Mismatch between Probiotic Benefits in Trials versus Food Products

Probiotic food products contain a variety of different bacterial strains and may offer different health effects. The objective was to document the prevalence and dosage of probiotic strains in the Canadian food supply and to review the literature investigating these strains in order to understand what health benefits these products may offer. The Food Label Information Program was used to identify probiotic-containing products in the food supply. PubMed, Web of Science, and Embase were searched for randomized controlled trials that tested the health effects of these strains in humans. There were six probiotic strains/strain combinations identified in the food supply. Thirty-one studies investigated these strains and found that they are associated with decreased diarrhea and constipation, improved digestive symptoms, glycemic control, antioxidant status, blood lipids, oral health, and infant breastfeeding outcomes, as well as enhanced immunity and support for Helicobacter pylori eradication. There were a limited number of studies investigating these strains. Many studies were funded by the food industry and tested dosages that were up to twenty-five times the dosage found in most food products. Probiotic food products could have health benefits not currently reported on their labels. However, many dosages are too low to provide the benefits demonstrated in clinical trials. Further research is needed to enable more effective use of these functional foods.


Introduction
Probiotics are "live microorganisms that when administered in adequate amounts confer a health benefit on the host" [1,2]. The benefits of consuming bacteria have been known since ancient times, when fermented milk was commonly prescribed to treat an upset stomach [3]. Today, the term "probiotic" has been defined and qualified by the World Health Organization, which put also forward guidelines to support their use. Accordingly, different probiotics have been shown to prevent or treat a wide range of health issues, including respiratory tract infections, infectious diarrhea, atopic eczema associated with cow's milk allergy, infant colic, necrotizing enterocolitis, pouchitis, bacterial vaginosis, Clostridioides (formerly Clostridium) difficile-associated diarrhea, and urinary tract infections [4][5][6].
Probiotic food products are one of the fastest growing product markets globally [7]. Currently, commercial probiotic food products contain a variety of different probiotic species and strains. Certain health benefits are common to most or all probiotic species. These effects are considered "core benefits"

Investigation of Probiotic Strains in the Food Supply
Data was derived from the Food Label Information Program (FLIP), a database of Canadian food package label information derived from major outlets of the three largest grocery chains in Canada (Loblaws, Metro, and Sobeys) and one major western retailer (Safeway) [13]. This database represents 75.4% of the grocery retail market share in Canada [14] and provides a detailed assessment of the nutrition information found on Canadian packaged food labels. Grocery store shelves were systematically scanned, and data for every food product with a Nutrition Facts table (NFt), including all available national and private label brands, were collected. Data for food products sold at multiple retailers were collected only once. When multiple sizes of a product were available, only one size was collected. However, all flavors and varieties of a product were collected. Information collected for each product included the Universal Product Code, company, brand, price, Nutrition Facts table information (serving size, calories etc.), ingredients, container size, nutrient content claims, disease risk reduction claims, function claims, front of pack symbols, children's marketing, and other claims (e.g., organic, natural, and gluten-free), in addition to the date and location of sampling. The FLIP database is updated every three years. Presently, two collections have been completed (in 2010 and 2013) and have been described in greater detail elsewhere [13,15]. The packages were visually inspected, and ingredient lists of the 15,341 unique products collected in 2013 were searched to identify probiotic-containing products. Fermented foods were not considered to be probiotic products unless they were labeled as being probiotic. The species, strain(s), and dosage found in the 92 probiotic-containing products were recorded and tabulated. In July 2016, Loblaws, Metro, and Sobeys were revisited to identify if any probiotic strain and dosage information had changed and to investigate if new probiotic products had entered the marketplace. Four new probiotic products were identified and included in this study. Companies that listed species names without strain information were contacted via e-mail to inquire whether strain data could be disclosed. One company provided strain information via e-mail.

Review of Randomized Controlled Trials Testing the Probiotic Strains Found in the Canadian Food Supply
A systematic search of the peer-reviewed literature investigating each strain/strain combination found in the food supply was conducted in 2016 in accordance with the preferred reporting items for systematic reviews and meta-analysis protocols (PRISMA) checklist (with the exception of items related to meta-analyses) [16]. The full detailed protocol for this is available at PROSPERO registry CRD42106042660 [17].

Eligibility Criteria
Study Design, Treatment, and Participants Double-blind randomized-controlled trials that tested the effects of probiotic strains in the food supply were considered. The probiotic strains were required to be administered in a food format similar to the formats found in the food supply.
Studies that administered probiotics in supplement form, that tested synbiotics, or investigated the safety, tolerance, persistence, or viability of probiotics were not included.
Studies on humans of all ages were considered, with the exception of infants under six-months. Individuals with a chronic disease (like diabetes), infections (such as Helicobacter pylori), or conditions (like constipation or Irritable Bowel Syndrome) were included.

Outcome Measures
This was not a traditional systematic review. This was an exploratory review and descriptive synthesis that aimed to understand what health effects these food products may offer. Therefore, any and all health-related outcome measures in humans were recorded. This ranged from serum lipid and glycemic levels, to incidence/duration of infections and illness, to markers of inflammation. Effects detected in-vitro were not included. Effects on cellular immunomodulation (e.g., increased number of lymphocytes) were not included.

Literature Search
PubMed, Web of Science, and Embase were searched by two independent reviewers (Beatriz Franco-Arellano and Sarah Murphy) from the earliest record to July 2016. The following keywords were searched in the title/abstract: (multiple iterations of each strain name) and (yogurt OR yoghurt OR milk OR fermented milk OR dairy) with (randomized controlled trial) in any field. When the strain was not found in a dairy product, the dairy keywords were omitted. The search was limited to full-manuscripts in English. Only randomized controlled trials in humans were searched.

Study Selection
After the removal of duplicates, two independent reviewers (Beatriz Franco-Arellano and Sarah Murphy) screened the title and abstracts of retrieved studies against the a priori selection criteria. The selection criteria included any double-blind randomized controlled trial reported in a peer-reviewed journal that included strain/strain combinations found in the food supply, a control group, a quantified dose of the probiotic, a quantified measure of the food treatment, and oral administration of the probiotic via a food format. The study could test any clinical health endpoint on any human population (healthy or sick, including pregnant and breastfeeding mothers). Full-text screening was completed independently by two of the authors of this paper (Mary Scourboutakos and Sarah Murphy), with consensus required for inclusion or exclusion.

Data Extraction
The following information was extracted from each manuscript; information related to the article (complete citation plus author, country, and year of publication), the probiotic species and strain(s) tested, strain dosage, food format, population characteristics (e.g., adults, children, male, female, both), health status of the population (e.g., healthy, population with constipation, diabetic population), sample size, study duration, primary outcome measure, secondary outcome measure(s), significant outcomes, and source of funding. Data was independently extracted by one author (Mary Scourboutakos) and verified by a different author (Sarah Murphy). When articles reported insufficient information, attempts were made to contact their authors via-e-mail to retrieve further information.

Assessment of Methodological Quality
The study quality was independently assessed by one author (Mary Scourboutakos) using Health Canada's quality appraisal tool for intervention studies [18] and independently checked by another author (Sarah Murphy). This tool is used to evaluate the quality of studies that provide evidence to support health claim submissions. The risk of bias was assessed using the Cochrane risk-of-bias tool (Table A1) [19].

Data Synthesis
All studies were grouped according to the strain/strain combination they investigated, and health outcomes were recorded accordingly.

Results
The probiotic strains found in the Canadian food supply and a summary of their health effects are shown in Table 1. The initial search of the probiotic strains found in the food supply and their health benefits yielded 188 papers, with 95 remaining after the removal of duplicates ( Figure 1). After reviewing the titles and abstracts, 59 remained for full-text review, 29 of which were eligible for inclusion ( Table 2). All studies were deemed to be of a 'high quality' according to Health Canada's quality appraisal tool for intervention studies. The majority of studies were judged to have an overall low risk of bias (Table A1). Full-text articles excluded n = 29 • Single-blinded n = 7 • Confounding ingredients n = 2 • Additional strains n = 6 • Chewing gum food format n = 1 • Immunomodulatory outcome n = 6 • Supplemental form n = 4 • Pilot study/protocol n = 2 • Not English n = 1  X = beneficial effects observed in healthy adults; x = beneficial effects observed in healthy children, O = studies that have investigated this outcome and have found no significant effect in adults, o = studies that have investigated this outcome and found no significant effect in children, $ = indicates that the research was funded by the company that uses that particular strain in their products. A blank square indicates that no research investigating the effects of that strain/strain combination was identified during the systematic review of all literature published up to 21 July 2016, as described in the methods. All effects reported in this table were found in healthy populations that were not diagnosed with a chronic disease or condition. Definition of health effects: Constipation = improved stool frequency, consistency, or condition; Acute diarrhea = decreased incidence or severity of acute diarrhea; Antibiotic-associated diarrhea = decreased incidence of antibiotic-associated or Clostridium difficile-associated diarrhea; Digestive symptoms = decreased abdominal pain/discomfort, bloating, flatulence, or overall GI well-being; Glycemic control = improved fasting glucose, insulin, HbA1c (marker of long-term glycemic control), or HOMA-IR (measure of insulin sensitivity); Helicobacter pylori eradication = enhanced eradication of Helicobacter pylori infections; Immunity = decreased incidence and/or duration of common infectious diseases, including fever, cough, common respiratory infections (rhinitis, sore throat), common gastrointestinal infections (gastroenteritis, vomiting), asthma, or days missed from school; Infant breastfeeding outcomes = infants (2-6 months old) of mothers who consume this strain while breastfeeding had decreased incidence of gastrointestinal episodes and lower medication-use rates; Inflammation = decreased levels of inflammatory markers (ex. C-reactive protein); Lipids = decreased serum total cholesterol, low density lipoprotein (LDL), triglyceride levels, or increased high density lipoprotein (HDL); Oral health = decreased levels of cavity causing bacteria. * CFU = colony forming units. ** Iogo's Probio reported two strains on its label in 2013 (Bifidobacterium lactis BB12 + Lactobacillus acidophilus LA-5) and only one strain on its label in 2016. † These products were available in 2013 but may no longer be available in the Canadian market. Note: All cited references were deemed to be of high quality according to Health Canada's quality appraisal tool for intervention studies [18].  1 All probiotic strains in the Canadian food supply were recorded and a systematic review of their health effects was conducted. All literature published up to 21 July 2016 was included, as described in the methods. All studies included in the review were deemed to be of a 'high quality' according to Health Canada's quality appraisal tool for intervention studies and thus are considered eligible to substantiate a health claim [18]. $ Indicates that funding was provided by the food industry HbA1c = hemoglobin A1c, a long-term measure of glycemic control; HOMA-IR = a measure of insulin sensitivity; LDL = low-density lipoprotein; HDL = high-density lipoprotein; VLDL = very low-density lipoprotein; IBS = irritable bowel syndrome; CID = common infectious diseases. DN 114-001). This was one of the most well studied strains in the food supply with eleven studies, all funded by Danone, investigating its effects [35]. Three studies showed decreased incidence [26,35] and duration [36] of common infectious diseases (ranging from upper respiratory tract infections to sore throats and influenza) in adults, children, and seniors. Of these, one study showed decreased duration of acute diarrhea in children [31]. One study of hospitalized elderly adults showed decreased incidence of Clostridium difficile and antibiotic-associated diarrhea [33]. Other effects associated with this strain included decreased asthma and rhinitis episodes [30] and increased Helicobacter pylori eradication rates in children [34]. One study tested the effect of this strain when consumed by breastfeeding mothers and showed that their infants had a reduced incidence of gastrointestinal episodes and a lower rate of medication use [37]. The probiotic dosage administered in these studies was up to three times the dosage found in one serving of this product.

Danone's DanActive contained one proprietary strain (Lactobacillus casei
Danone's Activia contained a different proprietary strain, Bifidobacterium lactis DN-173 010. This strain was associated with improved overall GI well-being, including decreased flatulence [38], decreased stomach rumbling, and improved stool consistency [39]. In one study of women with irritable bowel syndrome (IBS), this strain was shown to decrease overall IBS symptom severity and to decrease maximal abdominal bloating [40].
President's Choice's ProAdvantage contained Lactobacillus acidophilus NCFM. One study tested this strain in children and found decreased incidence of fever, cough, rhinorrhea, antibiotic use, symptom duration, and days missed from school [42]. However, the dosage tested in the study (20 billion colony forming units (cfu) per day) was twenty times the dosage found in the product (1 billion cfu per day). Astro's BioBest contained Lactobacillus acidophilus NCFM in combination with Bifidobacterium lactis Bi-07. This combination was tested in the same study reported above and was found to have the same effects and dosage discrepancy.
Bifidobacterium lactis BB-12 was found in two brands; Iogo's Probio and Yoplait's Minigo (a product intended for children). This strain was investigated in four studies. In one study, testing a dosage that was half of what is found in these products, this strain was associated with decreased levels of a cavity causing bacteria (mutans streptococci) in saliva [29]. Two studies tested the effect of this strain (at a dosage that was ten-times the dosage found in the product) on children's risk of illness and absences from school [27,48]. No effects were seen. One study tested a dosage that was thirty-five times the dosage found in the products containing this strain and showed no effect on inflammatory markers (C-reactive protein and cytokines) [28].
Bifidobacterium lactis BB-12 in combination with Lactobacillus acidophilus LA-5 was found in two brands (Yoplait's Yoptimal and Lucerne's Organics). Eleven studies investigated this strain combination. Three studies tested dosages that were substantially smaller than the dosage found in commercial products. Two of those studies showed reduced salivary levels of cavity causing bacteria (Streptococcus mutans) [24,25], while one showed decreased duration of antibiotic-associated diarrhea in patients infected with Helicobacter pylori [20]. Two studies investigated the impact of these strains on blood lipids and found no effects despite the fact that one study tested a dosage that was lower than found in commercial products and the other tested a dosage that was higher [22,23]. One study investigated the effect of these strains on glycemic control [21]. It tested a dosage that was three-times the dosage found in commercial products and found decreased insulin sensitivity. There were four studies that tested the effect of these strains on type-two diabetics and used dosages that were similar to those found in commercial products (Table 3). These studies showed improved glycemic control [43,47], improved blood lipid levels [44,45,47], and enhanced antioxidant status [43] in diabetics.
Nine brands labeled species names without identifying the strain (Table 4). Therefore, strain-specific health benefits could not be inferred for these products. A blank square indicates that no research investigating the effects of that strain/strain combination was identified during the systematic review of all literature published up to 21 July 2016, as described in the methods. Effects reported in this table were observed in populations that were diagnosed with a disease or condition Definition of health effects: Antioxidant status = activity of superoxide dismutase, glutathoine peroxidase, and total antioxidant status; Digestive symptoms: decreased abdominal distension/pain/discomfort, decreased fecal transit time, reduced IBS symptom severity; Glycemic control = decreased fasting blood glucose, insulin, and/or HbA1c (long-term measure of blood glucose control); Inflammation = Increased levels of anti-inflammatory markers (cytokines: IL-6, IL-10, TNF-α, adiponectin, and resistin) Liver damage = decreased serum levels of liver enzymes (alanine aminotransferase and aspartate aminotransferase) [a marker of decreased liver damage]; Radiation Induced Diarrhea = incidence and severity; Serum Lipids = decreased serum total cholesterol, LDL, or triglyceride levels; increased HDL; improved lipid ratios. * CFU = colony forming units. † This product may no longer be available in the marketplace Note: All cited references were deemed to be of high quality according to Health Canada's quality appraisal tool for intervention studies [18]. Table 4. Additional probiotic products with undetermined health effects.
A: Products whose specific health effects are undetermined because there is no research on the strain/strain combination in their specific food format  Lactobacillus acidophilus + Bifidobacterium lactis President's Choice's Greek Probiotic ** Yogurt 1 × 10 9 * Probiotic containing foods were identified in the Food Label Information Program (FLIP), a database of Canadian food package label information. FLIP data was collected in 2013 from major outlets of the four largest grocery retail chains in Canada. Probiotic species/strain information was obtained from the ingredients list and package of each probiotic product. Data was re-verified in 2016 to ensure that the species and dosage information had not changed. ** These products were not included in the 2013 database but were identified when grocery chains were revisited in 2016. † The strains associated with Liberte's effervescent Kefir are not listed on the product label. This data was obtained via an inquiry with the company. All companies that listed species names without strains were contacted to inquire whether strain data could be disclosed.
Most products contained one or two different strains. Kefir (here a fermented milk with added probiotics) products had the largest strain and species diversity, as well as the highest dosage (45 billion colony forming units per serving). However, not all kefir products contained this dosage and diversity.

Discussion
Probiotic food products in the Canadian marketplace contained bacterial strains that were associated with a wide variety of health benefits ranging from enhanced immunity to improved glycemic control in diabetics, suggesting that probiotic products could potentially offer health benefits that are not advertised on their labels. However, many of the current probiotic dosages in products were lower than the dosages tested in randomized controlled trials.

Dosage
In order to obtain many of the health benefits reported in the randomized controlled trials that were reviewed in this study, consumers would need to eat anywhere from two to twenty-five servings of these products each day. The WHO has recommended that "the suggested serving size (on the product label) must deliver the effective dose of probiotics related to the health claim" [8].
Currently most products contain one billion colony forming units (CFU) of probiotics because that is the minimum required in order to provide core benefits and thus be eligible to display the probiotic health claim "promotes a healthy gut flora" in Canada [12]. Therefore, if strain-specific health claims were implemented (in addition to the existing general probiotic health claim), companies would have greater incentive to provide the higher dosages needed to convey some of the health benefits reported in this review.

Strain Diversity
Most products contained one or two strains. However, research has shown that, in some cases, strain mixtures can be more effective than single strains [50][51][52][53][54], as "different strains (that are) targeted toward different ailments can be blended into one preparation", enabling cultures to complement each other's health effects and produce synergistic benefits [55]. For instance, Bifidobacterium lactis BB-12 (found in the food supply) has been shown to have greater gut-adherence when accompanied by Lactobacillus rhamnosus GG (one of the most well-studied probiotic strains [4], which is mainly available in supplement form) [56]. Furthermore, evidence from Leyer et al.'s investigation of Lactobacillus acidophilus NCFM alone and in combination with Bifidobacterium lactis Bi-07 showed that the combination of strains resulted in a lower risk of fever, coughing, and rhinorrhea when compared to the single strain [42]. It is understandable that our results found that fewer products contained strain mixtures, as, presently, food companies have no incentive to utilize strain synergies. Not to mention that single strain probiotics are more easily patentable than multi-strain probiotics [52]. Therefore, current health claims that are based on a single strain encourage the addition of single strains and could therefore be partially responsible for promoting a potentially suboptimal pharmaceutical-like approach to probiotic foods. That being said, it should be noted that not all strain mixtures are beneficial, as strains can antagonize one another. Therefore, research is needed to verify if mixtures are synergistic or antagonistic [50]. It has been previously noted that there is a lack of research on multi-strain probiotics because such research is more difficult to conduct and thus more expensive [57].

Strengths and Limitations
A strength of this study is the use of FLIP to derive information on all marketed products, which is why we chose to focus on the Canadian market as a model. Obviously, different markets will have different probiotic-containing products, which may come with benefits that overlap or differ from those discussed here. It is important that future research focuses on these other markets to the benefit of both the consumers and the industry. Furthermore, many probiotic benefits could vary depending on an individual's lifestyle and baseline microbiome. Therefore, it is expected that the health effects noted in this review may not benefit all consumers equivalently.
Limitations include the fact that studies in this review tested various strains, dosages, and health outcomes. Therefore, at this point in time, there is no consensus on what strain, dose, or product is best. For example, while our review showed that one strain (Bifidobacterium lactis DN-173 010) was associated with decreased digestive symptoms, this was the only strain for which this outcome was assessed. Hence, we cannot conclude that other strains/products would not also have these benefits. Therefore, these results show what is known according to the limited amount of literature that currently exists. Additionally, since much of the current research was funded by the companies that sell probiotic products and therefore dictated which strains were studied, there is a need for further research on a broader range of species/strains that is supported by alternate funding bodies.
Despite the WHO's recommendation that genus, species, and strain should be designated on a product's label [8], nearly half of the brands in this study did not disclose strain information. Thus, the potential health benefits for these products could not be deduced.
Previous research has shown that industry funded nutrition-related research may bias conclusions in favor of the sponsors' products [58]. Most of the studies included in this review were funded by the companies making the products [26,30,[32][33][34][35][36][37]40,59] or were published in journals that are funded by the food industry (Table 1) [45,46]. Many of these studies investigated a large number of outcome measures but did not make statistical adjustments to control for testing multiple hypotheses. Furthermore, in many of these studies, the primary outcome measure was not significant, and, instead, significance was detected in secondary outcomes or through post-hoc analyses. Therefore, while these studies were deemed to be of high quality and were published in peer-reviewed journals, elements of their analysis suggest that their results should be interpreted with caution.

Conclusions
Probiotic food products sold in Canada could offer a variety of health benefits depending on the strain(s) and dosage they contain. That being said, the probiotic dosages contained in most food products are currently too low to provide the benefits shown in clinical trials. Therefore, with higher dosages, or with trails substantiating the current dosage, there is potential for the strains that are already in food products to provide more benefits to the consumer.
Currently there is only a small volume of literature investigating the health benefits of the probiotic strains used in the Canadian food supply. Thus, additional clinical trials, particularly ones that are not sponsored by the food industry, are needed. Hopefully this work will encourage funding and regulatory agencies to fund more research investigating probiotics. A larger number of well conducted studies and clear evidence-based labeling regulations will ultimately help the consumers to make informed choices and derive substantiated benefits form the products they choose to consume.
Overall, considering the wide range of diseases and health conditions for which probiotics have been shown to have benefits, further research to promote the optimal design of probiotic food products is needed to enable more effective use of these functional foods.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A Table A1. Appraisal of the risk of bias of the included studies using the Cochrane risk-of-bias tool. Caglar et al. [29] L L L L L L Merenstein et al. [48] L L L U L L Merentstein et al. [27] L L L U L L Kekkonen et al. [28] U U L L L U

Bifidobacterium lactis DN-173 010
Pinto et al. [41] L L L L L L Tabbers et al. [38] L L L L L L Guyonnet et al. [39] U U L L L U Agrawal et al. [40] U U L L L U Lactobacillus acidophilus NCFM + Bifidobacterium lactis Bi-07 Leyer et al. [42] L U L L L L Bifidobacterium lactis BB12 + Lactobacillus acidophilus LA-5 Ivey et al. [21] L L L L L L Ivey et al. [22] L L L L L L Sadrzadeh-Yeganeh et al. [23] U U L U L U de Vrese et al. [20] U U L L L U Ashwin et al. [24] U U L U L U Singh et al. [25] L L L U L L Ejtahed et al. [45] L L L L H L Mohamadshahi et al. [44] L L L U L L Ejtahed et al. [43] L L L L H L Nabavi et al. [46] L L L L U L Tonucci et al. [47] L U L L L L

Lactobacillus casei DN 114-001
Guillemard et al. [36] L U L L L L Merenstein et al. [26] L L L L U L Guillemard et al. [35] L U L L L L Sykora et al. [34] L H L L L L Ortiz-Andrellucchi et al. [37] U L L L U U Agarwal et al. [31] L U L L L L Hickson et al. [33] L L L U L L Giovannini et al. [30] L H L L L L Giralt et al. [49] L L L L L L Note: H = high risk of bias, L = low risk of bias, and U = unclear risk of bias.