived from healthy participants. It might be expected that the gut microbiota of cancer patients has a distinct composition and activity. Therefore, ongoing studies also assess the effect of prebiotic fibre mixtures on faecal samples obtained from CRC patients during 5-FU-based chemotherapy. Given the study's limited statistical power due to only two runs per condition, statistical results should be interpreted cautiously. We have taken this into account as much as possible and used the nparLD R package which can be used to analyse longitudinal measurements in the factorial experiment. Nevertheless our pilot experiments provided valuable insights which need to be validated with larger sample sizes. Because the pH was kept constant in the model, the present experiments do not provide information concerning pH changes upon fibre fermentation, which should be addressed in other settings. An additional limitation is that we cannot exclude that conditions in the model might favour growth of specific gut bacteria, while other bacteria might react more sensitively to this environment, as already discussed in the context of butyrate.