Research Digest
The founding corpus: what the strain literature actually shows
The founding import indexed 217,246 papers across 445 strains. The finding that matters most is how little of that research has ever involved people.
The founding import was one long crawl. By the time it finished it had pulled in 217,246 papers covering 445 of the strains in the catalogue, and most of that literature, it turns out, has never been near a person.
Sort the corpus by how the work was actually done and the shape of the field is hard to miss. There are 82,776 laboratory studies and 28,781 in animals, against 4,554 randomized human trials and 415 meta-analyses. A 2024 review of the Lactobacillus evidence arrives at the same place from the other direction, cataloguing broad promise across many species while flagging, in its own title, how large the research gaps still are. The point is not that probiotics do nothing. It is that most of what has been published tests cells and mice, and the questions people actually ask have been answered far less often.
Where the human evidence holds up, and where it wobbles
The randomized trials are worth reading precisely because they are outnumbered, and the clearest thing they show is how little transfers from one strain to the next. Two twelve week trials tested closely related Lactobacillus strains on the same measures of body weight and body fat in overweight adults and reached opposite conclusions: L. curvatus HY7601 with L. plantarum KY1032 reported changes, while L. plantarum K50 found no significant difference from placebo at all. Same genus, same rough design, different strains, different answers.
Study design muddies things further. Many trials test a strain as an add-on to conventional treatment rather than on its own, as in this four strain trial run alongside standard medical care, which makes it hard to credit the strain for anything that shows up. Read each one as a single trial of a specific formulation under specific conditions, not as evidence of what any product will do for any person.
Association is not instruction
A lot of the interest in particular strains comes from observational work: a strain turns up more often in people who are doing well, so it gets cast as the cause. The catch is that a bacterium being common in healthy people is not evidence that adding it makes anyone healthier. That is a separate experiment, usually a randomized one, and for most strains it has either come out mixed or never been run.
The strains the literature keeps circling
Lactobacillus acidophilus and Lactiplantibacillus plantarum are the two most studied strains by a wide margin, tagged on roughly 16,400 and 15,800 papers, with L. casei close behind and L. rhamnosus a familiar name further down the list. A review of L. acidophilus is a good map of why. First isolated from the human gut in 1900, it tolerates acid and bile, sticks to the intestinal lining, and interacts with the immune system, which is more or less the same short checklist most of these strains are screened against. Depth of study is not proof of benefit, but it does show where the field has placed its bets.
What the founding corpus adds up to
The corpus indexes what has been published, careful work and thin work side by side, and the honest summary of the founding import is that probiotic research is broad, busy, and still front loaded with laboratory and animal studies. The human trials that do exist are small, strain specific, and often at odds with one another. That is not a reason to wave the field away. It is the reason to read it one strain and one study at a time.