Recurrent SIBO is not a treatment failure. It is the equilibrium state of a lesion that the standard treatment model does not address.

Rifaximin, herbal antimicrobials, and elemental diets reduce the microbial overgrowth. They do not restore the host substrate that determines which microbes can dominate the gut in the first place. When treatment ends, the substrate failure remains. The conditions that selected for the overgrowth select for it again. The microbes return to the niche the host is producing for them. This is the central pattern of recurrent SIBO, and it is the reason most patients cycle through three to six rounds of treatment before they stop trying.

The clinical pattern that defines recurrent SIBO

A patient presents with bloating, postprandial fullness, food sensitivities, and a positive breath test. They are prescribed rifaximin, sometimes with neomycin if methane is dominant. They feel substantially better within two to four weeks. Their breath test normalizes. Three to six months later, the symptoms return. They retreat. The same pattern repeats. By round four or five, the response to treatment is shorter and less complete than it was the first time. By round six, the patient is often more symptomatic than they were before the first treatment.

This pattern is so consistent across SIBO patients that it is essentially diagnostic. The clinical literature documents recurrence rates of 30 to 60 percent within nine months of rifaximin treatment, and higher in hydrogen-sulfide-dominant cases. Anecdotal experience among clinicians working with this patient population places the real number higher, because many patients do not return to the prescribing clinician once it becomes clear the treatment cycle is not converging.

The standard interpretation of this pattern is that recurrence reflects insufficient motility, anatomical predisposition, antimicrobial resistance, or inadequate treatment duration. Prokinetics are added. Treatment durations are extended. Antimicrobial classes are cycled. The pattern continues.

The alternative interpretation, the one this article will develop, is that recurrence is not a story about the microbes at all. It is a story about a host substrate that has lost the capacity to maintain the conditions that defined its own healthy microbial community. Until that substrate is addressed, the microbes will reorganize around whatever equilibrium the compromised host is producing. Antimicrobial intervention treats one half of the feedback loop. The other half stays running. Recurrence is the predictable result.

Why the standard model predicts what actually happens

The standard SIBO treatment model rests on a specific causal claim. The disease is the microbial overgrowth. The lesion is in the small intestine. The cure is microbial reduction. If this account is correct, antimicrobial treatment should produce durable remission, and recurrence should be the exception rather than the rule.

The data do not support this prediction. Recurrence is the rule. The model that produces a prediction inconsistent with the data is the model that needs to be re-examined.

The dysbiosis-first paradigm has had decades and tens of thousands of treated patients to validate itself. It has not. This is not a failure of clinical competence. It is a failure of the framework. Clinicians applying the dysbiosis-first model are not making mistakes in execution. They are working from a model that misidentifies the lesion. Better execution of the wrong model produces the same result faster.

The Host Capacity Model offers a different account. It begins from a different unit of analysis, makes different predictions, and points to a different set of interventions.

What the Host Capacity Model says is actually happening

The Host Capacity Model (HCM) is a systems-biology framework that reframes recurrent SIBO, and the broader pattern of chronic gut dysfunction, as downstream of colonocyte bioenergetic failure rather than primary microbial dysbiosis. The full framework is at the Host Capacity Model page. This section summarizes the parts that bear directly on SIBO recurrence.

The thesis is direct. The microbial composition of the colon, and the cecal and ileocecal region where SIBO presents, is determined by the bioenergetic capacity of the host's colonocytes. When the colonocyte can maintain its metabolic function, it sustains the conditions that allow the obligate anaerobic core community to dominate. When the colonocyte loses that capacity, those conditions collapse, and facultative anaerobes expand into the open niche. SIBO is the sentinel of that expansion. Recurrence is the equilibrium state of a host substrate that has not been restored.

Three specific failures in the colonocyte produce this pattern. Each is mechanistically distinct. Each is a potential point of intervention. Most patients with recurrent SIBO have some combination of all three.

Failure 1: SLC5A8 silencing

Butyrate is the dominant fuel of the colonocyte. The cell draws 70 to 80 percent of its energy from butyrate oxidation. For oxidation to occur, butyrate must first enter the cell, and the principal apical transporter for that entry is SLC5A8, a sodium-coupled monocarboxylate transporter expressed at the brush border.

SLC5A8 is epigenetically regulated. Its promoter is silenced by CpG methylation under conditions of chronic mucosal inflammation. The inflammation that drives this silencing has many sources: post-infectious immune activation, sustained barrier compromise with low-grade endotoxemia, food antigen exposure across a leaky barrier, or the broader inflammatory tone of post-viral illness. The mechanism is the same regardless of the trigger. Chronic inflammation methylates the SLC5A8 promoter. The transporter is silenced. The colonocyte loses its primary route for butyrate uptake.

The functional consequence is a decoupling. The patient may have adequate luminal butyrate. They may even be supplementing with high-dose butyrate. The substrate is present. The cell cannot access it. The oxidation machinery downstream of the missing transporter has nothing to oxidize.

This is one of the reasons butyrate supplementation produces such inconsistent results in this patient population. Patients with intact SLC5A8 expression respond. Patients with silenced SLC5A8 do not. The clinical heterogeneity in the literature is, under the HCM lens, exactly what the model predicts.

SLC5A8 silencing is partially reversible. When the inflammatory drive is removed, the promoter demethylates and expression returns. But the timescale is months, not days. This is one of the reasons substrate-restoration work is patient with timeline expectations.

Failure 2: Iron-sulfur cluster damage

Once butyrate enters the colonocyte, its oxidation depends on the electron transport chain. The electron transport chain depends on iron-sulfur cluster proteins. Complex I contains eight Fe-S clusters. Complex II contains three. The Rieske protein in Complex III is an Fe-S protein. Aconitase, the TCA cycle enzyme that feeds NADH and FADH2 generation, is an Fe-S protein. The entire oxidative machinery rests on dozens of Fe-S clusters distributed across the mitochondrial matrix and inner membrane.

Fe-S clusters are continuously damaged and continuously rebuilt. Under healthy conditions, the rebuild rate keeps pace with the damage rate, and the electron transport chain operates at capacity. Under inflammatory conditions, two things change. The damage rate accelerates, driven primarily by peroxynitrite, the product of nitric oxide and superoxide that accumulates during sustained inflammation. The rebuild rate slows, because Fe-S cluster biogenesis requires substrate (cysteine, iron, NAD+) and intact machinery (frataxin, NFS1, ISCU) that the inflammatory state degrades or depletes.

When damage outpaces rebuild, the electron transport chain operates below capacity. Butyrate that enters the cell encounters β-oxidation machinery that cannot dispose of the reducing equivalents it generates. The TCA cycle backs up. The cell shifts toward glycolysis. ATP output falls. Oxygen consumption falls.

This is the second mechanism producing the loss of colonocyte bioenergetic capacity. It is independent of SLC5A8 silencing, but in chronic illness, the two typically co-occur because the inflammatory drive that methylates SLC5A8 also damages Fe-S clusters.

Failure 3: The CD38-NAD+-SIRT3 cascade

The third failure operates at the regulatory layer. SIRT3 is the dominant mitochondrial NAD+-dependent deacetylase. It controls the activation state of dozens of mitochondrial enzymes including long-chain acyl-CoA dehydrogenase (controlling β-oxidation), succinate dehydrogenase (Complex II), isocitrate dehydrogenase (TCA flux), and manganese superoxide dismutase (matrix antioxidant defense). SIRT3 maintains these enzymes in their deacetylated, catalytically competent state.

SIRT3 activity depends on NAD+. The catalytic mechanism consumes NAD+ stoichiometrically. When NAD+ falls, SIRT3 activity falls in direct proportion. Hyperacetylated enzymes accumulate. Each operates at reduced rate. The aggregate is a mitochondrial proteome that has the structures and the substrates but has lost the cofactor that maintains the proteome in its active state.

CD38 is the enzyme that drives the NAD+ collapse. CD38 is a plasma membrane glycoprotein with NAD+ glycohydrolase activity. Its catalytic turnover for NAD+ is two to three orders of magnitude higher than the sirtuins. At baseline expression, CD38 is a modest contributor to NAD+ metabolism. At inflammation-induced expression, it dominates.

CD38 is upregulated by type-I interferons (the principal output of post-viral immune activation, which is why CD38 rises sharply after SARS-CoV-2 infection or EBV reactivation), by lipopolysaccharide (the principal output of barrier failure and endotoxemia, which is why CD38 tracks gut permeability in the absence of acute infection), and by the inflammatory cytokines of chronic low-grade inflammation. The induction is substantial. Five-fold to ten-fold increases are well-documented in aged and inflamed tissue.

When CD38 induction is sustained, the NAD+ pool cannot keep pace. SIRT3 is disabled. The colonocyte's mitochondrial regulation fails. Butyrate oxidation, even if substrate and machinery are intact, operates well below the rate the cell needs to maintain its energetic capacity. Oxygen consumption falls. Physiological hypoxia at the epithelial surface is lost.

Why the loss of epithelial hypoxia drives the recurrence

The healthy colonic environment is hospitable to obligate anaerobes because the colonocytes consume the oxygen at the surface. Faecalibacterium prausnitzii, Roseburia intestinalis, the Lachnospiraceae and Ruminococcaceae more broadly, the taxa most consistently associated with healthy gut function, are obligate anaerobes that cannot survive at meaningful oxygen concentrations. The healthy community is built by the host. The host's oxygen consumption creates the niche the community lives in.

When colonocyte oxidative metabolism falters, oxygen consumption falls. Oxygen leaks into the lumen. The obligate anaerobes lose their competitive advantage. Facultative anaerobes, particularly Enterobacteriaceae like E. coli and Klebsiella, and in some patients the sulfate-reducing taxa, expand into the new aerobic-tolerant environment.

This is the mechanism of dysbiosis viewed correctly. It is not a primary microbial disease. It is the readout of a host substrate that has lost the capacity to define its own niche. The microbial composition changes because the conditions that maintained the previous composition no longer exist.

Antimicrobial treatment in this state reduces the displaced community temporarily. The Enterobacteriaceae or sulfate-reducers are knocked down. But the colonocyte substrate failure that allowed them to expand has not been addressed. The luminal oxygen remains elevated because colonocyte consumption remains low. The niche stays open. As soon as antimicrobial pressure ends, the same facultative organisms expand back into the same available niche. Recurrence happens.

This is the mechanism of recurrent SIBO. It is not a story about resistance or insufficient duration. It is a story about a substrate-defined equilibrium that the treatment did not move.

The H2S Complex IV recurrence loop

In hydrogen-sulfide-dominant SIBO, an additional mechanism deepens the recurrence pattern. When the displaced community is dominated by sulfate-reducing taxa, particularly Desulfovibrio piger and Bilophila wadsworthia, the metabolic byproduct is hydrogen sulfide. At low concentrations, H2S is a signaling molecule with vasodilatory and anti-inflammatory effects. At sustained high concentrations, it is a respiratory toxin.

H2S binds the heme iron in the active site of cytochrome c oxidase, Complex IV of the electron transport chain. It competitively inhibits oxygen reduction at the terminal step of electron transport. When Complex IV is inhibited, the upstream complexes back up, electron flow stalls, ATP synthesis falls further, and the cell shifts further toward glycolysis. The colonocyte that was already struggling to oxidize butyrate now cannot perform oxidative phosphorylation either.

This produces the most distinctive recurrence pattern of H2S SIBO. The bioenergetic lesion that allowed sulfate-reducers to expand has been deepened by the metabolic byproduct of their expansion. Antimicrobial treatment reduces the population. H2S falls. Complex IV inhibition lifts. Colonocyte metabolism partially recovers. But the underlying substrate failure was never addressed. As antimicrobial pressure ends, the niche reopens, the sulfate-reducers expand back, H2S production returns, Complex IV inhibition returns, and the cycle resumes.

H2S SIBO patients often describe this cycle in their own words. They feel substantially better during treatment and for the first few weeks after. They feel worse over the following months, often arriving at a worse baseline than they had before the first treatment. The clinical impression that "H2S is harder to treat" is, under the HCM lens, exactly what the model predicts. The mechanism is not that the bacteria are harder to kill. The mechanism is that the bacteria are producing a metabolite that compounds the substrate failure that allowed them to expand in the first place.

Patterns I see in cases of recurrent SIBO

Across the cases I have worked, certain patterns appear consistently. Recurrent SIBO patients tend to share most of the following:

  • A prior infection or significant antibiotic course preceding the first SIBO episode, often by 6 to 18 months. The infection or antibiotic depleted the obligate anaerobic core community, which never fully recovered, and the recurrent SIBO is the readout.
  • A history of NSAID use, alcohol use, or other mucosal-permeability drivers preceding the first episode. These are the inflammatory drivers that initiate SLC5A8 silencing.
  • Concurrent or sequential MCAS, POTS, fatigue, or post-exertional malaise. These are the systemic expressions of the same CD38 induction that compromises the gut substrate, expressed in mast cells, autonomic regulation, and skeletal muscle.
  • Inconsistent response to high-dose butyrate supplementation. This is the SLC5A8 signature.
  • Response to elemental diet that does not hold once normal eating resumes. The elemental diet starves the overgrowth temporarily but does not restore the substrate.
  • A pattern of worsening response with each successive antimicrobial round. This is the cumulative damage to the substrate that each treatment cycle leaves untouched.
  • Normal or near-normal conventional gastroenterology workup. The lesion is functional and cellular. Standard imaging and standard biopsy do not see it.
  • A sense from the patient that they are getting worse on protocols that are supposed to be helping. This is often the most reliable signal that the protocols are addressing the wrong layer.

These patterns are not diagnostic in isolation. They are pattern-recognition heuristics that, taken together, identify cases where the Host Capacity Model is the relevant framing.

Tests that help read these cases

The standard SIBO breath test gives the surface diagnosis but says little about the upstream lesion. Tests that add mechanistic information include:

  • GI-MAP or comparable stool microbiome panel. Looking specifically for: Akkermansia muciniphila and Faecalibacterium prausnitzii levels (proxies for the obligate anaerobic core), Methanobrevibacter smithii (methane-dominant pattern), Desulfovibrio (H2S pattern), beta-glucuronidase activity (mucosal inflammation), and secretory IgA (barrier and immune status).
  • Organic Acids Test (OAT). Particular attention to: 3-hydroxypropionic acid (B12-related metabolism), succinic acid and fumaric acid (TCA cycle markers), kynurenic and quinolinic acid (inflammation and tryptophan diversion), and 8-OHdG (oxidative damage to DNA, a proxy for the inflammatory tone driving the substrate failure).
  • Standard inflammatory panel. hs-CRP, ferritin, fibrinogen. Elevated ferritin in the absence of iron overload is a particularly useful marker of low-grade inflammation that is driving the substrate failure.
  • Zonulin if available. Direct marker of barrier permeability.
  • LBP and CD14 if accessible. Markers of endotoxin translocation. These are not always available in standard panels but are useful in cases where they can be obtained.

I write more about reading lab panels through the Host Capacity Model lens in the Lab Result Interpreter tool and in the relevant articles.

Why the timeline is longer than people expect

A patient who has spent two years cycling through SIBO treatments is typically looking for an intervention that works faster than the cycle they have been in. Substrate restoration is not that intervention. It is slower, not faster.

The epigenetic processes that need to reverse, particularly SLC5A8 demethylation, operate on a timescale of months. The Fe-S cluster biogenesis machinery that needs to recover operates on a similar timescale. The mitochondrial proteome remodeling that follows SIRT3 reactivation takes weeks once NAD+ is restored. The shifts in luminal oxygen that allow the obligate anaerobic core community to re-expand take additional months once the host substrate is supplying the conditions.

The realistic timeline for substrate-restoring work is three to six months for the first measurable shifts and six to twelve months for the full clinical response. Faster response to substrate work is generally a sign that something else was dominant in the case, and the substrate work is not the principal mechanism. Slower response is the rule for cases where multiple mechanisms compound.

This timeline is a feature of the biology, not a feature of the consultation. It cannot be shortened by working harder. It can only be respected. Patients who are not ready for this timeline are often better served by continuing the antimicrobial cycle they are in, with full understanding that the cycle will continue.

What this means for treatment

A treatment approach anchored in the Host Capacity Model differs substantially from standard SIBO treatment. The shape of the approach is roughly:

First, identify the upstream drivers sustaining the inflammatory tone. This may be a persistent low-grade infection, an ongoing dietary trigger, environmental exposures, sleep disruption, or unaddressed metabolic dysfunction. Until these are addressed, no substrate intervention can outpace the inflammatory drive that is consuming the substrate.

Second, restore the regulatory cofactor pool, particularly NAD+. This typically involves NAD+ precursor supplementation (nicotinamide riboside or NMN at clinically validated doses), attention to the methylation cofactors that support both substrate function and regulatory machinery, and removal of NAD+-consuming drivers where possible.

Third, work the gut substrate directly. This involves addressing SLC5A8 silencing where present, supporting Fe-S cluster biogenesis, sequencing antimicrobial intervention into a state that can sustain the result, and supporting the recolonization of the obligate anaerobic core community.

Fourth, address the autonomic layer. Vagal afferent restoration, parasympathetic support, autonomic conditioning. The cholinergic anti-inflammatory pathway is a substantial regulator of the inflammatory tone that drives the cascade.

Fifth, manage downstream symptoms with containment measures whose value increases as the upstream work removes the driver.

The full sequence operates on a timescale of months. Substantial improvement is generally not expected before three months and may take six to twelve months to fully express.

This is the approach a Biomelogic consultation works through. The deliverable is a written mechanistic analysis that organizes the case in HCM terms, identifies the dominant mechanism in the specific patient, and provides intervention sequencing for the patient's existing clinical team to implement. Biomelogic does not prescribe and does not replace the patient's clinicians. The work is educational systems-biology analysis delivered in coordination with licensed care.

Frequently asked questions about recurrent SIBO

How long does substrate restoration take to produce relief from SIBO symptoms?

Three to six months for the first measurable improvements, six to twelve months for the full clinical response. The timeline is set by the biology of SLC5A8 demethylation, Fe-S cluster biogenesis, and microbial community reorganization. Faster response suggests a different mechanism was dominant in the case.

Do antimicrobials still have a role under the Host Capacity Model approach?

Yes, but sequenced differently. Antimicrobials are used after substrate restoration is underway, when the host can sustain the result. Antimicrobials used before substrate restoration produce the recurrence pattern this article describes. The timing matters more than the choice of antimicrobial.

Why has high-dose butyrate supplementation not helped me?

Most likely SLC5A8 silencing. The transporter that carries butyrate into the colonocyte is methylated, and the cell cannot access the substrate even when it is present. This is a known mechanism that is not measured by standard testing but can be inferred from the clinical pattern of butyrate non-response despite intact butyrate-producing taxa on stool testing.

Is the Host Capacity Model recognized by mainstream gastroenterology?

Not yet. The framework synthesizes established findings from microbial ecology, mitochondrial biology, and epigenetics into a coherent clinical framing. The individual mechanisms (CD38-NAD+ relationship, SLC5A8 epigenetic regulation, the Bäumler-laboratory work on oxygen-dependent microbial succession) are well-established in their respective primary literatures. The integration into a clinical framework is novel work. The framework is being developed and tested through case work and through collaboration with academic partners.

Is Mohammed Attallah a doctor?

No. Mohammed Attallah is an independent systems-biology researcher and the developer of the Host Capacity Model. He is not a licensed clinician. Biomelogic provides educational systems-biology analysis that operates alongside the client's existing licensed medical team. The deliverable is a written mechanistic analysis intended for review and discussion with the client's clinicians.

How do I know whether my case fits the Host Capacity Model framework?

The strongest fits are cases with multiple rounds of antimicrobial treatment producing transient relief followed by recurrence, concurrent symptoms across multiple systems (gut, autonomic, immune, energetic), and a sense that the standard model has run its course. A free 15-minute discovery call is the lowest-friction way to determine fit without committing to a full consultation.

What does a Biomelogic consultation cost?

The Standard Consultation is $650 one time, which includes the case review, the live session, and the written mechanistic analysis. The full service menu is at /services. HSA and FSA eligibility varies and clients should check with their administrator.

Working with Biomelogic on recurrent SIBO

If the patterns above resonate with the case you have been working through, a Biomelogic consultation may be useful. The work is appropriate for patients who have completed at least one round of standard SIBO treatment, who are working with a licensed clinician they trust, and who are interested in a mechanistic re-reading of their case rather than a faster protocol.

The lowest-friction starting point is the free 15-minute discovery call. The call is not medical advice and not a sales pitch. It exists to determine whether the case is the kind the Host Capacity Model is appropriate for. If the answer is yes, the next step is the Standard Consultation. If the answer is no, the call ends with a referral to a more appropriate resource.

For patients ready to proceed directly to a full case workup, the Gate 1 intake form is the starting point. Gate 1 is a brief triage that takes about five minutes and requires no files. After Gate 1, fit-appropriate cases proceed to the full consultation.

For practitioners working with recurrent SIBO patients, the Practitioner Collaboration service provides a mechanistic re-read of a single case with the practitioner present.

For readers interested in the broader framework that this article draws on, The Host Capacity Model is the canonical framework page.

Mohammed Attallah is an independent systems-biology researcher and founder of Biomelogic, where he develops and applies the Host Capacity Model to complex chronic illness cases. He is not a licensed clinician. The framework is educational systems-biology analysis delivered alongside the client's licensed medical team. Biomelogic is based in Bowie, Maryland and serves clients worldwide via remote consultation.