A probiotic trial declared its main outcome in advance, and that outcome moved while everything else stayed put

What stool form is

Stool form is rated on a seven point scale, where the low end means hard separate lumps and the high end means liquid. It is used as a cheap stand in for how long material has spent in the colon, because the longer the transit the more water is drawn out of it. It is a measure of consistency, not of how anyone feels or how often they go.

Drawn from background clinical practice, not from this paper.

Why they ran it

The authors note that constipation comes with a recognisable microbial pattern, fewer of the bacteria that make short chain fatty acids, more of the potentially inflammatory ones, and lower microbial diversity overall, which they say may impair the movement and water handling of the gut wall. Their stated gap is that the preclinical work on this strain is promising while high quality randomised trials in people, measuring symptoms and inflammation and the microbiome together, are still missing.

Drawn from the paper's introduction.

Randomised, double blind, placebo controlled trial. Eighty seven adults with functional constipation were assigned 1 to 1 to Pediococcus acidilactici PA53 at 3.0 times ten to the tenth colony forming units per day, blended into three grams of dextrin, or to three grams of dextrin alone, for eight weeks. Forty four were allocated to the strain and 43 to placebo, and 81 completed. The primary outcome was the Bristol Stool Form Scale.

At week eight the strain group scored higher on stool form than placebo, 2.76 plus or minus 0.93 against 2.03 plus or minus 0.85, p equals 0.004, with a significant within group improvement from a baseline of 1.93 plus or minus 0.80, p equals 0.001. The placebo group did not change, 2.00 to 2.03, p equals 0.438.

The proportion of participants achieving four to six spontaneous bowel movements a week rose inside the strain group from 16.7 percent to 38.1 percent, p equals 0.028, but the between group difference was not significant, p equals 0.144.

Within the strain group motilin and interleukin 10 rose while somatostatin and interleukin 6 fell, all p below 0.05. No between group difference was significant for any biomarker. Motilin rose in the placebo group too, p below 0.001 in both arms.

Sixteen S ribosomal RNA sequencing showed marked enrichment of the supplemented genus. Microbial diversity fell inside the strain group from baseline to week eight, p equals 0.003 after correction for ACE and Chao indices, while the placebo group did not change. Between group diversity differences at week eight did not reach significance.

The strain was well tolerated, with no adverse events and no change in routine clinical chemistry.

The numbers

Stool form at week eight	2.76 vs 2.03 on placebo, p=0.004
Stool form within the strain group	1.93 at baseline to 2.76 at week eight, p=0.001
Stool form within the placebo group	2.00 to 2.03, p=0.438
Four to six bowel movements a week, within the strain group	16.7% to 38.1%, p=0.028
Four to six bowel movements a week, between groups	no difference, p=0.144
Microbial diversity within the strain group	fell, p=0.003 after correction
Biomarkers between groups	no significant difference for any
Registration	NCT06761443

Why this might happen

Proposed by the authors This is the explanation the authors offer in their discussion. This study did not test it.

The authors propose that the strain established itself, at least temporarily, and that the enrichment they measured is the precondition for any lasting effect. The step they name after that is fermentation: more short chain fatty acid production lowers the acidity inside the colon, which in turn stimulates the wall to contract.

They also offer a hormonal route, through motilin, which drives the coordinated contractions that move material along and is reduced in constipation, and somatostatin, which works the other way by relaxing the smooth muscle and slowing transit.

Then they withdraw that second explanation themselves, because both hormones moved in the placebo group as well. They write that any effect of the strain on intestinal motility through neuroendocrine signalling remains speculative and should be considered hypothesis generating rather than confirmed mechanistic evidence.

Drawn from Discussion, PMC13628159.

What this does not show

This does not show that the strain changes how often people go. Bowel movement frequency rose inside the group taking the strain and did not differ from the placebo group when the two were compared. The authors name both the sample size and the categorical way frequency was recorded as reasons the comparison may simply have been unable to tell.
This does not show an effect on inflammation or on gut hormones. Every hormone and inflammatory marker that moved in the strain group also moved in the placebo group, and no between group difference reached significance for any of them.
This does not establish the mechanism it proposes. Short chain fatty acids, tight junction proteins and mucosal immune markers were not measured at all, so the fermentation step the explanation turns on was never observed.
This was not tested against an inert comparison. The placebo was three grams of dextrin, which is itself a soluble fibre with its own mild effect on bowel function. The authors raise this themselves and state that a completely inert placebo cannot be guaranteed.

Where this leaves us

A probiotic result can now be read against what the trial promised to measure, and here that separates one real finding about stool consistency from a row of changes that happened just as much in the placebo group.

Adults with functional constipation, over eight weeks, taking a single bacterial strain blended into a small amount of dextrin.

The same trial run against a genuinely inert placebo, with short chain fatty acids measured, since that is the step the proposed mechanism turns on and the one thing the dextrin comparison cannot settle.

Caveats worth holding

One outcome of many reached between group significance, and it is the one the trial was powered for. Every other comparison is an underpowered null, which the authors state.
The hormone and inflammatory findings are within group changes that also occurred on placebo.
The placebo was dextrin, a soluble fibre, so an inert comparison cannot be assumed. The authors note this biases toward underestimating the effect.
Diet was not controlled, which the authors name as a potential confounder.
Short chain fatty acids, tight junction proteins and mucosal immune markers were not assessed.
Eight weeks is too short to speak to lasting colonisation, which the authors state.
The fall in microbial diversity inside the strain group is unexplained.
A baseline imbalance in interleukin 4 was present, and no treatment effect remained after adjustment for it.
Both the investigational product and the placebo were supplied by the manufacturer, declared under a disclosure heading alongside a statement that the authors have no conflicts of interest.