In 2002, the FAO/WHO published guidelines for probiotics. In it, a table lists laboratory tests commonly used to characterize probiotics, including: resistance to gastric acidity; bile acid resistance; immunomodulatory activity; adherence to mucus and/or to cells in cell culture; antimicrobial activity; and bile salt hydrolase activity. Some in the field insist that such tests are essential to proper probiotic strain characterization (although the FAO/WHO paper states that these tests need to be validated). Indeed, the literature is replete with papers reporting results of such tests, which aim to justify that some strain has ‘probiotic properties’, which make it a better choice than another strain.
The problem with such tests is we do not know if they are predictive. We do not know if strains that show the ‘desired’ trait (e.g., high level resistance to acid or bile, adherence to mucus or good bile salt hydrolase activity) perform any better than strains testing as poor or mediocre on those traits.
These tests are not validated in humans. What’s missing is that strains that perform poorly in these tests are never compared in a human trial to strains that perform well. There’s a good reason for this: human trials are expensive, and it’s just not practical to dedicate research funds to comparing such strains. Research sponsors would prefer to compare a target strain to an inactive placebo, increasing the likelihood of a significant finding.
A recent paper titled “Correlation between in vitro and in vivo assays in selection of probiotics from traditional species of bacteria” reviewed this literature on in vitro characterization assays. After reading the paper, I felt that the title of the paper would have more precisely been “The lack of correlation between in vitro and in vivo assays in selection of probiotics from traditional species of bacteria”. The authors cite one study in fish, which suggested that strains that better adhered in cell line screens persisted better than strains that didn’t adhere well. However, the cutoff for strains that adhered well (>15% adherence) and strains that did not adhere well (<8% adherence) seems like a small difference, suggesting a precision in this assay that seems beyond the method’s proficiency. Another study the authors cite was a human study, which tested immune marker response to a Bifidobacterium strain in humans (O’Mahoney et al. 2005). Preclinical animal models showed the same impact on immune markers as was observed in human subjects. This suggests the relevance of the animal model in predicting the human response, but the absence of a negative control (use of a strain, or preferably an isogenic mutant devoid of this capacity) weakens confidence that the preclinical trial was predictive.
The paper does a good job of reviewing available information, but in the end, does not provide convincing evidence that any in vitro or animal models used for preclinical characterization of probiotic strains have been validated in humans, even though the authors conclude otherwise.
It’s frustrating is that some in the field think that the value of these tests is self-evident. Of course you want a strain that is more resistant to bile and acid. But the problem is that the in vitro tests are just not reflective of the exposures the microbe will experience once swallowed. Results of in vitro tests can change dramatically by altering test conditions. What is the source of bile? How long is the exposure? In a real human, conditions are dynamic, not static. There are immune cells, secretions, other microbes, undigested food, and other substances in the human environment that are not present in the model system. There is constant longitudinal movement through ever changing conditions. In vitro adherence tests are notoriously condition-specific – alter the conditions and get a different result. Probiotics probably don’t even see intestinal epithelial cells, which in the human are protected by layers of mucus.
Last I checked (which is often) the definition of a probiotic does NOT require any of these attributes. The definition states that the probiotic must be shown to confer a health benefit and is silent on the need for any in vitro performance criteria. If it can deliver a health benefit and ‘flunk’ all these in vitro tests, then that is all that is needed. Also, not all probiotics even need the same characteristics to achieve their particular benefit. Probiotics applied directly to the skin or designed for oral health benefits clearly have no need of properties that enhance survival in the intestine.
There is one area where in vitro tests are essential: measuring a strain’s ability to survive handling and storage processes. Processes such as fermentation, concentration, drying, and storage can dramatically impact a probiotic’s ability to stay alive. Since a probiotic must be administered alive, these types of in vitro assessments must be conducted to assure probiotic survival at an efficacious dose throughout the supply chain.
In conclusion, in vitro characterization assays have not been validated to predict the functionality of probiotic microorganisms in the human body. I have no doubt researchers will continue to run these tests, but they should at least acknowledge the methods’ limitations regarding predictability.