Does diet beat FMT for antibiotic recovery?

Breaking down the results of a recent study

Broad-spectrum antibiotics act like wildfire in your inner ecosystem, wiping out beneficial bacteria and leaving the gut susceptible to less-desirable opportunists. The functional medicine community has long believed that re-seeding the gut with probiotics might be a good solution for post-antibiotic recovery.

Then, a 2018 study brought some new information to the table. Suez et al. found that while a commercial probiotic delayed microbiome recovery, re-inoculating your gut with your own stool (aFMT, i.e. autologous fecal transplant) could restore the diversity and function of the human gut in a single day. (See my full writeup of the study here.)

Now, a new animal study by Kennedy et al. shows that providing the microbes via FMT might not be sufficient if you don’t also have the right diet in place, and that diet itself is a more powerful driver of microbiome recovery. So, what gives? Should we write off post-antibiotic aFMT altogether? In this article, I’ll break down the results of this latest study and discuss the implications.

The study design

Kennedy et al. split mice into two groups consuming different diets:

1. “Regular chow” — low-fat, low-sugar, high-fiber

2. “Western diet” — high-fat, high-sugar, low-fiber

After four weeks on the diets, the mice received three days of broad-spectrum antibiotics. The researchers then tracked microbiome recovery with regular fecal collections as the mice continued to consume their respective diets. They also tested how the mice responded to gut infection.

The results: diet drives microbiota recovery

The differences were pretty remarkable.

Regular chow-fed mice:

• Recovered microbial biomass (CFU counts) by day 4 post-antibiotics

• Began to recover microbial diversity by day 5, and recovered more than half of their initial diversity by day 11.

• Followed the expected ecological stages of succession, with early dominance of facultative anaerobes quickly followed by increasing diversification of obligate anaerobes, likely driven by complex carbohydrate availability.

• Recovered their overall pre-antibiotic microbial community composition and production of short-chain fatty acids (acetate, butyrate, and propionate) by day 14.

• Were less susceptible to gut infection by Salmonella.

Western diet-fed mice:

• Did not recover microbial biomass (CFU counts) until day 7 post-antibiotics

• Had severely diminished microbial diversity up to at least day 28, and up to 9 weeks post-antibiotics.

• Had delayed ecological succession patterns that didn’t begin until around day 14, with early dominance of the community by Streptococcaceae fueled by the availability of simple sugars.

• Only some of the mice had begun to approach their baseline microbial community structure by day 28. Short-chain fatty acid production was depleted at day 14 and remained low until day 28.

• Experienced “variable but occasionally severe gut inflammation”, and were more susceptible to gut infection by Salmonella.

The authors summed it up as follows:

“Across all taxonomic and functional metrics that we evaluated, mice on WD [Western Diet] experienced more severe ecosystem collapse with slower and less complete recovery than mice on RC [Regular chow].”

These findings were similar to previous studies that have shown impaired microbiome recovery in mice with a low-fiber, high-simple sugar diet.

Where does FMT fit in?

The researchers next looked at whether microbial exposures, via fecal microbiota transplant (FMT), could assist in microbiome recovery. They performed all manner of combinations, with some mice switching to the opposite diet post-antibiotics, some mice staying on the same diet but receiving FMTs from the opposite diet group, and relevant controls. The FMTs were administered at 24 hours post-antibiotics and again at day 14.

The research group reported that, regardless of which diet the mice consumed before antibiotics or which microbes they were exposed to after, gut microbiota recovery was largely driven by post-antibiotic diet.

Even mice that were given a healthy shot of RC microbes during the recovery period fared poorly if they consumed a Western diet post-antibiotics — they had lower microbial diversity and delayed community recovery.

Metabolic modeling provided the mechanism: the regular, fiber-rich, low-sugar chow restored the complex cross-feeding loops that characterize a healthy microbiome. Meanwhile, the Western diet starved them, letting simple sugar-loving opportunistic microbes monopolize the niche.

The authors wrote:

“These experiments indicate that an appropriate diet is both necessary and sufficient for rapid and robust gut microbiota recovery after antibiotic treatment, whereas microbial transplant is neither.”

But is that last bit really true? The global statistical model (in Supplementary Table 9) shows that at Day 14 of recovery, post-antibiotic diet accounts for 15% of total variance in beta-diversity, FMT explains 7% of the variance, and the diet x FMT interaction effect explains another 4%. All of which were statistically significant.

So yes, the benefits of FMT are overshadowed by the effects of diet here, at least at day 14. But FMT still explains a significant percentage of the variance.

The more relevant clinical question is, if we already assume we’re eating a healthy diet, does FMT confer any additional benefit?

Let’s look at the alpha diversity data. Here you can see ASV richness, which loosely translates as the number of observed species in a sample.

Figure 4c. ASV richness across treatment groups at day 14. [Orange significance bars are my own, added based on statistical analysis performed by the study authors and reported in Supplementary Table 9 of the original paper]

The two groups on the far left are the no-antibiotic controls (PBS = saline control), whereas all of the groups on the right received antibiotics plus various interventions. The authors chose not to annotate the significance on Figure 4c, but I’ve added it in orange so you can see the differences in the most relevant groups.

• On the regular chow diet, you can see that antibiotics collapsed richness from a median of ~98 ASVs (RC-PBS, mean ~89) to ~42 ASVs (RC-RC_D-PBS, mean ~32), a highly significant ~56-ASV drop (q <0.001).

• Adding a matched FMT rebounded richness to a median of ~92 ASVs (RC-RC_D-RC_M, mean ~97), indistinguishable from no-antibiotic controls (q = 0.58) and significantly higher than diet-only recovery (RC-RC_D-PBS, Δ ≈ 65 ASVs, q < 0.001).

In other words, regular chow-fed mice had a significant drop in alpha-diversity (two-thirds depleted), which was rescued by an FMT from mice that had consumed a regular chow diet.

Compositionally, the group that received FMT was also slightly closer to the no-antibiotic controls than the group that received diet alone at Day 14 (Bray-Curtis ANOVA data, Supplementary Table 9), though the difference was not statistically significant. This suggests that the gut was repopulated, but still reshuffling to see which members would ultimately stick (as might be expected with heterologous FMT).

A few other limitations make it hard to say that diet >> FMT, and FMT is not necessary or beneficial:

1. Timing: The authors collected fecal samples at day 7 of microbiome recovery, yet report no data from this timepoint. Might FMT have improved composition earlier, in week 1?

2. Function: The authors did not measure short-chain fatty acids, pathogen resistance, or mucosal cytokines in groups that received the FMT, so we don’t know how much the significant richness bump and modest numerical compositional improvement seen at 14 days translates into functional benefits. However, across dozens of antibiotic studies, higher post-antibiotic richness correlates with faster return of butyrate producers, secondary bile acids, and resistance to infectious pathogens.

3. Sample size: There were only six mice per treatment arm, making it difficult to see a strong effect of FMT.

4. Type of FMT. They used heterologous FMTs, pooling stool from all of the mice in the RC group, and using that as their transplant. Autologous FMTs, in contrast, are much more likely to engraft quickly and match pre-antibiotic composition.

5. Fiber equivalence. The diets chosen were pretty extreme in their fiber content. More on that below.

Taken together, it does appear that diet plays a larger role than heterologous FMT at day 14. However, a matched-diet FMT buys you an extra third of the journey back towards baseline diversity by day 14, possibly sooner, with likely functional benefits.

Does this translate to humans?

Given that the early predominance of facultative anaerobes also occurs in humans post-antibiotics, the authors said they expect that these conclusions about the strong effect of diet on post-antibiotic recovery will also hold true in humans.

However, it’s interesting to note that the 2018 study by Suez et al. showed rapid recovery of the human microbiome after antibiotics with autologous FMT despite zero dietary intervention. Much to my amazement, there was not a single reference to this high-impact paper in the discussion of the Kennedy et al. study (how this got through peer-review I’ll never understand), but I’ll offer my own two cents.

While Suez et al. did not report dietary data, I think we can reasonably assume that the participants’ fiber intake was similar to the typical Israeli intake (median 15-16 grams of fiber per day). Perhaps this was “enough” substrate for the incoming microbes to work with. It’s also possible that autologous transplants (derived from one’s own microbiome) may need less dietary support because the community is already matched to the host and their typical diet, whereas heterologous grafts (derived from other hosts, or pooled from many hosts, like in the Kennedy et al. study) may be more diet-sensitive.

The human microbiome may also just be more amenable to aFMT-supported recovery. Suez et al. found that aFMT-treated humans returned to their pre-antibiotic community composition within a single day, whereas mice given aFMT in their study did not return to baseline composition until day 28 (though this was still far better than mice in the spontaneous recovery group).

So, at least in humans, aFMT seems to be very effective, regardless of dietary intake.

It’s also important to note that the “regular chow” diet in the Kennedy et al. study represents an extreme, equivalent to 65 grams of fiber per day in humans (more than twice the recommended dietary guidelines in the U.S.) and might more accurately have been labeled a “very high-fiber diet”. The mouse “Western diet” was equivalent to 12 grams of fiber/day, slightly below the average intake in Western human populations. So, the picture of diet >> FMT may have been partly due to the diet extremes (particularly the high fiber diet) chosen for this study.

I say all of this realizing that aFMT is not common practice, and not currently feasible for most people. Safe storage of FMT material requires access to a -80 degree Celsius laboratory-grade freezer — something most people simply do not have. And I’m all for tuning up your diet for 2-3 weeks post-antibiotics, a very low-risk and potentially high-reward proposition. All I’m saying is, we shouldn’t be writing off aFMT, and I think the data are strong enough that we should be doing a lot more research on this AND working towards making them more widely available.

Dr. Eugene Chang, senior author on the Kennedy et al. study and professor at the University of Chicago, says a feasibility study to explore post-antibiotic diet vs. FMT influences in humans is “on the drawing board”, though no study protocol is registered yet.

If you do have the opportunity for an autologous FMT post-antibiotics, I certainly wouldn’t pass it up.

Things I will still keep in the toolkit

Other adjunct strategies can help support gut ecosystem recovery after antibiotics. In addition to diet (and aFMT), specific probiotics and gut metabolites have been shown to help support optimal gut and microbiome recovery. I cover these in detail in my guide to antibiotic recovery.

I’m hopeful that future research will bolster the emerging evidence that these are likely to be powerful supports for microbiome recovery, in lieu of aFMT becoming more widely available.

What if you already have dysbiosis?

So far, we’ve been assuming that we’re talking about someone with a healthy baseline microbiome. But what about someone who was already struggling before antibiotics and can’t tolerate complex carbohydrates? Could antibiotics open up a window in which to reintroduce these complex carbs and fibers, and reshape or diversify the microbiome? Or is a gut that was already struggling unlikely to be able to handle the extra fiber load, leading to abnormal fermentation, inflammation, and worsened dysbiosis?

The research here is still limited. In practice, I’d typically suggest trying to eat closer to what your optimal, more varied, diet might be (perhaps increasing fiber gradually in 3-5 gram/day increments), but resisting the urge to jump up 20 grams of fermentable fiber overnight. Diet is a powerful driver of recovery, but individual tolerance likely matters, too.

Could diet influence FMT efficacy?

The Kennedy et al. study may also shed light on why heterologous FMT (using screened stool from another person) often yields mixed results in conditions like ulcerative colitis, irritable bowel syndrome, or autism. Without the right gut environment (including but not limited to dietary substrates), incoming microbes may struggle to establish themselves. On the other hand, a period of intentional diet & gut environment optimization before FMT may increase the likelihood of successful engraftment and functional recovery.

If you haven’t already, be sure to check out my full guide to antibiotic recovery or read more about supporting the gut environment.