Cross-species examination of single- and multi-strain probiotic treatment effects on neuropsychiatric outcomes.

Study characteristics for the human trials are summarized by lifespan stage in Table 4 . Across trials, the most frequent single- and multi-strain combinations were L. casei (seven of forty-eight – 14.6% – six utilized substrain Shirota and one investigated ssp. rhamnosus ) and VSL#3 (two of forty-eight – 4.2%) which is comprised of S. thermophilus DSM 24731, B. longum DSM 24736, B. breve DSM 24732, B. infantis DSM 24737, L. acidophilus DSM 24735, L. plantarum DSM 24730, L. paracasei DSM 24733, and L. delbrueckii ssp. bulgaricus DSM 24734, respectively. Participants varied greatly by age range, physiological conditions, and psychiatric history. The number of investigations that verified medical or neuropsychiatric diagnoses, or lack thereof, with hospital or outpatient records could not be determined. Treatment and placebo group sample sizes were reported for all human studies. Within each trial, treatment and placebo groups were comparable in sex and age range distribution. Notable exceptions: one infant trial - a greater proportion of males were randomized to the placebo group, and one older adult investigation - a greater proportion of males were randomized to one of the L. helveticus IDCC3801 treatment conditions. one infant trial - a greater proportion of males were randomized to the placebo group, and one older adult investigation - a greater proportion of males were randomized to one of the L. helveticus IDCC3801 treatment conditions. Subjects in four out of the seven (57.1%) infant/child studies were defined as healthy. The three remaining trials enrolled infants considered to be premature, diagnosed with Wessel’s colic, or having a first-degree relative with dermatitis, atopic disease, or asthma. Five of the seven trials (71.4%) reported outcomes associated with L. reuteri (substrains DSM 17938 or ATCC5573). The sixth single-strain study determined the effects of L. rhamnosus GG ATCC53103 consumption. One (14.3%) multi-strain investigation was also conducted with S. thermophilus , B. animalis ssp. lactis BB-12, plus L. bulgaricus. Three of the five (60%) older adult studies enrolled individuals described as healthy. The two remaining reports conducted probiotic trials for post-surgical colorectal cancer patients or people diagnosed with Alzheimer’s disease. Probiotic strains tested varied for each of the older adult trials (see Table 1 ). Three (60%) single- and two (40%) multi-strain formulations were evaluated. Thirty-six young/middle-aged adult trials have been included in Table 1 . Thirty-two trials reported double- or triple-blinding. The last four investigations had indistinguishable blinding status and have been excluded from discussion (see Methods ). Eleven (34.4%) of these trials defined study subjects as generally healthy, of which one (9.1%) only enrolled female subjects. Seven (21.9%) investigations enrolled patients diagnosed with Rome II or III IBS. Two (6.3%) studies investigated probiotic treatment in the context of obese or overweight status. Two separate trials (6.3%) enrolled people diagnosed with hepatic encephalopathy. The twelve remaining reports recruited and investigated unique study samples. Precise characteristics have been noted in Table 1 . It is unclear whether an investigation of Bifidobacterium in hepatic encephalopathy patients should be characterized as a single- or multi-strain probiotic treatment. Eighteen out of thirty-two (56.3%) trials were conducted with single strains of probiotics. Six (33.3%) of the eighteen utilized L. casei (substrains - five of six Shirota and one ssp. rhamnosus ). Three (16.7%) of the eighteen investigated L. rhamnosus (substrains GG, HN001, or CGMCC1.3724). B. longum (substrains NCC3001 or ssp. infantis R0033) was tested three (16.7%) times. Two studies (11.1%) evaluated L. helveticus (substrains R0052 or Lafti L10). All other single-strain trials were conducted once as noted in Table 1 . Fourteen of the thirty-two (43.8%) young/middle-aged adult studies were multi-strain probiotic investigations. Two out of fourteen (14.3%) analyzed the effects of L. helveticus R0052 plus B. longum R0175. All other multi-strain formulations were tested once (see Table 1 ). Major characteristics by animal species are summarized in Table 5 . The most common single-strain probiotics tested were L. rhamnosus (substrains GG, JB-1, NC4007, and IMC501) and L. plantarum (substrains MTCC1325, MTCC9510, KY1032, PS128, and USDA-ARS; eight of fifty-eight trials each – 13.8%). Lacidofil ® – 95% L. rhamnosus R0011 and 5% L. helveticus R0052 – was the most frequently investigated multi-strain formulation (five of fifty-eight – 8.6%). Approximate animal age during probiotic treatment was reported for sixteen out of twenty-five (64.0%) rat, twenty-five out of twenty-seven (92.6%) mouse, and two out of five (40.0%) zebrafish studies. The age of most animals tested neurodevelopmentally coincided with human middle-adulthood. When reported, animal ages for treatment and control groups were comparable. Forty-two out of fifty-eight (72.4%) non-human studies utilized males. Three (5.2%) trials investigated females. Eight (13.8%) investigations observed both male and female animals. Numbers of animals by sex were not reported for five (8.6%) trials. Blinding status was indicated for twenty-five of the fifty-eight (43.1%) reports. Fifty-one (87.9%) trials administered probiotic in drinking water or provided control animals with an identical vehicle. In addition to the varying experimental paradigms further described by species below, three out of twenty-five (12.0%) rat, thirteen out of twenty-seven (48.1%) mouse, and one out of five (20.0%) zebrafish trials reported rearing animals in germ- or specific-pathogen-free environments concurrent to probiotic treatment. Multiple rat strains were utilized across trials. Fifteen of the twenty-five (60.0%) utilized Sprague-Dawley, eight (32.0%) Wistar, two (8.0%) Flinders Sensitive, and one (4.0%) Fischer 344. Paradigms varied among studies with some including multiple experimental manipulations: 1) eight models of stress (32.0%) – five maternal separation, two restraint, and one water avoidance; 2) six (24.0%) different diet modifications; 3) three (12.0%) aging/neurodegenerative models; 4) two (6.0%) myocardial infarction; and 5) two (6.0%) probiotic rescue of antibiotic-induced dysbiosis. All other designs were utilized once (see Table 2 ). Eleven out of the twenty-five (44%) rat studies employed single-strain probiotic therapy. Three out of eleven (27.3%) utilized B. infantis 35624. Three (27.3%) studies determined the effects of L. helveticus (substrains - two NS8 and one MTCC1325). Two (18.2%) trials each of L. plantarum (substrains MTCC1325 and unknown) and L. fermentum (substrains NS9 and CECT5716) were conducted. Additional single strain probiotic treatments were investigated once as noted in Table 2 . Fourteen (66%) rat studies were multi-strain formulation trials. Four out of fourteen (28.6%) trials investigated the effects of L. helveticus R0052 plus B. longum R0175. Two (11.8%) trials each of Ecologic ® Barrier, Lacidofil ® , and VSL#3 were conducted. The four remaining multi-strain combinations were tested once (see Table 2 ). The most frequently utilized mouse strain was C57BL6/J (sixteen of twenty-seven - 59.3%) followed by BALB/c (five of twenty-seven - 18.5%). Two (7.4%) trials tested probiotic effects in AKR/J mice. All other strains were investigated once each as noted in Table 2 . Two (7.4%) of the twenty-seven studies utilized knockout (−/−) mice: Rag1 on C57BL6/J and IL-10 on 129/SvEv. The most common experimental manipulation was stress exposure (seven of twenty-seven – 25.9%). Exposure types varied by trial and are noted in Table 2 . Aging, colitis, and vagotomy models were investigated three (11.1%) times each. Pathogenic infection, diet modification, antibiotic-induced dysbiosis, lipopolysaccharide (LPS) induced inflammation, antidepressant, or heat-killed probiotic comparisons were employed (7.4%) twice each. All other paradigms were evaluated once (see Table 2 ). Twenty-two out of twenty-seven (81.5%) reports can be stratified as single-strain probiotic trials. L. plantarum (substrains C29, CCFM639, MTCC9510, PS128) and L. rhamnosus (substrains GG, JB-1, NC4007) effects were evaluated five times (22.7%) each. Studies with B. longum (substrains NCC3001, 1714) were conducted four (18.2%) times. Three (13.6%) trials utilized L. casei (substrains 01, LABPC, DG) while two (9.1%) separate investigations determined the effects of B. breve 1205. All other single-strain probiotics were tested once (see Table 2 ). Five out of twenty-seven (18.5%) trials investigated multi-strain combinations of probiotics. Three of the five (60%) studies utilized Lacidofil ® while the two (40%) remaining analyzed VSL#3 intake effects. All zebrafish studies were conducted with wild-type animals. One of five (20.0%) reports also noted wild-type as “heterozygous”. Besides probiotic treatment, experimental designs included 1) stress, germ-free, and conventional environments; 2) high cholesterol diet; 3) ethanol exposure; and 4) chronic unpredictable stress. Probiotic strains tested included two (40%) L. rhamnosus (substrains GG and IMC501), two (40%) L. plantarum USDA-ARS, and one (20%) multi-strain formulation of P. acidilactici JN039350 plus L. plantarum JN039358 . The single female Japanese quail study investigated the effects of P. acidilactici R001 (MA 18/5M). Table 6 provides a cross-species comparison of single- and multi-strain probiotic trials in association with key neuropsychiatric outcome groupings. None of the trials measured anxiety-based phenotypes. Two of five (40.0%) trials assessed probiotic effects on anxiety with L. reuteri DSM 17938 or L. helveticus IDCC3801. No significant variations in anxiety symptoms were reported after 84 days of probiotic consumption. Twenty-four out of thirty-two (75.0%) trials determined anxiety outcomes associated with probiotic treatment. Twelve (50.0%) of the twenty-four were single-strain trials including five (38.5%) L. casei (substrains - four Shirota and one ssp. rhamnosus ), and two (15.4%) each of L. rhamnosus (HN001 and CGMCC1.3724) and B. longum ( ssp. infantis R0033 and NCC3001). Other single-strains were tested once as noted in Table 1 . Beneficial effects were observed with L. casei Shirota, L. gasseri CP2305, L. acidophilus NCFM, L. rhamnosus HN001, or B. bifidum R0071. The twelve (50.0%) multi-strain trials included two (18.2%) L. helveticus R0052 plus B. longum R0175 investigations. All other formulations were tested once (9.1%) as noted in Table 1 . Seven (58.3%) of these trials reported improved anxiety symptoms with 1) I.31 ( L. plantarum CECT7484 and CECT7485 plus P. acidilactici CECT7483); 2) L. gasseri SBT2055 plus B. longum SBT2928 in 100g yogurt containing S. thermophilus and L. delbrueckii subsp. Bulgaricus ; 3) L. helveticus R0052 plus B. longum R0175; 4) L. acidophilus LA5 plus B. lactis BB-12 in yogurt containing L. casei, L. acidophilus, L. rhamnosus, L. bulgaricus, B. breve, B. longum , and S. thermophilus ; 5) L. acidophilus plus L. casei and B. bifidum ; 6) S. thermophilus SGst01, B. animalis ssp. Lactis SGB06, S. thermophiles, B. bifidum SGB02, L. delbrueckii ssp. Bulgaricus DSM 20081, L. acidophilus SGL11, L. plantarum SGL07, L. reuteri SGL01; or 7) L. acidophilus , plus L. casei, B. bifidum , and L. fermentum intake. Of note, Lacidofil ® associated anxiety reduction was a within-group effect. No significant differences were observed between treatment groups. Thirteen out of twenty-five (52.0%) trials assessed anxiety behaviors. Six (46.2%) were single-strain trials that included two (33.3%) L. fermentum (substrains CECT5716 and NS9), two (33.3%) L. helveticus NS8, and the last two tested each of the following probiotics once: L. rhamnosus GG, B. breve UCC2003, B. infantis 35624, and L. salivarius UCC118. Four (66.7%) of the six studies reported significant improvements with L. helveticus NS8, L. rhamnosus GG or L. fermentum NS9 intake. Multi-strain (seven out of thirteen – 53.8%) evaluation of anxiety symptoms included three (42.9%) L. helveticus R0052 plus B. longum R0175, two (28.6%) Ecologic ® Barrier, one Lacidofil ® , and one VSL#3 trial. Two of the seven (28.6%) investigations noted a reduction in anxiety-based phenotypes with L. helveticus R0052 plus B. longum R0175 consumption. Notably, one of these studies indicated the beneficial effects of L. helveticus R0052 plus B. longum R0175 for anxiety symptom reduction was specific to the myocardial infarction (MI) model. Sixteen out of 27 (59.3%) trials assessed anxiety with L. rhamnosus (substrains NC4007, GG, and JB-1), B. longum (substrains NCC3001 and 1714), L. plantarum (substrains – two PS128 and one MTCC9510), B. breve 1205, L. reuteri MM4-1A ATCC-PTA-6475, L. johnsonii ATCC33200, E. faecium CFR3003, B. fragilis NCTC9343 , L. helveticus R0052, or Lacidofil ® . Thirteen out of fourteen (92.9%) single-strain studies reported a reduction in anxiety-based outcomes (see Table 2 ). One of the L. plantarum PS128 trials observed improvement in anxiety-like phenotypes specifically in mice naïve to maternal separation stress. Although one trial observed anxiety behavior reduction after B. longum NCC3001 intake (10 days), L. rhamnosus NC4007 consumption did not confer the same benefits. A concurrent investigation of B. longum 1714 and B. breve 1205 demonstrated that anxiety reduction was specific to B. breve 1205 treatment. The two Lacidofil ® trials also indicated a significant improvement in anxiety behavior scores post-treatment. Three out of five (60%) trials investigated probiotic treatment effects on thigmotaxis and novel tank diving using L. plantarum USDA-ARS or L. rhamnosus GG. Both of the L. plantarum USDA-ARS studies (66.7%) reported decreased anxiety-based behaviors in conventionally raised zebrafish after two or thirty days of treatment. However, L. plantarum USDA-ARS was unable to modify anxiety-like phenotypes in a germ-free environment. P. acidilactici R001 (MA 18/5M) did not have a significant effect on anxiety-based behaviors. Five out of seven (71.4%) trials assessed depression behaviors with single-strain probiotics including four (80%) L. reuteri (substrains DSM 17938, ATCC55730, or unknown) and one (20%) L. rhamnosus GG trial. The L. reuteri ATCC55730 trial conducted a parallel investigation with B. lactis BB-12. One of the five (20%) studies reported improvement in fussiness/crying after 90 days of L. reuteri DSM17938 treatment. Two (40%) reported negative effects, i.e. increased crying and irritability, with L. reuteri DSM 17938 intake. The negative findings were not replicated during post-treatment assessment and specific to formula-fed infants for one of the two trials. Two out of five (40%) trials investigated the effects of L. casei Shirota or L. helveticus IDCC3801 on depression outcomes. Although both studies observed no significant differences as a whole, L. casei Shirota consumption for twenty-one days improved depression symptoms for a subset of individuals experiencing poor baseline mood. Nineteen of the thirty-two (59.4%) trials assessed depression phenotypes. Ten (52.6%) of the nineteen were single-strain investigations that included two (20.0%) each of B. longum (substrains NCC3001 or unknown), L. casei (substrains ssp. rhamnosus LCR35 or Shirota), L. rhamnosus (CGMCC1.3724 or HN001) and one (10.0%) each of B. animalis ssp. lactis -07, S. boulardii , L. acidophilus NCFM, and L. helveticus Lafti L10. S. boulardii and L. casei ssp. rhamnosus LCR35 had no significant effect (20.0%) on depression outcomes. The eight remaining studies (80.0%) reported a reduction in depression symptoms after probiotic intake. Nine (47.4%) of the nineteen young/middle-aged adult studies were multi-strain trials. Two investigated the effects of L. helveticus R0052 plus B. longum R0175. All other combinations were tested once as noted in Table 2 . The Lacidofil ® and one L. helveticus R0052 plus B. longum R0175 trial did not observe significant differences in depression outcomes after probiotic intake. Of note, the Lacidofil ® trial reported significant within- but no between-group differences after 84 days of probiotic consumption. The seven (77.8%) remaining studies reported improvements in depression symptoms. Nine out of twenty-five (36%) trials investigated three (33.3%) single-strain – two B. infantis 35624 and one L. plantarum MTCC1325 – and six (66.7%) multi-strain – three (50.0%) L. helveticus R0052 and B. longum R0175, two (33.3%) Ecologic ® Barrier, and one (16.7%) L. rhamnosus plus B. longum – combinations of probiotics on depression-based outcomes. Two of three (66.7%) single- and five of six (83.3%) multi-strain trials reported significant beneficial changes in depression-like phenotypes post-probiotic intake. One B. infantis 35624 and one L. helveticus R0052 plus B. longum R0175 trial did not modify depression behaviors. Seven out of twenty-seven (25.9%) trials analyzed probiotic effects on depression-associated outcomes. Six (85.7%) were single-strain trials – two (33.3%) L. rhamnosus (substrains JB-1 or GG), two (33.3%) L. plantarum (substrains PS128 or MTCC9510), one (16.7%) L. casei DG, and one (16.7%) comparison of B. longum 1714 and B. breve 1205. Single-strain treatments (83.3%) improved depression-like phenotypes with the exceptions of L. rhamnosus GG and B. longum 1714. A VSL#3 (14.3%) trial also indicated a significant reduction in depression-based behaviors. None of the trials assessed depression-like phenotypes. P. acidilactici R001 (MA 18/5M) improved emotional reactivity after thirty-six days. One of seven (14.3%) trials investigated a combination of S. thermophilus, B. animalis ssp. lactis BB-12, and L. bulgaricus with 1 g inulin on pediatric Quality of Life (QOL) social functioning. This study reported improved social behavior after one hundred twelve days of probiotic consumption. One of five (20%) trials investigated VSL#3 with significant improvements in SF-36 social functioning at post-treatment assessment (twenty-eight days). Six out of thirty-two (18.8%) trials assessed the effects of probiotics on social ability. Four of the six (66.7%) utilized single probiotic strains that included B. longum (substrains NCC3001 or unknown), L. rhamnosus GG, or S. boulardii. Three of the four (75.0%) noted improvement in IBS, IBD, or SIP social function after twenty-eight, twenty-eight, or fifty-six days of probiotic intake, respectively. The two (42.9%) multi-strain trials were Lacidofil ® or a combination of S. thermophilus SGst01, plus B. animalis ssp. Lactis SGB06, S. thermophiles, B. bifidum SGB02, L. delbrueckii ssp. Bulgaricus DSM20081, L. acidophilus SGL11, L. plantarum SGL07, and L. reuteri SGL01. Both multi-strain formulations were not effective at modifying social behaviors. Three out of twenty-five (12.0%) rat trials evaluated L. helveticus R0052 plus B. longum R0175 in relation to social interactions. Two of the three (66.7%) indicated improved social function with probiotic consumption for fourteen days post-MI induction or seven days pre- and seven days post-MI. Five out of twenty-seven (18.5%) trials assessed the effects of probiotics on social ability. Two of the four (50%) single-strain investigations observed increased social interactions after twenty-eight days of L. rhamnosus JB-1 or L. reuteri MM4-1A-ATCC-PTA-6475 consumption. B. fragilis NCTC9343 , L. casei DG, or L. johnsonii ATCC33200 intake did not alter social behaviors. A twenty-day trial of VSL#3 had positive effects on social interactions. One out of five (20.0%) trials investigated and observed increased shoaling after twenty-eight days of L. rhamnosus IMC501 exposure. No social functioning phenotypes were assessed. One out of seven (14.3%) trials investigated psychomotor development effects associated with L. reuteri treatment. No significant changes were detected. Three of the five (60%) trials determined the effects of L. casei Shirota , L. helveticus IDCC3801, or a multi-strain combination of L. acidophilus , L. casei , L. fermentum , plus B. bifidum. The L. helveticus IDCC3801 and L. acidophilus , L. casei , L. fermentum , plus B. bifidum trials observed modifications in Rapid Visual Information Processing (RVIP) and Mini Mental Status Exam (MMSE) performance post-probiotic consumption. Eight out of thirty-two (25.0%) trials evaluated the effects of probiotic treatment on cognitive indices. Four of the eight (50.0%) were single-strains investigations of L. rhamnosus (substrains JB-1 or CGMCC1.3724), L. casei Shirota, or B. longum NCC3001. Multi-strain combinations included VSL#3, S. thermophilus SGst01, B. animalis ssp. Lactis SGB06, S. thermophiles, B. bifidum SGB02, L. delbrueckii ssp. Bulgaricus DSM20081, L. acidophilus SGL11, L. plantarum SGL07, L. reuteri SGL01, or combination of B. animalis ssp. lactis , plus L. lactis ssp. lactis, S. thermophilus , and L. bulgaricus. Except for one multi- and one single-strain trial, all of the young/middle-aged adult reports noted improved cognitive performance with probiotic consumption. One trial investigated the effects of Bifidobacterium and noted greater block design ability and shorter Trail Making A & B completion times after sixty days of probiotic intake. However, it is unclear if a single- or multi-strain formulation of Bifidobacterium was tested. Seventeen out of twenty-five (68.0%) trials assessed cognitive functioning. Six (35.3%) trials were single-strain investigations that included two (33.3%) L. plantarum (substrains MTCC1325 or unknown), two (33.3%) L. helveticus NS8, and one (16.7%) each of L. paracasei HII01, L. rhamnosus , and B. B94. Five out of the six (83.3%) studies noted cognitive improvement with probiotic treatment. Twenty-eight days of L. plantarum or B. B94 consumption did not modify spatial ability. The eleven (64.7%) multi-strain combinations were two each of (18.2%) VSL#3, Lacidofil ® , and L. helveticus R0052 plus B. longum R0175. All other multi-strain combinations were tested once (see Table 2 ). Two (18.2%) of the multi-strain trials did not report significant differences in cognitive performance after Ecologic ® Barrier or L. helveticus R0052 plus B. longum R0175 consumption. One (9.1%) study reported a decline in memory with Lacidofil ® intake in rats exposed to maternal stress. The eight (72.7%) remaining trials observed improvements in cognitive indices post-probiotic treatment. Sixteen out of twenty-seven (59.3%) trials reported probiotic effects on cognition. All sixteen studies observed improved cognitive task performance with probiotic treatment. Four out of sixteen (25%) trials were multi-strain designs - three Lacidofil ® and one VSL#3. One of the Lacidofil ® trials observed beneficial effects on memory in Rag −/− mice naïve to water avoidance stress (WAS). Probiotic treatment reduced memory ability in WAS exposed Rag −/− mice. One out of five (20.0%) trials investigated cognition-based outcomes. Forty-nine days of P. acidilactici JN039350 plus L. plantarum JN039358 consumption improved high cholesterol diet-induced decline in spatial memory. P. acidilactici R001 (MA 18/5M) improved memory performance during treatment days two and three.

The authors of this article are not associated with any of the trials that were examined. This review compared neuropsychiatric outcomes associated with single- and multi-strain probiotic treatments in humans and translational non-human animal models while adhering to systematic review (PRISMA) guidelines ( Moher et al., 2009 ; Shamseer et al., 2015 ). Studies comprising this article were ascertained with the PubMed Advanced Search Builder http://www.ncbi.nlm.nih.gov/pubmed/advanced and filtered by the English language. Publication dates were unrestricted and ranged from 2006 to 2018. The most recent search was conducted on April 18, 2018. The PubMed searches were supplemented with a collection of original reports from prior systematic review or hypothesis articles obtained from the Cochrane Database of Systematic Reviews http://www.cochranelibrary.com/cochrane-database-of-systematic-reviews/ with the search terms noted below. Database search keywords were selected to align this review with anxiety, mood, and psychotic disorder phenotypes. Exact search terms included probiotic and (PubMed/Cochrane Database) well-being (32/13), psychological stress (104/10), anxiety (145/30), worry (2/1), depression (171/1), mood (1133/12), bipolar disorder (7/1), mania (8/1), schizophrenia (19/4), psychosis (4/5), post-traumatic stress disorder (1/0), obsessive-compulsive disorder (4/0), negative symptoms (325/27), learning (46/7), memory (82/11), cognition (52/10), motivation (22/3), reward (2/0), social behavior (38/5), social function (16/1), and sickness behavior (16/0), an inflammation-mediated depression phenotype ( Brydon et al., 2009 ). Most trials selected for review utilized established probiotics belonging to the Bifidobacterium or Lactobacillus genera ( Fijan, 2014 ; Hill et al., 2014 ). Investigations of Bacillus subtilis, Clostridium butyricum, Enterococcus faecium, Escherichia coli Nissle, Lactococcus lactis, Pediococcus acidilactici, Saccharomyces boulardii, Saccharomyces cerevisiae , and Streptococcus thermophilus , additional species with recognized probiotic properties, were also included in this review. Investigations with Leuconostoc genera were not available with our defined search criteria. Precise substrains varied by trial. All case reports, retrospective studies, review articles, non-experimental studies (e.g. internet or survey-based), non-randomized or placebo-controlled trials (human), or reports that failed to include a control group (non-human), were eliminated. Since probiotics are defined as live microorganisms ( Hill et al., 2014 ), studies that only investigated heat-killed probiotics were removed (e.g. Shinkai et al. (2013) . Additional reports were rejected if the only microbial treatment was classified as a pathogenic or engineered strain ( Miyazawa et al., 2015 ; Shinkai et al., 2013 ). Infant trials in which probiotic outcomes were assessed in the mothers but not the consuming infants ( Mi et al., 2015 ; Sung et al., 2014 ) were also excluded. All trials that failed to assay neuropsychiatric phenotypes were omitted from this review. However, the physiological outcomes from otherwise eligible studies that exposed non-human animals to neuropsychological stress as part of the experimental paradigm have been recorded in Table A.1 (Appendix). Figure 1 delineates the number of articles from each stage of our literature search strategy to yield the final set of human ( Table 1 ) and non-human ( Table 2 ) trials for review. Due to profound differences in experimental designs, neuropsychiatric outcome assessments, physiological indices, probiotic strains, and substrains reported by the trials ascertained for review (see Tables 1 and 2 ), meta-analyses were not conducted. Most trials indicating significant differences associated with probiotic treatment reported beneficial effects on neuropsychiatric outcomes. If a probiotic treatment led to a decline in neuropsychiatric performance or functioning, these findings have been highlighted in the Results. Both review authors assessed the study quality and risk of bias for all trials in Tables 1 and 2 with 100% consensus. Human study quality and risk for bias were evaluated with the PEDro scale ( Maher et al., 2003 ), Quality Index ( Downs and Black, 1998 ), and the Cochrane Collaboration Tool ( Higgins et al., 2011 ). Individual item scores, mean total scores, and interquartile ranges for these assessments are noted in Tables A.2 - A.4 (Appendix) for each study. Study quality and risk of bias for the non-human animal trials were estimated with modified criteria from Macleod et al. (2004) . The amended scale items, individual item scores for each study, total mean scores, and interquartile range can be examined in Table A.5 (Appendix). Specific-pathogen-free or germ-free conditions were not taken into consideration when scoring the non-human animal study environment. If such conditions were reported, they are noted in Table 2 and summarized in the Results. For all study quality and bias scales except the Cochrane Collaboration Tool, total scores were normally distributed and higher scores corresponded with greater study quality or lower risk for bias. Each item of the Cochrane Collaboration Tool was dichotomized as having a high or low risk for bias, except when the item criteria could not be obtained from the study report. If this condition was met, the item was recorded as having an unclear risk for bias. Because the validity and reliability of excluding investigations for systematic review based on these types of assessments continues to be actively debated in the extant literature ( Ilgen et al., 2015 ; Juni et al., 2001 ), our global assessment of human and non-human trial methodology suggested excessive heterogeneity, and since we were unable to obtain meaningful minimum scores with the quality and risk of bias assessments, these scores were not considered when evaluating studies for final inclusion in Tables 1 and 2 . However, to maximize scientific rigor and minimize bias, we excluded reports noted in Table 1 from Tables 4 - 6 and the Results below if we could not confirm at minimum double-blinding. Notably, the standardized scores for the Quality Index and PEDro Scale were within 2.5 deviations of the mean (i.e. Z-Score Range −2.5 to +2.5). Due to the varied and limited reporting of blinding status for the non-human trials (43.1%), we did not employ the double-blind exclusion strategy for these investigations. Table 3 is a human vs. non-human comparison of the multiple neuropsychiatric scales and behavioral measures utilized as indices for stress, anxiety, depression, cognition, and social functioning.

Across all species, L. rhamnosus (substrains GG, JB-1, NC4007, IMC501) was the most common single-strain probiotic treatment investigated (12.3%). VSL#3 and Lacidofil ® were the most frequently tested multi-strain formulations (5.7% each). A gross difference between the human and non-human trials was the greater enrollment of human female subjects ( Table 4 ), whereas non-human studies conducted most experiments with male animals ( Table 5 ). The cross-species overview ( Table 6 ) indicates that both single- and multi-strain combinations of probiotics may influence cognition, social function, anxiety, depression, or other emotional behaviors with similar efficacy in humans. For the non-human studies, multistrain combinations were more likely to modify cognition and social behavior, whereas single- and multi-strain combinations may be comparable in ability to regulate anxiety, depression, or emotional behaviors. Because variation in probiotic combinations and experimental designs among the human and non-human trials hindered our ability to conduct meaningful meta-analyses ( DerSimonian and Kacker, 2007 ), these and the other findings reported in this review should be considered crude and preliminary estimations. There are profound gaps in our understanding of probiotic treatments that should be addressed in forthcoming human and non-human translational trials. Non-neuropsychiatric outcomes assayed in conjunction with each probiotic treatment varied widely and included neural, immune, anthropometric, gastrointestinal, neuroendocrine, or metabolic markers. Since these biomarkers are likely to be critical components in our mechanistic understanding of gut-brain interactions, thoughtful incorporation and replication of precise biomarkers is germane to the success of future probiotic investigations. Specific to the human reports, several healthy adult trials noted in this review did not observe significant probiotic treatment effects on neuropsychiatric outcomes. Although healthy subject investigations minimize illness-related study confounds, these trials are unlikely to capture a sufficient range of neuropsychiatric phenotypes. This coincides with findings from a recent meta-analysis of probiotic trials in relation to depression symptoms ( Ng et al., 2018 ). Additional neuropsychiatric case-control trials with multiple treatment outcome and compliance measures are necessary to confirm prior reports and conduct more rigorous meta-analyses of probiotics in relation to specific symptoms. The total number of studies conducted with quail, zebrafish, and specific rat and mouse stains are very limited for a comprehensive review. The reporting of the experimental microbial environment (i.e. specific-pathogen-free, germ-free, etc.) for animal trials was inconsistent and should be improved. Significant limitations common to all probiotic trials include reporting bias, sample size, confirmation of probiotic activity and administration vehicle, precise probiotic dosage and treatment duration, differential neuropsychiatric assessments, and an insufficient number of trials for early and late neurodevelopmental stages across the lifespan. Assessment of study quality and bias had some effectiveness for capturing dropout rates and quality variation. Approximately 59% percent of the human trials reported less than 15% study subject dropout. Intent-to-treat analyses were reported for 45.8% of the human trials. Measures of probiotic treatment adherence (i.e. intake of 75% or greater doses) and exclusion criteria for poor compliance were reported for 29.2% of the human trials. Specific blinding status was reported for 93.1% and 43.1% human and non-human studies, respectively. Although we aimed to minimize bias when evaluating all non-human trials, excessive positive finding reporting for animal studies has been acknowledged ( Sena et al., 2010 ; Tsilidis et al., 2013 ) and is not easily illuminated with existing study quality and bias assessments. These are critical factors that need to be addressed to improve study quality and develop successful treatments. The sample sizes for approximately 50% of the human and 90% of the non-human trials were fewer than thirty and fifteen subjects per group, respectively. In addition, studies reporting a priori designations of primary and secondary outcomes customary for rigorous clinical trial designs or sample size estimations to reflect sufficient study power were limited across species. This implies most of the trials evaluated would be considered exploratory investigations. Therefore, larger, adequately powered, replication studies are required to confirm observations from individual trials and those compiled in this systematic review. Verification of probiotic strain/substrain activity (i.e. in vitro culture) prior to conducting the investigation or post-probiotic treatment fecal sample sequencing was inconsistent across trials. In addition, probiotic intake vehicle varied across trials (i.e. capsule, in yogurt, in water, per oral, etc.). The combination of these factors can lead to significant experimental confounds and thereby influence the validity and reliability of our systematic review observations. Although the exact mechanisms by which probiotics proliferate within the intestinal tract are unclear, administration route may influence successful probiotic colonization. Therefore, future studies may consider confirmatory procedures such as fecal (human and non-human) or intestinal biopsy (non-human) sequencing. Three human, two rat, and one mouse trial investigated varying “doses” (i.e. colony forming units-CFU) for the same probiotic strain/substrain in relation to neuropsychiatric outcomes. However, more extensive dose-finding experiments will need to be conducted across species, probiotic strains/substrains, and study populations prior to large-scale implementation of probiotics for gut-brain-behavior based outcomes. Treatment duration for most probiotic trials reviewed was less than sixty days with a limited number of trials assessing long-term neuropsychiatric outcomes. A pilot adjunctive probiotic trial in treatment-resistant depressed patients reported the efficacy of an L. acidophilus, B. bifidum, S. thermophiles (2 × 10 10 CFU) plus 1600 mg magnesium orotate therapeutic and demonstrated symptom reemergence upon treatment cessation ( Bambling et al., 2017 ). While the study was ineligible for inclusion in this report based on our systematic review criteria; their observations highlight the need for developing longitudinal probiotic investigations and utilizing probiotic treatments as a long-term health and wellness lifestyle modification, rather than a short-term intervention. Thus far, two double-blind investigations have been conducted with patients diagnosed with significant anxiety or depression symptoms ( Akkasheh et al., 2016 ; Romijn et al., 2017 ). Investigations of probiotics in patients with prominent mania or psychoses are also limited ( Dickerson et al., 2018 ; Dickerson et al., 2014 ). While we aimed to distinguish neuropsychiatric outcomes by specific symptoms in the Results section of this review, the sizeable overlap in neuropsychiatric phenotypes, especially anxiety and depression ( Sartorius et al., 1996 ), are well-recognized. Most of the human and non-human probiotic studies utilized varying stress or depression-based assessments. In addition, limited assessments or assays in relation to neuropsychiatric symptoms were conducted within most trials. Standardized neurobehavioral measures and novel symptom models should be developed and incorporated into future trials. To differentiate subsyndromes, future trials could employ multiple experimental strategies (e.g. psychophysical task plus anxiety assessment interview in human studies). Across species, trials were primarily conducted to coincide with the middle-age adult stage of the lifespan. While this review aimed to highlight the limited studies that enrolled older adults, young/middle-aged adults, and children; early- (12-14 years) and middle- (15-17 years) adolescent stage human or non-human animal trials were not reported in the neuropsychiatric literature. Gut-brain-pathways are highly likely to be modified by gut microbes and probiotics during these critical periods of the lifespan ( McVey Neufeld et al., 2016 ) which should be underscored for future probiotic and gut-brain-behavior studies. Although neuropsychiatric illnesses are largely characterized by significant cognitive and behavioral alterations; sociodemographic and lifestyle factors are prominent and may modify the effects of probiotics on anxiety, mood, and psychotic disorders. There are preliminary indications that probiotic treatment efficacy may vary by participant sex ( Sanchez et al., 2017 ). Other sociodemographic differences were not reported by the studies reviewed. Non-probiotic neuropsychiatric trials have reported differential treatment response by race or ethnicity ( Ellis et al., 2015 ). Therefore, future studies may need to consider variation in probiotic treatment efficacy by subject race/ethnicity or animal strain. Our review indicates that several human and non-human probiotic treatment trials have been conducted with direct or indirect intent to target gut-brain-behavior interactions. However, the large variance in experimental design, sample characteristics, and assessment methodology hinders true comparison and computation of effect sizes for specific phenotypes and probiotic strain or substrain combinations. Although a wide-range of mediating biomarkers and neuropsychiatric assessments can have significant utility as secondary or tertiary outcome analyses, it is important for future human and non-human investigations to improve study power and minimize risk of bias by clearly delineating all post hoc analyses and/or incorporating as many of these variables a priori when computing sample size estimates. Improving the reporting of trial methodology with these crucial details will substantially improve the quality of subsequent systematic reviews and meta-analyses. Future studies will need to replicate existing findings and develop additional randomized, placebo-controlled, double-blinded, and case-controlled, or cohort probiotic trials. Since probiotic trials are unable to clarify all gut-microbe based treatment mechanisms, large-scale normative data obtained from various human cohorts and non-human animal species and strains are necessary to obtain a more veridical representation of gut microbial composition and variation across the lifespan. Concurrent investigations with special emphasis on neurodevelopmental and neuropsychiatric illness trajectory will be especially valuable. Data obtained from these combined sources will better inform future probiotic treatment study design and hypothesized neuropsychiatric targets. In brief, the ability to 1) elucidate gut microbe-brain-behavior pathway mechanisms; 2) disentangle unique effects of single- and multi-strain formulations of probiotics, and; 3) implement novel probiotic treatments to target precise neuropsychiatric phenotypes, will require comprehensive review and meta-analyses of future probiotic trial outcomes.

Neuropsychiatric therapies are heterogeneous in their long-term efficacy ( Geddes et al., 2000 ; McEvoy and Nathan, 2007 ; Serretti and Mandelli, 2010 ), with some having considerable harmful effects ( Correll et al., 2009 ; De Hert et al., 2012 ; Lozano et al., 2008 ). Consequently, the biomedical research community has placed a great emphasis on developing novel treatments that target more objectively quantifiable brain and peripheral biomarkers ( Insel, 2014 ; Niciu et al., 2014 ; Sanislow et al., 2015 ). Across organ systems, an increasing number of studies continue to highlight the importance of connections among the brain, mind, and body ( Gallagher, 2004 ; Gold and Charney, 2002 ; Jones et al., 2006 ; Muehsam et al., 2017 ). Therefore, scientists and clinicians are designing studies to expand our knowledge of the bidirectional signaling mechanisms between gut microbes and the brain and their subsequent influence on behavior and mental health for potential neuropsychiatric treatment development ( Mayer et al., 2014 ). The precise relationship between dysbiosis (i.e. altered gut microbial composition) and neuropsychiatric symptoms in various cohorts is unclear and continues to be investigated. Initial studies in people diagnosed with Major Depressive Disorder ( Aizawa et al., 2016 ; Jiang et al., 2015 ) indicate a relative increase in Bacteroidetes, Proteobacteria, and Actinobacteria phyla coinciding with a decline in Firmicutes (including Lactobacillus and Bifidobacterium ). A relative decrease in Actinobacteria, Lentisphaerae, and Verrucomicrobia phyla has been associated with neuropsychiatric symptom severity in a sample of individuals with Post-Traumatic Stress Disorder ( Hemmings et al., 2017 ). In a preliminary study in people with schizophrenia, a relative reduction in Proteobacteria ( Haemophilus, Sutterella , and Clostridium ), with a concurrent increase in Firmicutes ( Anaerococcus ) has been observed ( Nguyen et al., 2018 ). This report also indicates negative symptoms may be uniquely linked to Firmicutes ( Ruminococcaceae ) colonization whereas current depression could vary by Bacteroidetes ( Bacteroides ) frequency. Preliminary animal studies indicate that gut microbial alteration can influence a wide range of neurobehavioral phenotypes across the developmental trajectory ( Bruce-Keller et al., 2015 ; Clarke et al., 2013 ; Park et al., 2013 ; Pyndt Jorgensen et al., 2015 ). One such study demonstrated that ampicillin was successful at restoring phencyclidine-induced gut microbe-cognitive dysregulation ( Pyndt Jorgensen et al., 2015 ). However, there is a heightened awareness of the significant adverse effects antibiotics such as ampicillin can have on gut microbial diversity and physiological function ( Dethlefsen et al., 2008 ; Dethlefsen and Relman, 2011 ), limiting its neuropsychiatric treatment utility. Moreover, the long-term beneficial versus detrimental effects of antibiotics on gut-brain-behavior interactions have not yet been characterized. This has accelerated the pursuit of probiotics as a potential neuropsychiatric intervention. Probiotics are live microorganisms that confer health benefits to its host ( Hill et al., 2014 ) with a minimal incidence of adverse effects ( Marteau and Shanahan, 2003 ). Probiotic consumption to maintain gut and overall health has now been implemented in routine medical practice ( Gareau et al., 2010 ). Probiotics have shown significant promise for improving atopic dermatitis, necrotizing enterocolitis, pouchitis, and irritable bowel syndrome - IBS ( Sanders et al., 2013 ), which is postulated to be a gut-brain dysregulation disorder ( Blankstein et al., 2010 ; Kennedy et al., 2014 ; Kennedy et al., 2012 ). However, probiotic trials designed to target stress levels, mood, cognitive, or psychosocial functioning, have only been conducted during the past ten to fifteen years. In addition, the utility of probiotic treatments to improve neuropsychiatric symptoms have not been established. Moreover, while multi-strain probiotic combinations may provide greater health benefits in comparison to single-strains of probiotics for several infections and gastrointestinal disorders ( Chapman et al., 2011 ; Timmerman et al., 2004 ), this has not yet been evaluated for human or non-human neuropsychiatric outcome trials. This review examined published probiotic trials conducted across animal species to determine if single- or multi-strain formulations of probiotics have differential effects in modifying neuropsychiatric symptoms or phenotypes.