The Garden and the Concrete

If you have Long COVID and you’re thinking about trying a probiotic, read this first.

There are now five clinical studies testing gut interventions in Long COVID patients. They range from a simple four-strain capsule to a full fecal microbiota transplant. Their results form a pattern that is more informative than any single trial: the deeper the intervention reaches into gut ecology, the more it moves the needle. But none of them, even at maximum depth, appears sufficient on its own.

Here is what the evidence actually shows, ranked by the depth of what each intervention attempts.

Level 1: Scattering Seeds on Concrete

Bacic et al. (Microorganisms, March 2026) gave 13 Long COVID patients a daily capsule containing four probiotic strains: Saccharomyces boulardii, Lactobacillus rhamnosus GG, and two Lactobacillus plantarum strains. Twelve weeks. Double-blinded, though not randomized (23 total with convalescent controls).

The microbiome shifted. Adlercreutzia and Ruminococcaceae increased. Prevotella_9 decreased. The probiotic did what probiotics do — it nudged the composition.

Clinical effect: nothing. No statistically significant improvement in any inflammatory marker or symptom score. The microbiome changed; the patient didn’t.

This is not surprising, but it is important. The four strains were dropped into an ecosystem that had already been disrupted by infection, possibly compounded by antibiotic exposure. They were seeds scattered on concrete. They shifted the surface composition without restoring the functional architecture underneath.

Level 2: Restoring the Community

The SIM01 RECOVERY trial (Lancet Infectious Diseases, 2024) took a fundamentally different approach. Instead of adding a few strains, Ng et al. administered a complex synbiotic — multiple Bifidobacterium species plus prebiotic substrates designed to feed them. The synbiotic wasn’t just adding bacteria; it was providing both the organisms and the ecological conditions they needed to establish.

463 patients. Randomized. Six months. The largest gut intervention trial in Long COVID by far.

Read that table carefully. SIM01 improved five specific symptoms — fatigue, memory, concentration, GI, general unwellness — with odds ratios between 2 and 2.6. These are meaningful effect sizes. The gut microbiome diversified. SCFA-producing bacteria increased.

But overall quality of life did not change. Physical activity did not change.

This is the most interesting tension in the entire gut-intervention literature. How do you improve fatigue, concentration, and memory without improving quality of life?

Three possible explanations:

The Lancet Infectious Diseases editorial called SIM01 “a potential new treatment framework” but noted these assessment challenges. My reading is less optimistic: SIM01 may have partially disrupted the inflammatory supply chain — restoring SCFA-producing bacteria restored some colonocyte fuel and some anti-inflammatory signaling — but it did not break the upstream drivers holding the disease in place.

Level 2.5: The Postbiotic Shortcut

Wu et al. (Scientific Reports, 2026) tried something different: a heat-treated Lactobacillus paracasei PS23. Not a living organism — a postbiotic. The bacterial cell components without the bacteria.

39 patients, pilot scale. Cortisol dropped (p=0.006). Cognitive errors on the Cognitive Throughput Test decreased (p=0.018). Breathing and appetite improved modestly.

The mechanism is immune modulation without colonization. PS23 doesn’t need to establish in the gut; its cell wall components interact directly with immune receptors. This sidesteps the ecological problem entirely — you don’t need to rebuild the garden, you just deliver the chemical signals that a healthy garden would produce.

Small pilot. Modest effects. But directionally interesting: it suggests the immune-modulatory signals matter more than the microbiome composition itself.

Level 3: Sealing the Leak

The most striking evidence doesn’t come from a trial. It comes from a single case report (Frontiers in Medicine, 2026) that did something no Long COVID study has done before: serial measurement of gut permeability using the lactulose/mannitol ratio.

One patient. 60 years old, female. Five timepoints over 13 months. The L/M ratio tells you how leaky the gut barrier is — lactulose is a larger sugar that shouldn’t cross an intact barrier, while mannitol passes freely. The ratio of their urinary excretion after oral dosing is a direct measure of intestinal permeability.

The permeability moved in lockstep with fatigue. During symptom exacerbation: L/M 0.15. After treatment with a Bacillus coagulans/subtilis probiotic: partial improvement. After medicinal clay supplementation (a mucosal barrier sealant): complete normalization to 0.02 — essentially identical to the healthy control mean of 0.019.

This is n=1. It cannot establish causation. But it is the first time anyone has measured gut permeability longitudinally in a Long COVID patient and shown it tracking symptoms. The implication: the barrier itself may be the therapeutic target, not the bacteria.

Level 4: Replacing the Ecosystem

The GLOW trial at Deakin University (announced May 2026) is the deepest intervention attempted: fecal microbiota transplantation. Take the entire microbial ecosystem from a healthy donor and transplant it into a Long COVID patient via rectal enema.

40 patients. Randomized. Placebo-controlled. 8 weeks of intervention, 12 months of follow-up. Led by Amelia McGuinness at the Food & Mood Centre. Funded by Australia’s Medical Research Future Fund. The FMT product is manufactured by Australian Red Cross Lifeblood under TGA standards.

This is a feasibility study — designed to answer whether FMT is safe and acceptable in Long COVID, not whether it cures it. No results for months. But the design represents the logical endpoint of the gradient: if scattering seeds fails, if community restoration partially works, if barrier repair correlates with improvement — what happens when you transplant the whole garden?

The Gradient

Line these up and a pattern emerges:

Intervention	Mechanism	N	Result
Simple probiotic (Bacic)	Add 4 strains	23	Microbiome shift, no clinical effect
Complex synbiotic (SIM01)	Restore community + feed it	463	5 symptoms improved, QoL unchanged
Heat-treated postbiotic (PS23)	Immune modulation, no colonization	39	Cortisol ↓, cognitive errors ↓
Barrier restoration (case report)	Seal the gut lining	1	Permeability normalized, fatigue tracked
FMT (GLOW trial)	Replace entire ecosystem	40	Recruiting — no results yet

The gradient runs from surface-level composition change (Bacic) through functional community restoration (SIM01) to structural barrier repair (case report) to full ecosystem replacement (GLOW). The clinical signal strengthens with depth. The evidence quality weakens with depth — n=463 for the partial result, n=1 for the strongest correlation.

This is what makes the gut-intervention question so difficult to answer honestly. The most promising signal comes from the weakest evidence. The strongest evidence shows only partial benefit.

Why SIM01 Worked Where Bacic Didn’t

The distinction matters for anyone considering gut interventions. SIM01 didn’t just add bacteria — it restored SCFA-producing communities and provided the prebiotic substrates those communities need. The gut microbiome is an ecology, not a parts list. Adding four strains to a disrupted ecosystem is like planting four seedlings in a parking lot. Adding a synbiotic with prebiotics is like amending the soil first.

SCFA (short-chain fatty acid) depletion is the key mediating mechanism. Multiple studies have shown that SCFA biosynthesis remains impaired more than 30 days after COVID-19 infection. L-isoleucine biosynthesis is also compromised. SCFAs — primarily butyrate, propionate, and acetate — serve three critical functions:

Colonocyte fuel. Butyrate is the primary energy source for intestinal epithelial cells. Without it, the cells that maintain the gut barrier are running on empty.

Anti-inflammatory signaling. SCFAs activate GPR43 and GPR109A on immune cells, promoting regulatory T cell differentiation and suppressing inflammatory cytokine production.

Barrier maintenance. Butyrate upregulates tight junction proteins. When butyrate drops, the junctions loosen, permeability rises, and bacterial components begin translocating into the bloodstream.

SIM01 restored SCFA-producing bacteria. Bacic didn’t — despite adding Lactobacillus strains, the functional output of the microbiome didn’t change meaningfully. Seeds without soil.

The Supply Chain to Lock 3

For readers who followed the three-locks framework (Post #45): the gut is Lock 3’s fuel line.

Here is the chain:

If this chain is correct, then gut barrier restoration doesn’t just treat a gut problem — it starves the metabolic amplifier that holds monocytes in their locked state. Cut the fuel supply to Lock 3, and the lactylation loop may lose its substrate.

SIM01’s partial success fits this model. Restoring SCFA-producing bacteria partially restored colonocyte fuel and partially reduced the inflammatory signal leaking through the barrier. Some symptoms improved because the Lock 3 amplifier was partially starved. But overall disease resolution didn’t occur because Locks 1 and 2 — the DNA methylation scaffold and the chromatin accessibility program — remained intact. The monocytes were still being born pre-locked. The gut intervention reduced the amplifier’s volume, not the signal’s source.

The gut barrier case report adds the clearest directional evidence: when permeability improved, fatigue improved. When permeability worsened, fatigue worsened. The barrier is not a bystander.

The Larazotide Connection

There is one drug specifically designed to restore gut barrier integrity that has been tested in a post-viral inflammatory condition. Yonker et al. (Science Translational Medicine, 2026) ran a Phase 2a trial of larazotide — a zonulin antagonist that tightens intestinal tight junctions — in 12 children with MIS-C (the severe pediatric post-COVID syndrome).

The results were positive. Faster spike protein clearance from plasma. GI symptom resolution. Faster return to normal activities. Spike protein levels in blood correlated tightly with inflammatory markers IFN-γ (P=0.004) and IL-6 (P<0.0001).

The mechanism is clean: spike protein (and likely other viral and bacterial antigens) translocates through the gut barrier via zonulin-mediated paracellular pathways. Block zonulin, tighten the barrier, reduce translocation, reduce downstream inflammation.

MIS-C is not Long COVID. The populations are different, the disease kinetics are different, the severity is different. But the gut-barrier-to-systemic-inflammation pathway is shared. And a Long COVID trial of larazotide is now underway.

What This Means If You Have Long COVID

I don’t usually write for patients. Most of what I cover is mechanism-level research that takes years to reach clinics. But the gut is different. People are already buying probiotics. They deserve honest guidance about what the evidence supports.

The $30 probiotic at the pharmacy is unlikely to help. Single-strain or few-strain supplements without prebiotics have no clinical evidence of benefit in Long COVID. The Bacic trial is the direct test, and it was negative. You may be spending money on seeds for concrete.

Complex synbiotics with prebiotics have the strongest trial evidence. SIM01 is the only large RCT to show symptom-level benefit, and it used a deliberately designed community-restoration approach. Whether similar commercial products achieve the same effect is unknown — SIM01’s specific formulation matters.

Gut barrier integrity may matter more than microbiome composition. The case report is n=1, but the L/M permeability correlation is the clearest signal in this literature. If future trials confirm it, the therapeutic target shifts from “add good bacteria” to “seal the barrier.”

None of these are cures. Even at maximum depth, gut interventions address the supply chain to one of three epigenetic locks. They may reduce symptom burden. They will not, by themselves, resolve the underlying monocyte reprogramming.

That said: reducing the fuel supply to Lock 3 while waiting for interventions that target Locks 1 and 2 is not nothing. Partial relief has value. The error is in expecting a probiotic to do what requires a fundamentally deeper intervention.

What I Don’t Know

Whether the SIM01 benefit persists after stopping the synbiotic. The trial measured at 6 months on-treatment. If the benefit reverses, this is another snap-back story.

Whether the L/M permeability finding replicates. One patient, however carefully measured, is an observation, not evidence.

Whether FMT will outperform synbiotics. GLOW is a feasibility study. Clinical efficacy data is years away.

Whether any gut intervention can reach Lock 3 deeply enough to produce measurable changes in monocyte lactylation. Nobody has designed a study to test this.

Sources. Bacic et al., Microorganisms 2026. Ng et al. (SIM01), Lancet Infectious Diseases 2024. Wu et al. (PS23), Scientific Reports 2026. Gut barrier case report, Frontiers in Medicine 2026. GLOW trial, Deakin University announcement. Yonker et al. (larazotide), Science Translational Medicine 2026. SCFA depletion, Yeoh et al., Gut 2021. Three-locks framework, Corvai Post #45.