The effect of a high-fiber diet on my microbiome
Recently, I listened to an excellent podcast episode of the quantified body where Damien Blenkinsopp discusses the various microbiome testing options with well-known self-experimenter Richard Sprague. It made me think a lot about my own approach to learning more about the microbiome. If you’re even remotely thinking about getting your microbiome tested, I highly recommend listening to it.
I also recently took two uBiome gut microbiome tests. Once I got my results, just like most people, I was left scratching my head with the long list of unfamiliar bacterial groups they list out. Before I took the tests, I was really only familiar with two beneficial bacterial genera (i.e. bacterial groups at the “genus” level) — Lactobacillus and Bifidobacterium — which are commonly found in most probiotics supplements. It turns out that when it comes to colonizing the human gut, these two genera are very small players. There’s a whole host of other “colonizers” that make a home for themselves in the gut, some of which provide numerous benefits. So, this led to some natural questions: Am I feeding my gut microbiome the right way, and do my probiotic supplements even work?
I’ve been following blog posts by Richard Sprague for a while now. However, I was left with more questions than answers. There are a host of confusing (and conflicting) studies on the effect of specific microbial genera on human health. Trying to correlate a particular genus or species with any one condition is a lost cause. The microbiome field is changing so rapidly, it seems that almost every other day, a new study comes out that either discovers a new trend or refutes a prior claim.
From an individual standpoint, I’m committed to learn more about what kinds of foods feed specific known beneficial organisms. I want to know how eating certain kinds of foods (especially fiber) influence specific groups my gut microbiome, and how that translates to an overall feeling of better health. In this post, I’ll summarize some of my findings and cite some of my references.
“Good” vs. “Bad”
First of all, I want to stress that I’m not looking for “good” or “bad” species in my microbiome. Based on their interaction with humans, bacteria can be grouped as follows:
- Commensal: These are the organisms that live within us, but do not cause us any harm or benefit
- Mutualistic: These organisms benefit us while they themselves benefit
- Parasitic: These are the “pathogenic” organisms that cause us harm
In addition to these groups, there are also “opportunistic pathogens”, wherein a particular species that was commensal becomes pathogenic due to perturbations in their host (age, disease, wounds, medication and immunodeficiency).
It’s clear that there is no such thing as an “optimum” gut microbiome, since human beings are so diverse and diets are so varied. This is where individual microbiome tests can tell us a lot about ourselves and our daily choices.
I’ve always had a relatively high fiber diet, with plenty of fruits, vegetables, lentils and whole grains. I’ve recently been reading studies that look into how different kinds of fiber influence specific bacteria in our guts. I was delighted to find evidence that my dietary habits are well within acceptable logic, and it gave me a good sense of direction for what I plan to eat in future.
Before each test, I made a note of what I was eating for couple of weeks before taking my sample. In general, my diet was heavily plant-based, with plenty of vegetables, whole-grains and lentils. I ate limited amounts of animal protein (mostly eggs and grass-fed dairy), and limited fruit. My primary methods of cooking were sautéing, baking, steaming and broiling. My primary sources of fat were saturated and mono-unsaturated fats (butter, ghee or clarified butter, olive oil, avocados). I only snacked on nuts and dried fruits (if at all). I also consumed fermented foods (small-batch yogurt, kefir and kimchi, all store-bought but hopefully with some live organisms in them). I completely eliminated processed foods, processed sugar and fried foods from my diet. Oh, and I didn’t take any medication or antibiotics throughout the last year.
The major difference between my samples taken on January 23rd and March 3rd was that before my first sample, I had been eating a higher proportion of whole grains (whole wheat bread and parboiled rice) and lentils (red/yellow lentils, chick peas, kidney beans) and a relatively lower proportion of vegetables. Before my second sample, I had eaten way more green vegetables (spinach, kale, broccoli, asparagus), a higher amount of vegetable fiber overall (sweet potatoes, parsnips, green onions), and far lower amounts of fat and whole grains.
- In January, I ate a diet containing larger amounts of whole grains and lentils, with some vegetables (around 5–8 kinds) the few days before taking my sample.
- In March, I ate much larger amounts of vegetables, including tubers, cruciferous and green vegetables, and a much larger variety of vegetables (more than 15 kinds) in the few days before taking my sample.
In both cases, I was consuming roughly the same, limited amount of healthy fats and avoiding processed sugar for at least a week before taking my sample.
My Microbiome Analysis (Gut)
I started off by reading Richard Sprague’s detailed instructions on how to begin working with uBiome raw data. First, I used a fork of Richard’s
ubiome-opensource/microbiome-tools repo to plot my diversity (at the family level)using Simpson’s index. The Python code and raw data available here.
While it looks like my diversity decreased between my first and second samples, it’s important to note that the diversity can vary greatly over the course of even one day. Not too much can be read into the decreasing trend with just two samples. I’d need many more samples to see the overall range over which my diversity varies with time.
However, I did notice that according to uBiome’s website, my diversity was on the lower end when compared to their customer data in my category. For my first test, I was in the 32nd percentile, and my second test showed an even lower diversity, just in the 16th percentile! That’s definitely lower than I’d like, even if the phenomenon is simply transient.
I do have some guesses as to why I think my diversity is on the lower end of uBiome’s data spectrum:
- I have an indoor desk job, and my life mostly revolves around my home and my office.
- It’s winter in the northern US where I live, and I don’t get much time outdoors amidst nature.
- Most of my meals are made at home (same kitchen, similar food groups).
- I live on my own and have no pets, so this minimizes any exchange of microbes.
- I normally repeat meals through the week (leftovers, meal prep for weekdays, etc.).
So, it kind of makes sense that I see low diversity in my samples. I’d definitely like to do another test in the summer, when I’ve had the chance to spend more time outdoors on the grass and exposed to the elements. It’ll be interesting to see if that helps.
Logically, I might conclude that if I had some sort of catastrophic loss of beneficial microbes (due to antibiotics, a stomach bug, ill health or a poor diet), a lower diversity would make me much more susceptible to “colonization” by pathogenic species. It has been well documented that a more diverse microbiome is more generally associated with good health, but there is also a high degree of variability of diversity, even within healthy subjects. So, for now, I’ll just be watching my diet and looking for any warning signs with regard to my health.
Results: Phylum level
To plot individual results and a comparison of my two results, I wrote some Python code as per the 2018 uBiome JSON file format. All my code and raw input files are on my GitHub repo.
The phylum level is the topmost classification of bacterial groups. Here’s a look at my initial (January) sample’s breakdown at the phylum level.
It looks like my entire gut bacterial population is made up of just four phyla — Firmicutes, Bacteroidetes, Actinobacteria and Proteobacteria. Almost 80% of those are just Firmicutes!
Here’s a summary of what I looked for:
- Firmicutes: This is too broad a grouping to gain any insights, but this group contains many of the “gram-positive” beneficial species.
- Bacteroidetes: This is also a really broad group with many different interacting species, which are also mostly beneficial.
- Actinobacteria: This is the group that contains the well-known Bifidobacteria genus, a beneficial group.
- Proteobacteria: While not all Proteobacteria are pathogenic, a number of them, especially those in the “Enterobacteriaceae” family are pathogenic. Many people with inflammatory bowel disease show increased levels of Proteobacteria, and numerous studies have been done on this particular phylum linking it to diseased states.
At the phylum level, it’s preferable that the ratio of Actinobacteria to Proteobacteria stays as high as possible (to minimize the possibility of contracting pathogens). Other than that, it’s hard to draw any more useful conclusions from this level.
Outside of the phyla identified in my sample, the following phyla are also commonly found in uBiome samples:
- Verrucomicrobia: This contains the genus Akkermansia that is commonly cited in studies as beneficial to help reduce obesity.
- Fusobacteria: This has been associated with carcinogenesis in the colon
- Spirochaetes: This phylum has known pathogens, that cause Syphilis and Lyme disease.
Of course, just like Proteobacteria, there are potentially harmless or beneficial organisms in the Fusobacteria and Spirochaetes phyla as well. New research is always coming out, and there is no one “good” or “bad” group. There are still other phyla and even archaea (evolutionary cousins of bacteria) that are found in the gut, but at present, too little is known about their effects on human health.
Below, I compare the change between my results from January and March using a horizontal bar chart.
In the above plot, a positive (rightward) bar means that my new sample shows a higher relative percentage than the old sample, and vice versa. The Proteobacteria levels are almost non-existent. There’s a large shift in the Firmicutes to Bacteroidetes ratio, but again, there are so many conflicting studies on the significance of this that I’m choosing to ignore it, at least at the phylum level. It is clear from the above plot that my Proteobacteria levels remain very low (less than 1%) between January and March, which is desirable. Another obvious observation is that my Actinobacteria levels increased by a couple percent — so I’ve been able to push up my Actinobacteria to Proteobacteria ratio by a small amount!
No new phyla showed up in my March sample. Considering my stable living conditions and diet and the fact that I haven’t traveled anywhere, this makes sense.
Results: Genus level
I’ll jump straight down to the genus level (ignoring order, class and family) for ease of comparison and to avoid making too long a post. At this level, some interesting trends and groups are visible.
For clarity, I only plotted the top 20 genera in my January sample. What immediately stands out is that almost 80% of my sample is made up of just the top six genera!
- Faecalibacterium: This is a very common genus in healthy people, and is one of the primary “butyrate” producers in the gut. Butyrate is a short-chain fatty acid (SCFA) that is highly beneficial in maintaining a healthy gut mucosal lining, and leads to “intestinal homeostasis”, which correlates strongly with good health. In my case, they made up a large proportion (more than 20%!) of all genera in my sample.
- Bacteroides: This genus (not to be confused with the phylum Bacteroidetes) helps break down vegetable/whole grain fibers, and is common in people with lower fat diets.
- Roseburia: Just like Faecalibacterium, this is a butyrate producer, which is hugely beneficial for preventing intestinal dysbiosis. An interesting 2017 uBiome paper also links this genus with reduced likelihood of type-2 diabetes, IBS and inflammatory bowel disease.
- Blautia: This genus is a strong indication of a healthy gut — just like Bacteroides, they help digest complex carbohydrates.
- Megamonas: This is a mysterious genus. I’ve not seen too many other articles or papers discussing this, and it’s definitely not as widely studied as the other common beneficial genera. Considering I seem to harbor a significant proportion of this microbe (almost 10%), I’d be interested to see what effect this has on health or metabolism.
- Fusicatenibacter: Richard Sprague noticed a sharp spike in the levels of this genus when he began consuming kefir. His results are published here. Could it be that this bacterium finds a niche in the guts of people who harbor other beneficial species and who consume kefir regularly? It’s unclear what exactly the kefir has to do with it — it might just be that Fusicatenibacter thrives on the byproducts of other species that coexist with it, and kefir somehow provides it with that environment. I showed as much as 6% of this one genus in my January sample, and I consume a lot of kefir (at least once a day), so it makes for an interesting observation!
Most of the remaining genera, present in smaller proportions, appear to be relatively benign and not associated with disease, as far as I can tell. Bifidobacteria are way down at <1%, meaning that they are most likely out-competed by the other fiber-degrading groups. I was a little surprised at this, considering I take probiotic supplements with Bifidobacteria, and consume kefir and yogurt on a regular basis. It looks like diet plays a larger role in deciding how much Bifidobacteria can take hold.
Next, I compared my March sample with the January sample at the genus level. The below plot shows only those genera that were common across both samples, and tracks the percent change (positive or negative).
The first thing that stands out is the whopping increase in the genus Bacteroides — almost 25%! With this increase, there is a correspondingly significant decrease in Roseburia, Megamonas and Fusicatenibacter. There is also a noticeable decrease in Faecalibacterium and Lachnospira. It looks like by consuming a lot more vegetable fiber and reducing my fat and whole-grain intake, my gut (at least temporarily) shifted heavily towards the Bacteroides genus. Also, interestingly, my Bifidobacterium levels increased. By not only increasing the amount of vegetable fiber, but also the variety of vegetables I ate, there was a noticeable increase in the levels of Bifidobacteria!
Granted, I only have two samples to make conclusions from, but I fully intend to do another test or two to see for myself if I can maintain higher Bifidobacteria levels over time using a dietary approach.
Each of the top three (and corresponding bottom three) trends align well with the dietary changes I made in the few days before taking my samples. By consuming much more vegetable fiber and beans while reducing lentils and whole-grains, I “traded” some of my butyrate producers (Faecalibacterium, Roseburia, Lachnospira) for the fiber degraders (Bacteroides, Blautia, Bifidobacteria). Along the way, my Megamonas and Fusicatenibacter levels also dropped as collateral damage. Interestingly, I hadn’t consumed kefir the night before my second sample, so I’m not sure if that alone might have played a role in reducing my Fusicatenibacter.
It is clear that the different beneficial genera are competing for the same resources in the same limited space. Depending on what we eat and what they are fermenting, it appears that their levels will greatly vary over time. This study describes some research regarding the different ways fiber in the gut is utilized by Bacteroides.
I’m going to be tracking the two less common genera — Megamonas and Fusicatenibacter over time. I don’t typically see much of these two genera in the online chatter regarding uBiome results. I’m curious as to what role they play, and what dietary changes affect them specifically. Also, I want to see if I can maintain Bifidobacteria levels over time by consuming a varied whole foods diet and a large variety of vegetables and legumes.
Along with comparing the common genera, I also plotted which genera were unique to each sample.
At first glance, it looks like I dropped a number of genera between January and March (which makes sense considering my diversity “dropped”). However, the relative percentages are really small, so it’s just possible that my stool sample (being localized and not “homogenized”) might not have captured all the groups that are present in trace amounts.
However, it looks like many of the Enterobacteriaceae family (possible pathogens) like Kluyvera, and other Proteobacteria (Campylobacter, Neisseria, Haemophilus and Actinobacillus) are not present in my March sample, which seems like a good thing. Consistently maintaining low levels of Proteobacteria (especially Enterobacteriaceae) is a sign that the high fiber diet I’m following is allowing for the beneficial microbes to thrive and minimizes the risk of pathogenic species taking hold.
Another interesting observation is that the common genus Prevotella, which is associated with carbohydrate-rich diets — already scant in my January sample (<0.1%) — completely disappeared in my March sample. This points to my relatively high vegetable intake and lower intake of grains, which may have allowed the vegetable fiber degraders like Bacteroides to dominate.
Results: Species level
Although uBiome does not publish the species information on their website, they do provide the information in their raw data. I thought it would be interesting to see which specific Bacteroides species increased so much in my March sample.
What’s interesting is that one particular species, Bacteroides Plebeius, shows the biggest increase with the vegetable-rich diet. While I take this result with a grain of salt, it’s still interesting to read about this species and what it’s commonly associated with. There are a number of studies stating that B. Plebeius is present in the guts of Japanese people and that this species genetically adapted itself to digesting seaweed. Richard Sprague noticed relatively higher levels (almost 16%) of this species in his samples back in 2015. Why do I show such high levels of this species considering I’m from South Asia, and have never lived in Japan? Sure, I visited Japan four years ago, and I ate a Korean meal the night before I gave my sample in March (that was rich in nori, kimchi and green vegetables), but I’d be surprised if this one species managed to take hold with such limited exposure. Could it be that my diet rich in green vegetables led to a temporary spike in a particular strain of B. Plebeius? Could it also be that those studies that link it with Japanese guts are outdated, and that there are other world populations that show this species in high numbers as well? In reality, only one particular strain of B. Plebeius is likely to have picked up the mutation that allows it to break down seaweed. It’s highly possible that the strain I (may or may not) have in my gut is a different one; one that is tuned to break down tough fibers in green vegetables. Overall, it was interesting to note this particular species, and I’ll be watching for it in my future samples.
After spending the better part of the last two months educating myself about the wonders of the gut microbiome, I’ve come out of the experience much more conscious of the power that our microbes hold over us. I noted the key players in my microbiome, and drew patterns regarding which genera are most influenced by my specific diet. Every time I buy vegetables in the supermarket, I am that much more cognizant of the kinds of fibers that I need to consume, because at the end of the day, some amount of indigestible fibers need to reach my gut.
Although I only took two uBiome tests so far, I fully intend to take more tests keep a record of my gut microbiome over time. Based on my limited tests, I came to following conclusions:
- I have a healthy proportion of butyrate producers and fiber-degrading bacteria and very low level of pathogenic (or disease-associated) bacteria in my gut.
- I have very low, or almost non-existent levels of Prevotella, which means my vegetable-eating habit has more of an impact on my microbiome than my grain-based carbohydrate consumption.
- Because most of my protein is from plants, I harbor non-existent levels of sulfur-degrading bacteria (like Desulfurovibrio) and have lower risks of inflammatory bowel disease.
- My probiotic foods and supplements don’t seem to directly influence any particular species, including the Bifidobacteria in my gut. However, this isn’t to say that probiotics are useless; it’s highly possible that they might be indirectly having a beneficial effect, such as influencing the immune system of setting up a more favorable environment for other beneficial organisms to thrive.
- The low diversity, while not an immediate concern, is something I want to look at as time goes on. It will be interesting to see how it changes with seasons, and if I’m living with other people or spending more time outdoors.
Having a high-fiber diet with plenty of vegetables and legumes — with limited amounts of whole grains, fermented foods and healthy fats, and no processed sugar — could possibly be the reason for my overall well-being of late. I have no “aches and pains”, healthy skin and hair, and I’ve been sleeping much better than I used to. I’ve also noticed an improvement in my immunity — I’ve only gotten one or two mild colds (recovering in a day or two) over the past year, with no other health problems.
While a high-fiber diet may not work for everyone, a few quick microbiome tests, some rethinking of dietary choices and bit of research online can lead to some life-changing decisions!
Along the way, I came across some golden material, which I felt put me on the right track. I highly recommend reading the work of the following people — I know I’ll be reading/listening to their material and making my own logical conclusions with any new microbiome research that comes out.