Table 5.

Advantages and limitations of gastrointestinal (GI) digestion model systems.

In Vitro Bioaccessibility Methods	Advantages	Limitations
Solubility and Dialysability	Simple and inexpensive to conduct with enzymes and dialysis filters that chemically mimic oral, gastric and small intestinal digestion Inexpensive No human or animal subjects required	Does not represent peristaltic movements, secretions, or gastric emptying of the GI tract No gut microbial component
Static GI models	Simple to conduct in single bioreactor or flask with stirring and pH adjustments Inexpensive No human or animal subjects required	Broad variance in results due to reagent diversity, particularly digestive enzymes which differ in activity dependent on their source (human, porcine, rabbit, bacterial, or fungal) Continuous mechanical agitation is not representative of complex peristaltic movements, secretions, or gastric emptying of the GI tract No gut microbial component
INFOGEST static in vitro model	Addresses worldwide lack of cohesion in simulated digestive methods Standardised static method suitable for food based on physiologically relevant conditions which can be applied for various endpoints Pepsin determined to be the factor causing most variation—activity determination may be improved by pH stabilisation Inexpensive No human or animal subjects required	Continuous mechanical agitation is not representative of complex peristaltic movements, secretions, or gastric emptying of the GI tract No gut microbial component
Dynamic GI models	Holistic in vitro gastrointestinal model incorporating the large and small intestine More representative of human GI digestion as changing physicochemical conditions and peristaltic forces are simulated in functionally distinct zones Human faecal inoculum included to study the effect of colonic fermentation on the food sample and nutrient absorption Samples can be taken at any stage of the digestive process without pausing the experiment Bioaccessibility results of dynamic models have been shown to correlate with bioavailability of the same nutrient in vivo No human or animal subjects required	More costly and lower throughput than static models Lack of in vivo factors such as first pass effect, renal clearance, and metabolisation by intestinal epithelial cells.
Cell models	Representative of intestinal epithelial cells Parallels human in vivo absorption studies May be used to mimic the ability of food components to be actively or passively transported and assimilated across the intestinal epithelium Human cell lines can be used as well as animal cells Mucus-producing cell lines can be co-cultured to more closely resemble in vivo conditions	Time-consuming to culture cell lines Costly First pass effect, renal clearance, interaction of the food sample with other nutrients and anti-nutrients, and different absorptive capacities at each stage of the gastrointestinal tract are not represented
Ex vivo bioavailability methods	Multi-cell systems are more representative of intestinal epithelial behaviour in terms of food absorption than single cell lines Animal organ or tissue models can measure the oral bioavailability of bioactive food components Mimics arterial blood haemoglobin delivery by maintaining oxygen and carbon dioxide levels Precise measurement of electrical and transport parameters of intact epithelium Any type of intestinal epithelium from duodenum to colon can be studied, as well as other epithelia, such as the placental barrier No human subjects required	Extensive preparation Lack of inclusion of gut microbial influence Low throughput (mounted tissue models, such as Ussing chambers) Intestinal segment models have greater throughput, but no distinction between apical and basolateral side of the epithelium as tissue segments are fully submerged Short viability–epithelial intestinal tissue must be excised from animal within ~5 min of sacrifice Viability of intestinal tissues once the experiment begins is only ~150 min and not suitable for many oral bioavailability studies that require more time Limited range of measurements that do not fully describe the complex physiological system of the intestinal mucosa
In vitro fermentation models	Static batch or dynamic fermentation models can be used Batch models are simple to set up and inexpensive Evaluates the impact of gut microbial populations on food bioaccessibility and bioactivity without using invasive human or animal methods Dynamic multistage models overcome the issue of fermentation product build-up in static batch models. pH and nutrient availability within each chamber are controlled throughout fermentation Computerised dynamic models such as TIM-2, SHIME and SIMGI create an anaerobic environment representative of the upper and lower digestive tracts rather than the colon alone in terms of bacterial populations and SCFA production Long-term stability of the microbiome—can be assessed as it adapts SHIME has option to set parameters found in diverse groups—humans, animals, diseased, healthy, elderly, or infants, and compare alternate treatments in parallel Possible to create a luminal or a mucosal microbiome Easier to obtain ethical approval compared to in vivo studies	Dynamic multistage models are costly and complex to set-up In static sealed batch models, fermentation products such as SCFA and p-cresol can accumulate and there is a finite amount of substrate available for the bacteria Lack of realistic peristalsis; expensive set-up costs; and absence of a dialysis component and mucosal cells (in the original SHIME model) Lack of realistic peristalsis in SHIME model and absence of a dialysis component and mucosal cells (in the original model) Lack of intestinal epithelial and immune cells in some systems. No feed-back mechanisms Use of parameters such as pH, redox potential, and transit time based on healthy individuals may not be representative of many groups
In vivo bioavailability methods	Considered the gold standard and most accurate method for measuring bioavailability – analysis of metabolites in blood plasma and/or urine after a single dose, or controlled long-term consumption Reflects complete effect of digestion, first pass metabolism, Phase I/II biotransformation, host microbiota and fermentation on an orally consumed nutrient Balance studies collecting urine and stools to measure oral bioavailability are accurate Tissue distribution studies provide bioavailability data on the extent of absorption Data from in vivo studies is more clinically relevant and any side-effects induced by the consumed sample can be observed	Balance studies are laborious and more suited to laboratory animal models than human subjects Tissue distribution studies almost exclusively conducted on animals due to invasive nature Difficult to obtain ethical approval due to potential harm to animal or human participants and sacrifice of animals Usually more expensive and time-consuming than other methods Not suitable for high-throughput screening of bioavailability More difficult to control all variables due to naturally occurring differences in living organisms In vivo trials involving small cohorts may not be reflective of the bioavailability of a nutrient in the wider population