Carnivore Diet and Ankylosing Spondylitis: The Klebsiella Hypothesis, the Low-Starch Evidence, and What Hasn't Been Tested

The honest one-sentence version of this article: ankylosing spondylitis has a real, specific gut-bacteria diet hypothesis behind it, the actual human trial data for that hypothesis is small and uncontrolled, the diet that was studied is not the same as carnivore, and no published research has tested carnivore itself for AS. Everything below expands on that sentence without softening it.

This matters because AS content online often blurs three different things together: (1) a genuine, published, mechanistically-grounded low-starch diet research program, (2) the modern carnivore diet, and (3) individual success stories. We're going to keep those three separate the whole way through.

1. The Hypothesis: HLA-B27, Klebsiella, and Why Starch Specifically

Ankylosing spondylitis is a chronic inflammatory arthritis that primarily affects the spine and sacroiliac joints, causing pain, stiffness, and — in more advanced or undertreated cases — new bone formation that can fuse vertebrae together. It is strongly associated with the HLA-B27 gene, though most people who carry HLA-B27 never develop AS, and current mainstream treatment targets the immune pathways involved in the disease, particularly the TNF and IL-23/IL-17 signaling axis.

Separately from that mainstream immunology, a specific and longer-running research program has focused on the gut. Starting in the 1980s, rheumatologist Alan Ebringer and colleagues proposed what is commonly called the Klebsiella hypothesis:

• People with AS who carry HLA-B27 also tend to show elevated antibodies against Klebsiella pneumoniae, a common gut bacterium.
• Ebringer's group reported that certain Klebsiella molecules carry amino acid sequences resembling parts of the HLA-B27 molecule and other joint tissue — a phenomenon called molecular mimicry, where the immune system's response to a microbe cross-reacts with the body's own structures.
• Because Klebsiella can use starch as a growth substrate, the theory concludes that reducing dietary starch may reduce Klebsiella populations in the gut, and in turn reduce the antibody response and disease activity.

This is a coherent, testable mechanistic story, and it is meaningfully more specific than the generic "elimination diet might help autoimmune disease" framing used for most other conditions. It is described across several of Ebringer and colleagues' review papers (Rashid & Ebringer, 2007, PMID 17186116; Rashid, Wilson & Ebringer, 2013, PMID 23781254).

It is also, in current terms, a minority hypothesis, not an established mechanism. The dominant model in rheumatology today centers on HLA-B27 protein misfolding, gut microbiome dysbiosis more broadly (not Klebsiella specifically), and the IL-23/IL-17 immune axis — which is why the newest AS biologics target IL-17 and IL-23, not Klebsiella. The Klebsiella-specific literature has been produced largely by the same research group over several decades and has not become the standard model taught as AS pathophysiology, even though it has not been disproven either. Readers should treat it as a live, minority research thread — plausible, actively studied, but not settled science.

Crucially, the diet this hypothesis actually generated — the "low-starch diet" used in Ebringer's clinic — is not carnivore. It restricted bread, potatoes, pasta, and cakes/confectionery, while still permitting vegetables, salad, fruit, and dairy. A carnivore diet removes starch too (by removing all plant food), so it is mechanistically compatible with the "reduce starch" logic — but it also removes many things the tested diet did not remove (all non-starchy vegetables and fruit, and their fiber, vitamin C, and polyphenols) and is a far more extreme intervention than anything that has actually been studied.

2. The Actual Evidence: A Real Hypothesis, Old Human Data, Zero Carnivore-Specific Studies

The adjacent evidence: Ebringer's low-starch diet studies

Unlike most autoimmune conditions, AS does have real (if old and limited) human trial data for a related, non-carnivore intervention:

• Ebringer A, Wilson C. "The Use of a Low Starch Diet in the Treatment of Patients Suffering from Ankylosing Spondylitis." Clin Rheumatol. 1996;15 Suppl 1:62-66. PMID 8835506; DOI 10.1007/BF03342649. This report describes AS patients placed on a low-starch diet (restricting bread, potatoes, pasta, cakes) for several months, with reported reductions in serum IgA levels and in symptoms/inflammation. Secondary sources describing this work cite a cohort of roughly 36 patients followed for an average of about nine months, and note it was later used more broadly in a London AS clinic; we were not able to independently confirm the exact sample size and statistics from the abstract text alone, so treat that specific figure as reported-elsewhere rather than independently verified by us.
• Rashid T, Ebringer A. "Ankylosing Spondylitis Is Linked to Klebsiella — the Evidence." Clin Rheumatol. 2007;26(6):858-864. PMID 17186116; DOI 10.1007/s10067-006-0488-7. A review of genetic, microbiologic, and serologic evidence for the Klebsiella hypothesis; it explicitly calls for further prospective studies rather than presenting the theory as proven.
• Rashid T, Wilson C, Ebringer A. "The Link between Ankylosing Spondylitis, Crohn's Disease, Klebsiella, and Starch Consumption." Clin Dev Immunol (now Journal of Immunology Research). 2013;2013:872632. PMID 23781254; DOI 10.1155/2013/872632. Another review from the same group restating the starch-Klebsiella-HLA-B27 mechanism and proposing low-starch diets as a complementary (not replacement) measure alongside conventional treatment.

Why this is weak-to-moderate evidence at best, not proof:

• No blinding, no placebo/control arm in the original 1996 report — patients knew what diet they were following and there was no comparison group eating their normal diet over the same period.
• Concentration of authorship. Most of the primary Klebsiella/low-starch literature comes from Ebringer's own research group across several decades, rather than being independently replicated by unrelated labs at scale.
• Age of the primary data. The main clinical report is from 1996 — three decades old, using older disease-activity measures and without modern biologic-era context (most AS patients today are also on TNF or IL-17 inhibitors, which the original cohort was not).
• Surrogate/self-reported outcomes. IgA and symptom reports are informative but are not the same as objective imaging or validated composite disease-activity scores in a controlled design.

A registered randomized trial exists — but we could not find published results

A larger, more rigorous test of this idea was registered: "Effect of a Low Starch Diet in Patients With Ankylosing Spondylitis" (ClinicalTrials.gov NCT04386538), sponsored by Universidade do Porto in collaboration with the Instituto Português de Reumatologia in Lisbon. The design randomized AS patients to either a WHO-recommended balanced diet or an individualized diet reducing starch intake by at least 40%, measuring Klebsiella pneumoniae levels alongside disease activity, functional impairment, and quality of life. Its registry status was last verified in January 2022 (recruiting, estimated completion December 2022), with no update or posted results since — the registry entry appears to have gone stale rather than showing a recent active or completed trial, and we were unable to locate a published peer-reviewed results paper for it. This is a real evidence gap worth flagging rather than papering over.

Separately, we found a 2024 conference abstract (Annals of the Rheumatic Diseases, abstract AB1181-HPR) describing a smaller, different, observational (non-interventional) cross-sectional study from Hospital de Egas Moniz in Lisbon, which measured existing dietary starch intake (via food diaries) against disease-activity scores (BASDAI, BASFI, CRP) in AS patients already on their usual diets, and reported a correlation between higher starch intake and higher disease activity. This is correlational, not a diet intervention — it cannot show that lowering starch caused improvement, only that people already eating less starch tended to have lower disease-activity scores, which could reflect many other factors (overall diet quality, disease severity affecting appetite, etc.). We were only able to review this abstract via search-indexed summaries, as the full text sits behind a publisher paywall we could not access directly — treat our description of it as lower-confidence than the PMID-verified items above.

3. What This Means, Concretely

Putting the pieces together honestly:

• The mechanistic story (Klebsiella molecular mimicry) is genuine, published, peer-reviewed research — not something carnivore marketing invented. But it remains a minority hypothesis within rheumatology, not the consensus model of AS.
• The diet that was actually tested in humans is a moderate low-starch diet, not carnivore. It kept vegetables, fruit, and dairy; carnivore removes all of that.
• The human evidence for even that moderate diet is old, small, and uncontrolled — one 1996 clinical report plus supportive review articles from the same investigators, and a modern randomized trial whose results we could not find published.
• There is no published carnivore-specific study of AS at all — carnivore is a further, untested extrapolation on top of an already-thin evidence base.

That combination — a real mechanism, weak old human data for a related-but-different diet, and zero data for carnivore itself — is a meaningfully different evidence picture than "anecdote only," but it is still far short of "proven treatment." Anyone telling you carnivore is a clinically validated therapy for AS is overstating what has actually been published.

4. Risks and Responsible Framing

AS carries risks that make the stakes of "diet instead of treatment" higher than in many other conditions.

Never stop or change prescribed treatment on your own

NSAIDs, biologics (TNF inhibitors and IL-17 inhibitors), and DMARDs are not optional based on a blog post or online testimonial. Untreated or undertreated axial inflammation in AS can progress to new bone formation and fusion of spinal joints (sometimes described as "bamboo spine"), reduced chest expansion and lung capacity, increased fracture risk in a rigid spine, and involvement of other organs (including the eyes — acute anterior uveitis — and, less commonly, the heart and lungs). These are structural and sometimes irreversible changes. No diet, carnivore or otherwise, has been shown to reverse existing spinal fusion.

Physical therapy and exercise are not optional either

Unlike some autoimmune conditions where diet is the main non-drug lever people ask about, AS management is built around regular postural and mobility exercise alongside medication — this is standard rheumatology practice, not something a diet substitutes for. A dietary experiment is not a reason to skip physical therapy.

Nutritional and practical risks of carnivore specifically

• Micronutrient gaps. Removing all plants removes vitamin C, most fiber, and typical sources of folate, vitamin K1, and polyphenols — a stricter removal than the low-starch diet that was actually studied. (See our nutrient-deficiency guide.)
• Overshooting the tested intervention. If the mechanism is "less starch feeds less Klebsiella," a moderate reduction (as studied) and a total elimination (carnivore) are not proven to be equivalent — more restrictive is not automatically better, and has not been tested.
• Lipid changes. Broader carnivore-diet survey data shows meaningfully elevated LDL cholesterol in many adherents, which is relevant to cardiovascular risk monitoring regardless of joint symptoms.
• Masking, not treating. Feeling less stiff or less fatigued is not the same as imaging-confirmed inflammation resolving. AS can progress structurally with relatively modest symptoms in some people.

When a supervised dietary trial might be reasonable to discuss

A dietary trial is a conversation to have with your rheumatologist, not a decision to make alone. It may be worth raising if all of the following hold:

If you cannot meet those conditions, the responsible answer is to wait and keep working with your rheumatology team.

5. What You Could Track Yourself, If You Test This

Nothing below is a claim that tracking will make you feel better. It is a way to find out, in your own case, whether anything actually changed — instead of relying on a vague impression of "feeling better," which is easy to misjudge in either direction when a disease naturally waxes and wanes.

What this is not: a promise that any of these numbers will move in a good direction. Recording your BASDAI before and after a diet change tells you what happened in your specific case — it does not by itself tell you the diet caused it (AS activity fluctuates on its own, and a placebo or attention effect from any structured intervention is well documented), and it does not replace the imaging and lab monitoring your rheumatologist already performs. The only purpose of tracking is to replace "I think I feel better" with a number you and your clinician can look at together and decide what, if anything, it means.

How CarnivOS Fits (and Where It Stops)

CarnivOS is a tracking tool, not a treatment. If you and your rheumatologist decide to run a structured, supervised dietary trial, the app can log your food, symptoms, and stiffness/fatigue patterns over the trial window so the trend data is ready for your clinical appointment. That is the entire scope: the app organizes data; your rheumatologist interprets it and makes the medical decisions. CarnivOS does not diagnose, treat, or claim to improve ankylosing spondylitis.

Sources

1. Ebringer A, Wilson C. The Use of a Low Starch Diet in the Treatment of Patients Suffering from Ankylosing Spondylitis. Clin Rheumatol. 1996;15 Suppl 1:62-66. PMID 8835506; DOI 10.1007/BF03342649. (Uncontrolled clinical report on a low-starch — not carnivore — diet; reported reductions in serum IgA and symptoms/inflammation. Exact sample size not independently confirmed by us from the abstract alone; secondary sources describe ~36 patients over ~9 months.)
2. Rashid T, Ebringer A. Ankylosing Spondylitis Is Linked to Klebsiella — the Evidence. Clin Rheumatol. 2007;26(6):858-864. PMID 17186116; DOI 10.1007/s10067-006-0488-7. (Review of genetic/microbiologic/serologic evidence for the Klebsiella hypothesis; calls for further prospective study.)
3. Rashid T, Wilson C, Ebringer A. The Link between Ankylosing Spondylitis, Crohn's Disease, Klebsiella, and Starch Consumption. Clin Dev Immunol (Journal of Immunology Research). 2013;2013:872632. PMID 23781254; DOI 10.1155/2013/872632. (Review restating the starch-Klebsiella-HLA-B27 mechanism; proposes low-starch diet as complementary to conventional treatment, not a replacement.)
4. ClinicalTrials.gov. Effect of a Low Starch Diet in Patients With Ankylosing Spondylitis. NCT04386538. Sponsor: Universidade do Porto / Instituto Português de Reumatologia, Lisbon. (Registered randomized trial of low-starch vs. WHO-recommended diet; registry status last verified January 2022 with no update or posted results since; we could not locate a published peer-reviewed results paper as of this writing.)
5. Annals of the Rheumatic Diseases (EULAR abstract). AB1181-HPR: Starch Intake and Parameters of Disease Activity, Functional Impact and Quality of Life of Patients with Ankylosing Spondylitis. 2024. (Observational cross-sectional cohort correlating existing dietary starch intake with BASDAI/BASFI/CRP; not a diet intervention. Reviewed via search-indexed summary only — full text was paywalled, so treat as lower-confidence than the PMID items above.)
6. Lennerz BS, Mey JT, Henn OH, Ludwig DS. Behavioral Characteristics and Self-Reported Health Status among 2029 Adults Consuming a "Carnivore Diet." Curr Dev Nutr. 2021. PMID 34934897; DOI 10.1093/cdn/nzab133. (Self-reported survey; reports an aggregate "autoimmune disease" category. We could not directly confirm whether ankylosing spondylitis is broken out as its own line item due to paywall access limits — flagged as unverified.)
7. Klement RJ, Matzat JS. Subjective Experiences and Blood Parameter Changes in Individuals From Germany Following a Self-Conceived "Carnivore Diet": An Explorative Study. Cureus. 2025;17:e82521. PMID 40385902; DOI 10.7759/cureus.82521. (n=24 explorative survey; does not mention ankylosing spondylitis.)

Frequently Asked Questions