Hostname: page-component-cd9895bd7-jkksz Total loading time: 0 Render date: 2024-12-24T12:28:26.425Z Has data issue: false hasContentIssue false

The impact of botanical fermented foods on metabolic syndrome and type 2 diabetes: a systematic review of randomised controlled trials

Published online by Cambridge University Press:  26 October 2023

Miin Chan
Affiliation:
School of Agriculture, Food and Ecosystem Sciences, University of Melbourne, Melbourne, VIC, Australia
Nadja Larsen
Affiliation:
Department of Food Science, University of Copenhagen, Frederiksberg, Denmark
Helen Baxter
Affiliation:
Austin Health Science Library, Austin Health, Heidelberg, VIC, Australia
Lene Jespersen
Affiliation:
Department of Food Science, University of Copenhagen, Frederiksberg, Denmark
Elif I. Ekinci
Affiliation:
The Australian Centre for Accelerating Diabetes Innovations (ACADI), Melbourne Medical School, The University of Melbourne and Department of Endocrinology, Austin Health, Heidelberg, VIC, Australia
Kate Howell*
Affiliation:
School of Agriculture, Food and Ecosystem Sciences, University of Melbourne, Melbourne, VIC, Australia
*
*Corresponding author: Kate Howell, email: [email protected]
Rights & Permissions [Opens in a new window]

Abstract

Our systematic review assessed the impact of botanical fermented food (BFF) consumption on glucose, lipid, anthropometric, inflammatory and gut microbiota parameters, in adults with metabolic syndrome (MetS), MetS components or type 2 diabetes mellitus (T2DM). Embase, MEDLINE, Cochrane CENTRAL and Google Scholar were searched with no language limits, from inception to 31 August 2022, for eligible randomised controlled trials (RCTs). Two independent reviewers screened 6873 abstracts and extracted relevant data. Risk of bias (ROB) was assessed using the Cochrane Collaboration’s ROB2 tool. The final review included twenty-six RCTs, with thirty-one reports published between 2001 and 2022. Significant (p < 0·05) within-group and between-group changes in cardiometabolic outcome means were reported in twenty-three and nineteen studies, respectively. Gut microbiota composition was assessed in four studies, with two finding significant between-group differences. No significant difference between groups of any measured outcomes was observed in five studies. There were fourteen studies at low ROB; ten were of some concern; and two were at high ROB. In 73% of included studies, BFF consumption by participants with obesity, MetS or T2DM led to significant between-group improvements in discrete cardiometabolic outcomes, including fasting blood glucose, lipid profile, blood pressure, waist circumference, body fat percentage and C-reactive protein. BFF consumption increased the abundance of beneficial gut bacteria, such as Bifidobacterium and LAB, whilst reducing potential pathogens such as Bacteroides. To determine the clinical significance of BFFs as therapeutic dietary adjuncts, their safety, tolerability and affordability must be balanced with the limited power and magnitude of these preliminary findings.

Type
Review Article
Creative Commons
Creative Common License - CCCreative Common License - BY
This is an Open Access article, distributed under the terms of the Creative Commons Attribution licence (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted re-use, distribution and reproduction, provided the original article is properly cited.
Copyright
© The Author(s), 2023. Published by Cambridge University Press on behalf of The Nutrition Society

Introduction

Metabolic syndrome (MetS) is an increasingly prevalent constellation of cardiometabolic derangements. Signified by central obesity, individuals with MetS suffer from atherogenic dyslipidaemia (raised triacylglycerols, lowered HDL-C), hypertension and impaired fasting glucose/prediabetes(Reference Alberti, Zimmet and Shaw1,Reference Grundy2) . These interrelated risk factors are associated with an increased likelihood of developing type 2 diabetes mellitus (T2DM) and cardiovascular disease(Reference Alberti, Zimmet and Shaw1,Reference Alberti, Eckel and Grundy3) . Gut microbiota appears to be important in the pathogenesis of metabolic disorders(Reference Festi, Schiumerini and Eusebi4); adults with obesity and T2DM display markedly reduced gut microbiota diversity and low bacterial gene richness(Reference Ley, Turnbaugh and Klein5Reference Le Chatelier, Nielsen and Qin8). Diet regulates gut microbiota composition and function(Reference Muegge, Kuczynski and Knights9,Reference David, Maurice and Carmody10) , and is a well-established modifiable factor in the management of MetS and T2DM(Reference Fabiani, Naldini and Chiavarini11). Globally, MetS affects >25% of all adults(Reference O’Neill and O’Driscoll12). Projections indicate that 1 billion adults will develop obesity by 2030(13), with 700 million adults developing T2DM by 2045(Reference Saeedi, Petersohn and Salpea14). As such, dietary modulation of the gut microbiota may be a cost-effective approach to reducing the global health burden of obesity, MetS and T2DM.

Fermented foods are microbially transformed foods traditionally consumed worldwide. Recent studies suggest that these foods, whether dairy or plant-based, deliver health benefits through transient integration of food-associated live micro-organisms into gut commensal communities(Reference Zhang, Derrien and Levenez15); microbial enzymatic substrate transformation in the intestinal lumen(Reference Marco, Heeney and Binda16); the release of bioactive compounds including those with insulinotropic or immune-regulatory effects(Reference Derrien and van Hylckama Vlieg17); or the bacterial biosynthesis of vitamins(Reference Gille, Schmid and Walther18). Recent multiomic approaches have found that fermented foods may be an important source of commensal lactic acid bacteria (LAB) in the gut microbiome(Reference Pasolli, De Filippis and Mauriello19), and that consumption of large amounts of fermented foods leads to subtle, persistent differences in human gut microbiota composition and faecal metabolome(Reference Taylor, Lejzerowicz and Poirel20). In a recent study, diets which were high in fermented food increased microbiota diversity and reduced inflammatory markers compared with plant-based diets with twice the fibre and no fermented foods(Reference Wastyk, Fragiadakis and Perelman21). Abundant clinical studies investigating the effect of fermented dairy products on cardiometabolic health have been systematically reviewed and meta-analysed(Reference Gille, Schmid and Walther18), with mixed results. Overall, fermented milk consumption is associated with a reduced risk of T2DM and may assist with weight maintenance and obesity(Reference Barengolts22Reference Savaiano and Hutkins24). However, less well studied are botanical fermented foods (BFFs), which may be even more effective at exerting health effects than their dairy-based counterparts(Reference Nguyen25).

BFFs are globally consumed plant-based foods and beverages(Reference Marco, Heeney and Binda16,Reference Hutkins26) , produced through the fermentation of vegetables, fruit, cereals, nuts and pulses. Traditional BFFs include kimchi, sauerkraut, water kefir and tempeh. Compared with dairy-based fermented foods, BFFs may contain more diverse microbial communities, with more microbial genes (gastrointestinal survival, gut colonisation, immune modulation) associated with potential health benefits to the host(Reference Leech, Cabrera-Rubio and Walsh27,Reference Lavefve, Cureau and Rodhouse28) . BFFs are also lower in fat; are cholesterol free; and contain higher levels of polyphenols, other antioxidants and microbiota-associated carbohydrates, including dietary fibre(Reference Panghal, Janghu and Virkar29,Reference Rodzi and Lee30) . They are also better tolerated and more palatable for many consumers(Reference Luckow and Delahunty31), especially as 75% of the world’s population is lactose intolerant(Reference Silanikove, Leitner and Merin32). For these reasons, BFFs are potentially more suitable than fermented dairy products for the prevention and management of MetS, its components and T2DM.

Recent critical reviews of fermented foods have stated that more human studies are required to establish the role of BFFs as interventions for noncommunicable diseases(Reference Gille, Schmid and Walther18,Reference Diez-Ozaeta and Astiazaran33,Reference Rul, Béra-Maillet and Champomier-Vergès34) . However, as these reviews did not use systematic review methodology, many existing clinical trials of BFFs were not identified. On the other hand, we found several up-to-date systematic reviews/meta-analyses of randomised controlled trials (RCTs) investigating the impact of red yeast rice/monacolin/xuezhikang(Reference Xiong, Wang and Li35Reference Yuan, Yuan and Wang39) or vinegar(Reference Siddiqui, Assam and de Souza40Reference Cherta-Murillo, Pugh and Alaraj-Alshehhi45) on metabolic parameters; these RCTs were thus excluded during abstract screening to prevent the waste of resources(Reference Bhide and Acharya46). We have undertaken a systematic review of RCTs administering other BFFs to participants with T2DM, MetS or any MetS components. Our review asks: ‘Does the consumption of BFFs have any impact on the gut microbiota, T2DM or MetS outcomes?’.

Methods

Search strategy

We performed a qualitative systematic review of randomised controlled trials (RCTs) conducted in adults with type 2 diabetes mellitus (T2DM) or metabolic syndrome (MetS) components or any combination of these components (Supplementary Material 1). This systematic review was registered at the International Prospective Register of Systematic Reviews (PROSPERO) website on 18 December 2018 (CRD42018117766). The detailed methods of this review were reported in a protocol paper(Reference Chan, Baxter and Larsen47). Embase, MEDLINE, Cochrane CENTRAL and Google Scholar (first 200 relevancy ranked results) were searched from inception to 30 August 2022. Our search strategy (Supplementary Material 2) combined subject heading terms and text words for BFF (e.g. fermented food, fermentation), and MetS or T2DM (e.g. MetS, obesity, hypertension, blood pressure, diabetes, prediabetes, hyperlipidaemia, microbiota, dysbiosis, inflammation). To retrieve RCTs, the Cochrane Highly Sensitive Search Strategy for MEDLINE(Reference Dickersin, Scherer and Lefebvre48) was used. No language limits were applied. Reference lists in identified articles were searched with Scopus. We also searched grey literature via trial registries and conference papers. When a study had unreported data, authors were contacted for further information.

Selection criteria and data management

Reviewers M.C. and N.L. independently selected and extracted data from eligible publications. Selection was based on the PICOS criteria(Reference Higgins and Thomas49), as shown in Table 1. The following data were extracted: first author’s name, publication year and study location; study design; BFF and comparator type, dosage and duration; subject characteristics (sample size, population, age, gender, condition); metabolic and gut microbiota parameters (mean ± standard deviation) in each group before and after the intervention. Study populations described in more than one published article were counted as single studies, with relevant data extracted from all articles. Any discrepancies were resolved through discussion. Authors of trials were contacted for clarification when necessary. All processes and data were recorded using Covidence software (www.covidence.org).

Table 1. PICOS (population, intervention, comparison, outcome, study design) criteria for inclusion of studies

T2DM, type 2 diabetes mellitus; MetS, metabolic syndrome; BP, blood pressure; GM, gut microbiota; QOL, quality of life.

Primary outcomes were changes in any T2DM and MetS clinical parameters. Anthropometric measures were body weight (BW), body mass index (BMI), waist circumference (WC), waist–hip ratio (WHR), body fat percentage (BFP), total fat area (TFA), visceral fat area (VFA), subcutaneous fat area (SFA), systolic and diastolic blood pressure (SBP, DBP). Lipid profile consisted of total cholesterol (TC), triacylglycerols (TG), high-density lipoprotein (HDL-C) and low-density lipoprotein cholesterol (LDL-C), and free fatty acids (FFA). Glucose metabolism was measured via fasting blood glucose and insulin (FBG, FBI), 2-h postprandial glucose and insulin (2hPPG, 2hPPI), haemoglobin A1c (HbA1c), Homeostatic Model Assessment for Insulin Resistance (HOMA-IR) and C-peptide (C-pep). Secondary outcomes were shifts in: inflammatory markers interleukin-6 (IL-6), interleukin-10 (IL-10), tumour necrosis factor alpha (TNF-α), C-reactive protein (CRP) and high sensitivity-CRP (hs-CRP); obesity markers apolipoproteins A1 (ApoA1) and B (ApoB), and ApoB/A1; gut microbiota composition/function (α-diversity, β-diversity, relative abundance, correlations with cardiometabolic parameters and gene expression, short-chain fatty acid production); and liver function tests.

Risk-of-bias assessment

All included studies were independently qualified by M.C. and N.L. using the Cochrane Collaboration’s Risk of Bias 2 (ROB2) tool(Reference Higgins, Altman and Gotzsche50) within Covidence. Each study was assessed as having a low risk of bias, some concerns, or a high risk of bias.

Data synthesis strategy

We deemed a meta-analysis inappropriate due to between-study clinical heterogeneity in participant groups, diagnostic cut-off points, interventions, comparators and outcomes measured. We presented a narrative synthesis, organised according to fermented food/micro-organism types: (1) lactofermented foods, produced through the fermentative action of LAB, such as kimchi and ash kardeh; (2) jangs, such as kochujang, doenjang and chungkookjang, produced by fermenting soybeans with Bacillus and Aspergillus spp.; (3) tempeh, produced by fermenting soybeans with Rhizopus spp.; (4) Aspergillus oryzae-fermented products, such as shiokoji, amazake and miso; and (5) others. For more detail on the production of these fermented foods, please see Supplementary Material 3. Results were then described according to relevant biochemical and clinical parameters (lipid, glucose, anthropometric, inflammatory and gut microbiota parameters), as well as populations studied (overweight/obese; impaired fasting glucose/prediabetic/T2DM; hypertension; hyperlipidaemia; and MetS/cardiovascular risk factors).

Patient and public involvement

No patients involved.

Results

Our search yielded a total of 6873 citations. Following removal of duplicates, abstract screening and full-text review, thirty-one published reports of twenty-six RCTs met our inclusion criteria. Relevant data were extracted from all reports and combined for the same study when duplicates were present. The PRISMA literature search(Reference Page, McKenzie and Bossuyt51) results are represented in Figure 1; the PRISMA checklist was also completed (Supplementary Material 4).

Fig. 1. PRISMA flow diagram of literature search.

Assessment of risk of bias

Using the Revised Cochrane risk-of-bias tool (ROB 2), we assessed all included studies for risk of bias. The results of judgments by two investigators, according to each of the tool’s five domains, are summarised in Figures 2 and 3, produced using the Robvis tool(Reference McGuinness and Higgins52) Bias due to randomisation was high risk in 8% of studies, of some concern in 31%, with 61% at low risk. The studies showed a low risk of bias in the following domains: deviations from intended interventions (100%), missing outcome data (92%), outcome measurement (100%) and selection of reported result (92%). Overall risk of bias was low in 54% of studies, of some concern in 38% and high in 8%.

Fig. 2. Risk-of-bias summary: domain-level judgements for each individual study.

Fig. 3. Risk-of-bias graph: weighted distribution of risk-of-bias judgements within each bias domain across all included studies.

Risk of bias during the randomisation process was low in all but ten studies: eight were of some concern(Reference Kim, An and Lee53Reference Nadia, Wati and Isnawati60), and two were at high risk(Reference Somanah, Aruoma and Gunness61,Reference Akamine, Millman and Uema62) . These ten studies did not describe randomisation or allocation concealment methods in enough detail. All studies had almost equal baseline numbers in each intervention group, with the exception of one study(Reference Somanah, Aruoma and Gunness61) which did not use block randomisation and had substantial differences in group size allocation (intervention group n = 49; control n = 78). Akamine et al. (Reference Akamine, Millman and Uema62) reported significant baseline imbalances in FBG levels (brown rice amazake 5·6 ± 0·2 versus white rice amazake control 6·3 ± 0·2 mmol/L; p = 0·018); the number of subjects with T2DM (brown rice amazake n = 3 versus control n = 10; p = 0·017); and, gut microbiota composition at genus and species levels (Bacteroides intestinalis, Faecalibacterium spp. DJF VR20, Sutterella wadsworthensis, Parabacteroides distasonis and Alistipes onderdonkii were significantly higher in the brown rice amazake group; p < 0·05). Kim et al. (Reference Kim, An and Lee53), Somanah et al.(Reference Somanah, Aruoma and Gunness61) and An et al. (Reference An, Lee and Jeon55) did not provide sufficient information on baseline imbalances between groups, as well as An et al.(Reference An, Lee and Choi54) and Lee et al.(Reference Lee, An and Jeon56), which were abstracts only. Han et al.(Reference Han, Bose and Wang57) stated that ‘CRP and DBP are significantly different between groups’. All other studies showed no baseline imbalances between groups.

No studies reported toxicities or serious adverse events, likely due to the interventions being foods or food based. As such, although there were eleven open-label studies(Reference Kim, An and Lee53Reference Salehi, Karimpour and Imani63), no studies were at risk of bias due to deviation from intended interventions.

No dropouts were reported in four studies(Reference Kim, An and Lee53,Reference Nadia, Wati and Isnawati60,Reference Kim, Kim and Lee64,Reference Shin, Kwon and Jeong65) . Akamine et al.(Reference Akamine, Millman and Uema62) also reported no dropouts, but only analysed 88% of participants for plasma SCFA, and 85% of participants for faecal microbiota composition, due to missing or uncollected samples. Analysis of over 90% of randomised participants (91–96%) was conducted in ten studies(Reference An, Lee and Jeon55,Reference Han, Bose and Wang57Reference Afifah, Nabilah and Supraba59,Reference Salehi, Karimpour and Imani63,Reference Fujita, Yamagami and Ohshima66Reference Han, Jeong and Ryu70) ; all of these studies used per-protocol analysis. Between 80% and 90% of participant data was analysed in five studies(Reference Cha, Yang and Back71Reference Mohammadi, Mazloomi and Niakousari75), with only Cha et al.(Reference Cha, Kim and Yang72) and Oh et al.(Reference Oh, Park and Kim73) using intention-to-treat analysis for missing data. Tenore et al.(Reference Tenore, Caruso and Buonomo76), with only 69% of participants completing the study, used a ‘negative binomial, generalised linear mixed effects model with a per-protocol set’ to account for missing data. The remaining studies included data analysis for 79%(Reference Somanah, Aruoma and Gunness61), 69%(Reference Byun, Yu and Cha77) and 61%(Reference Nakamura, Kitagawa and Yamamoto78) of participants. Somanah et al.’s(Reference Somanah, Aruoma and Gunness61) study was high risk in the domain of missing outcome data as there were large numbers of dropouts, combined with unexplained inconsistencies in the number of participants analysed for each outcome. Nakamura et al.(Reference Nakamura, Kitagawa and Yamamoto78) provided primary analysis data (shiokoji n = 23; placebo n = 24) as per their original inclusion criteria, finding no significant differences between groups in glucose parameters. On completion of the study, investigators excluded participants who had significant fluctuations in physical activity or alcohol intake during the trial (seven from each group), and/or those with high insulin resistance (HOMA-IR >5·0) (two from placebo group HOMA-IR >5·0; one from placebo group with both factors). Secondary post-hoc analysis using the same pre-specified statistical methods was performed, resulting in a significant finding. Although both analysis sets were provided, reasons were documented and limitations were stated, it is difficult to ascertain if these decisions were made to elicit a positive result. As such, this study was assessed as having some concerns due to missing outcome data and selective reporting. The study by Han et al.(Reference Han, Jeong and Ryu70) was found to be of some concern regarding selective non-reporting, as it did not present any between-group analysis data; we were unable to ascertain, despite inquiries, whether analysis was performed, or if no significant differences were found and therefore unreported. In all other studies with dropouts, the authors documented the reasons for missing outcome data, which were all unrelated to the outcomes (e.g. noncompliance, personal reasons). None of the other studies used selective reporting, and all studies had a low risk of bias in measurement of outcomes.

Overall, we judged fourteen of the included RCTs to have a low risk of bias(Reference Salehi, Karimpour and Imani63Reference Jung, Haddad and Kaur69,Reference Cha, Yang and Back71Reference Byun, Yu and Cha77) . Eight of the ten studies assessed to be of some concern(Reference Kim, An and Lee53Reference Nadia, Wati and Isnawati60) would likely be upgraded to low risk if more information regarding randomisation and allocation concealment was provided. Han et al.(Reference Han, Jeong and Ryu70) likely engaged in selective non-reporting of between-group analysis; more information from authors may change the level of risk assigned. Nakamura et al.(Reference Nakamura, Kitagawa and Yamamoto78) would be deemed low risk if the primary analysis is used. Two studies were at high risk of bias: Somanah et al.(Reference Somanah, Aruoma and Gunness61) had a high risk of bias in the domains of randomisation and incomplete outcome data; Akamine et al.(Reference Akamine, Millman and Uema62) had a high risk of selection bias as they did not conceal allocation, used ‘pseudo-randomisation’, and had a significant baseline imbalance in FBG levels, with unequal group numbers (19 versus 21). In future studies, to reduce the risk of bias, intention-to-treat analysis should be utilised to account for missing data, and more details regarding randomisation, allocation concealment and baseline imbalances need to be included.

Study characteristics

For the full characteristics of included studies, please see Supplementary Material 5. Tables 26 summarise the characteristics of included RCTs, organised by BFF subgroup. RCTs by An et al.(Reference An, Lee and Choi54) and Lee et al.(Reference Lee, An and Jeon56) were described only in abstracts. There were seven crossover studies(Reference Kim, An and Lee53Reference Lee, An and Jeon56,Reference Jung, Haddad and Kaur69,Reference Mohammadi, Mazloomi and Niakousari75,Reference Byun, Yu and Cha77) , with the rest being of parallel design; fifteen were double-blinded or more(Reference Kim, Kim and Lee64Reference Nakamura, Kitagawa and Yamamoto78), whilst eleven were open-label studies. Women only were recruited in three studies(Reference Han, Bose and Wang57,Reference Afifah, Nabilah and Supraba59,Reference Nadia, Wati and Isnawati60) ; all other studies included both genders. Participants were between the ages of 19(Reference Byun, Yu and Cha77) and 65(Reference Fujita, Yamagami and Ohshima66). The sample sizes were between 16(Reference Lee, An and Jeon56) and 127(Reference Somanah, Aruoma and Gunness61), with intervention periods ranging from 2(Reference Afifah, Nabilah and Supraba59) to 14(Reference Somanah, Aruoma and Gunness61) weeks, with variable BFF doses, and no statement of microbial counts, except in two studies(Reference Kim, An and Lee53,Reference Tenore, Caruso and Buonomo76) . A variety of BFFs fermented with different micro-organisms, from LAB, Bacillus, Aspergillus and/or Rhizopus spp., were used as interventions. Comparators included placebos, no intervention and active controls. Some studies provided meals to both groups, and dietary intake data were collected and analysed across all studies. The populations investigated in the studies originated from South Korea (n = 14), Japan (n = 4), Indonesia (n = 3), Iran (n = 2), USA (n = 1), Italy (n = 1) and Mauritius (n = 1).

Table 2. Characteristics of included studies – lactofermented foods

n, number; R/C, randomised/completed; M/F, male/female; m(SD), mean ± standard deviation; p/day, per day; BMI, body mass index; CO, crossover; 4/2/4, 4-week interventions with 2-week wash-out period in between; Pl, placebo; FBG, fasting blood glucose; BFP, body fat percentage; SBP, systolic blood pressure; DBP, diastolic blood pressure; FBI, fasting blood insulin; WHR, waist-to-hip ratio; TC, total cholesterol; BW, body weight; T2DM, type 2 diabetes mellitus; NS, not stated; NSD, no significant difference; HbA1c, haemoglobin A1c; 8/4/8, 8-week interventions with 4-week wash-out period in between; MetS, metabolic syndrome; HOMA-IR, Homeostatic Model Assessment of Insulin Resistance; WC, waist circumference; CRP, C-reactive protein; IFG, impaired fasting glucose; DB, double blinded; 2hPPG, 2-h postprandial glucose; 2hPPI, 2-h postprandial insulin; HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; CVD, cardiovascular disease; LAB, lactic acid bacteria; L. plantarum, Lactiplantibacillus plantarum; RPG, random plasma glucose; Cntrl, control; TG, triacylglycerols.

a Prediabetes with MetS: ⩾3: TG ⩾1·7; HDL-C M ⩽1·0, F ⩽1·3; FBG ⩾5·6 (mmol/L); BP ⩾130/85 mmHg; WC M ⩾90 cm, F ⩾80 cm.

b TC 5·2–6·7; HDL-C 0·8–1·2; LDL-C 3·9–4·7; TG 1·9–3·2; FBG 5–6·9 (mmol/L).

*p < 0·05

**p < 0·01

***p < 0·001.

Table 3. Characteristics of included studies – Chungkookjang, kochujang and doenjang (Bacillus and Aspergillus species)

n, number; R/C, randomised/completed; M/F, male/female; m(SD), mean ± standard deviation; p/day, per day; BMI, body mass index; WHR, waist-to-hip ratio; DB, double blinded; Pl, placebo; ApoB, apolipoprotein B; IFG, impaired fasting glucose; FBG, fasting blood glucose; TC, total cholesterol; LDL-C, low-density lipoprotein cholesterol; ApoA1, apolipoprotein A1; BW, body weight; BFM, body fat mass; VFA, visceral fat area; BFP, body fat percentage; TFA, total fat mass; SFA, subcutaneous fat area; TG, triacylglycerols; WC, waist circumference; CO, crossover; 12/12/12, 12-week interventions with 12-week wash-out period in between; HDL-C, high-density lipoprotein cholesterol.

*p < 0.05

**p < 0.001

***p < 0.001.

Table 4. Characteristics of included studies – Shiokoji, miso, touchi, amazake and kochujang (Aspergillus oryzae)

n, number; R/C, randomised/completed; M/F, male/female; m(SD), mean ± standard deviation; p/day, per day; T2DM, type 2 diabetes mellitus; FBG, fasting blood glucose; HbA1c, haemoglobin A1c; DB, double blinded; LDL-C, low-density lipoprotein cholesterol; TC, total cholesterol; BP, blood pressure; TG, triacylglycerols; H/T, hypertension; SBP, systolic blood pressure; DBP, diastolic blood pressure; BW, body weight; NSD, no significant difference; MetS, metabolic syndrome; SCFA, short-chain fatty acids; BMI, body mass index; HDL-C, high-density lipoprotein cholesterol; spp., species.

a High-normal BP (130–139/85–89 mmHg) to untreated stage I H/T (140–159/90–99 mmHg);

b BMI ⩾25 kg/m² and any 2 of: TG ⩾1.7 mmol/L; HDL-C ⩽1.0 mmol/L; SBP ⩾130 mmHg +/− DBP ⩾85 mmHg; or FBG ⩾5.6 mmol/L.

*p < 0.05

**p < 0.001

***p < 0.001.

Table 5. Characteristics of included studies – tempeh products (Rhizopus species)

n, number; R/C, randomised/completed; M/F, male/female; m(SD), mean ± standard deviation; p/day, per day; hyperchol, hypercholesterolaemia; TC, total cholesterol; LDL-C, low-density lipoprotein cholesterol; HDL-C, high-density lipoprotein cholesterol; BMI, body mass index; BW, body weight; FBG, fasting blood glucose; HOMA-IR, Homeostatic Model Assessment of Insulin Resistance; FBI, fasting blood insulin; hsCRP, high sensitivity C-reactive protein; TG, triacylglycerols.

*p < 0.05

**p < 0.001

***p < 0.001.

Table 6. Characteristics of included studies – others

n, number; R/C, randomised/completed; M/F, male/female; m(SD), mean ± standard deviation; p/day, per day; FBG, fasting blood glucose; OGTT, oral glucose tolerance test; DB, double blinded; TC, total cholesterol; LDL-C, low-density lipoprotein cholesterol; WC, waist circumference; AST, aspartate aminotransferase; ALT, alanine transaminase; HbA1c, haemoglobin A1c; SBP, systolic blood pressure; CRP, C-reactive protein; TG, triacylglycerols; BMI, body mass index; DBP, diastolic blood pressure; HDL-C, high density lipoprotein cholesterol; CVD, cardiovascular disease; ; 12/12/12, 12-week interventions with 12-week wash-out period in between; ApoA1, apolipoprotein A1; ApoB, apolipoprotein B; CO, crossover; H/T, hypertension; 6/3/6, 6-week interventions with 3-week wash-out period in between; NSD, no significant difference.

a CVD risk biomarkers: ⩾2: smoking; SBP ⩾140/90 mmHg; LDL-C ⩾2.9 mmol/L; HDL-C ⩾1.0 mmol/L; TG ⩾1.7 mmol/L; FBG ⩾6.1 mmol/L; BMI ⩾25 kg/m²; family history of heart disease

*p < 0.05

**p < 0.001

***p < 0.001.

Study findings

The studies measured some combination of the following relevant outcomes: blood lipids (TC, TG, HDL-C, LDL-C, FFA, ApoA1, ApoB, ApoA1/ApoB, hs-CRP), twenty-five studies; glucose parameters (FBG, 2hpp1, HbA1c, HOMA-IR, C-pep, 2hPPG, 2hPPI), nineteen studies; anthropometric parameters (BW, BMI, WHR, BFP, TFA, VFA, SFA, WC, DBP, SBP), twenty studies; inflammatory cytokines (CRP, IL-6, IL-10, TNF-α), ten studies; and liver function tests, four studies. Impacts of BFF interventions on gut microbiota composition (relative abundance, correlation with genes/clinical parameters) were investigated in four studies,(Reference Han, Bose and Wang57,Reference Akamine, Millman and Uema62,Reference Han, Jeong and Ryu70,Reference Tenore, Caruso and Buonomo76) with only Akamine et al.(Reference Akamine, Millman and Uema62) measuring plasma short-chain fatty acid levels as a functional outcome of microbiota changes.

Post-intervention, all twenty-six studies reported significant within-group changes in means of some target outcomes, compared with baseline. Statistically significant impacts of BFFs on discrete cardiometabolic outcomes when compared to controls were observed in nineteen studies(Reference Kim, An and Lee53,Reference Lee, An and Jeon56,Reference Wirawanti, Hardinsyah and Briawan58Reference Somanah, Aruoma and Gunness61,Reference Salehi, Karimpour and Imani63Reference Jung, Haddad and Kaur69,Reference Cha, Yang and Back71Reference Lim, Jung and Choi74,Reference Tenore, Caruso and Buonomo76,Reference Byun, Yu and Cha77) . The number of studies with significant (p < 0·05) between-group differences in each measured outcome, arranged according to body system, is shown in Figure 4. The p-values of significant (p < 0·05) between-group differences for each study are shown in Supplementary Material 6. Tenore et al.(Reference Tenore, Caruso and Buonomo76) and Akamine et al. (Reference Akamine, Millman and Uema62) reported significant intergroup differences in gut microbiota composition. No significant difference between groups in any measured outcomes were found in five studies(Reference An, Lee and Choi54,Reference An, Lee and Jeon55,Reference Han, Bose and Wang57,Reference Mohammadi, Mazloomi and Niakousari75,Reference Nakamura, Kitagawa and Yamamoto78) , and one study(Reference Han, Jeong and Ryu70) did not perform between-group analysis. No significant adverse effects were reported. To allow for suitable comparisons within BFF subtypes, the results are presented according to the major micro-organisms involved in the production of each fermented food.

Fig. 4. Schematic diagram summarising the results of the studies reviewed. Number of studies that had outcomes with significant (p < 0·05) between-group changes post intervention are given in brackets next to the body system affected.

Lactofermented kimchi, red ginseng, ash kardeh and Annurca apple

Lactofermented foods (Table 2) are produced through lactic acid fermentation with LAB. These foods were used as interventions in eight studies: 10-d-fermented kimchi versus 1-d-fermented fresh kimchi(Reference Kim, An and Lee53Reference Han, Bose and Wang57,Reference Oh, Park and Kim73) ; fermented red ginseng versus yeast placebo(Reference Salehi, Karimpour and Imani63,Reference Oh, Park and Kim73,Reference Tenore, Caruso and Buonomo76) ; lactofermented Annurca apple puree(Reference Tenore, Caruso and Buonomo76); and ash kardeh(Reference Salehi, Karimpour and Imani63). The target population was adults with T2DM in three studies(Reference An, Lee and Choi54,Reference Salehi, Karimpour and Imani63,Reference Oh, Park and Kim73) ; overweight/obesity in two studies(Reference Kim, An and Lee53,Reference Han, Bose and Wang57) ; MetS in two studies(Reference An, Lee and Jeon55,Reference Lee, An and Jeon56) ; and cardiovascular risk factors in one study(Reference Tenore, Caruso and Buonomo76). South Koreans were the target population in six studies(Reference Kim, An and Lee53Reference Han, Bose and Wang57), with Tenore et al.(Reference Tenore, Caruso and Buonomo76) and Salehi et al.(Reference Salehi, Karimpour and Imani63) conducted in Italian and Iranian populations, respectively.

Blood lipid parameters were measured in seven studies(Reference Kim, An and Lee53Reference Han, Bose and Wang57,Reference Salehi, Karimpour and Imani63,Reference Oh, Park and Kim73,Reference Tenore, Caruso and Buonomo76) . Kim et al.(Reference Kim, An and Lee53) conducted a crossover trial of 300 g/d fermented (FeK) versus fresh kimchi (FrK) in twenty-two participants with BMI ≥25 kg/m². TC was significantly reduced in the FeK group compared with baseline; this change was significantly different from FrK (FeK −0·3 versus FrK −0·1 mmol/L; p < 0·05). However, in Han et al.’s(Reference Han, Bose and Wang57) 8-week 180 g/d parallel study of a similar female population (n = 24), HDL-C significantly improved in the FeK group compared with baseline, but this change was not significantly different from the FrK group. Notably, some subjects in both studies had elevated TG levels (≥1·7 mmol/L). In An et al.(Reference An, Lee and Choi54) and Lee et al.(Reference Lee, An and Jeon56) (abstracts only, dosages unclear), 8-week crossover trials with FeK and FrK in participants with T2DM (n = 20) or MetS (n = 16) respectively, there were no reported significant between-group differences in blood lipids. Lactofermented ash kardeh (AK) was consumed for 6 weeks, along with routine diabetic treatment, by forty-eight participants with T2DM in an open-label parallel trial(Reference Salehi, Karimpour and Imani63), many of the participants would fit MetS criteria (obese, abnormal TG and HDL-C levels, hypertension). In comparison with the routine diabetic treatment control group, 250 g/d AK consumption resulted in significant reductions in TC (AK −0·2 ± 0·5 versus control 0·1 ± 0·3 mmol/L; p = 0·025) and TG (AK −0·3 ± 0·3 versus control −0·1 ± 0·2 mmol/L; p = 0·003), as well as a significant increase in HDL-C (AK 0·04 ± 0·1 versus control −0·05 ± 0·2 mmol/L; p = 0·048). Consumption of 2·7 g/d fermented red ginseng (FRG) versus placebo for 4 weeks by forty-two T2DM/prediabetic participants did not elicit any significant between-group differences in measured blood lipids(Reference Oh, Park and Kim73), although the FRG group had significant improvements in TC, HDL-C and LDL-C compared with baseline. Tenore et al.(Reference Tenore, Caruso and Buonomo76) compared consumption of 125 g unfermented Annurca apple puree (AAP) with lactofermented AAP (lfAAP) and a probiotic containing an identical dosage of the same LAB (3·0 × 108 CFU/g Lacticaseibacillus rhamnosus LRH11 and L. plantarum SGL07), in ninety Italian adults with cardiovascular risk factors. After 8 weeks, HDL-C was significantly increased in all groups compared with baseline; both lfAAP and AAP groups in relation to each other; and to the probiotic alone (lfAAP 0·6 versus AAP 0·5 versus LAB 0·2 mmol/L; p < 0·05). Notably, there was a 31% drop-out rate. Overall, three included studies of lactofermented foods found significant (p < 0·05) between-group differences in the blood lipid parameters TC, TG or HDL-C.

Glucose parameters were measured in all lactofermented food studies. Most participants in Kim et al.’s(Reference Kim, An and Lee53) study had elevated FBG at baseline (5·7 ± 0·6 mmol/L); fermented kimchi (FeK) significantly reduced FBG and FBI compared with baseline, but only the change in FBG was significantly different to fresh kimchi (FrK) (FeK −0·3 versus FrK −0·2 mmol/L; p < 0·05). In An et al.(Reference An, Lee and Choi54), only the FrK group had improved HbA1c compared with baseline, whilst in An et al.(Reference An, Lee and Jeon55), both groups showed improvement in HbA1c, HOMA-IR and FBI. However, there were no significant between-group differences in either of these studies, or in Lee et al.(Reference Lee, An and Jeon56), Han et al.(Reference Han, Bose and Wang57) or Tenore et al.(Reference Tenore, Caruso and Buonomo76). Ash kardeh (AK)(Reference Salehi, Karimpour and Imani63) significantly reduced FBG levels compared with control (AK −1·1 ± 1·4 versus control −0·2 ± 0·5 mmol/L; p = 0·003). In Oh et al.(Reference Oh, Park and Kim73), the only significant changes from baseline were found in the FRG group (FBG, 2hPPG); after 4 weeks, the FRG group had significantly reduced 2-h postprandial glucose (FRG −1·6 versus placebo −0·5 mmol/L; p = 0·008), and increased 2-h postprandial insulin (FRG 17·1 versus placebo −2·6 µU/mL; p = 0·040) compared with placebo. Overall, three studies observed a significant improvement in the glucose parameters FBG or 2hPPG/2hPPI compared with control.

All lactofermented food studies collected anthropometric data, except for Oh et al.(Reference Oh, Park and Kim73). Salehi et al.(Reference Salehi, Karimpour and Imani63) found that ash kardeh (AK) significantly reduced SBP compared with the control group (AK −13·47 ± 14·01 versus control −0·43 ± 5·62 mmHg; p < 0·001). In Kim et al.(Reference Kim, An and Lee53), both groups had significant improvements in BW, BMI and BFP compared with baseline, whilst fermented kimchi (FeK) also reduced WHR. Compared with fresh kimchi (FrK), FeK consumption significantly reduced BFP (FeK −0·7 versus FrK −0·3%; p < 0·05), SBP (FeK −4·8 versus FrK −3·7 mmHg; p < 0·05), and DBP (FeK −4·2 versus FrK −1·7 mmHg; p < 0·05). Lee et al.(Reference Lee, An and Jeon56) also showed that FeK significantly reduced DBP (FeK −3·7 mmHg versus FrK not stated; p = 0·037). Compared with baseline, An et al.(Reference An, Lee and Jeon55) found that both the FeK and FrK groups had significant reductions in BW, BMI, WC and BFP, with only the FeK group having significant improvements in SBP and DBP. However, there were no significant intergroup differences. Similarly, Han et al.(Reference Han, Bose and Wang57) found significant within-group improvements in WC, BFP, DBP (FrK) and SBP (FeK) but no significant between-group differences. An et al.(Reference An, Lee and Choi54) observed significant within-group change in DBP in the FeK group, but no significant between-group differences. Although all studies measured inflammatory markers, only Lee et al.(Reference Lee, An and Jeon56) showed a significant between-group difference in CRP (FeK 208·5 µg/mL versus FrK not stated; p = 0·048). Overall, three studies reported significant between-group reductions in SBP and DBP compared with control, with one study reporting a significant reduction in CRP.

Han et al.(Reference Han, Bose and Wang57) and Tenore et al.(Reference Tenore, Caruso and Buonomo76) collected faecal samples for assessment of changes in gut microbiota composition. In Han et al.(Reference Han, Bose and Wang57), faecal DNA pyrosequencing and blood RNA microarray were performed on ten and four randomly selected participants per group respectively. Identified bacterial taxa were also correlated with relevant obesity genes and clinical parameters. Bifidobacterium was significantly increased in both groups after 8 weeks. In correlation analysis, significant negative correlations were found between Actinobacteria and BFP; B. longum and WC; Bifidobacterium and BW; and Bifidobacterium and BMI. In the fermented kimchi group, the versican (regulates inflammation) and acyl-CoA synthetase long-chain family member 1 (regulates lipid metabolism) genes were significantly upregulated (p < 0·05); these genes were found to negatively correlate with SBP and DBP (p < 0·05). Tenore et al.(Reference Tenore, Caruso and Buonomo76) also examined changes in gut microbial composition, and found that, when comparing lfAAP, AAP and a LAB probiotic, there were significant (p < 0·05) increases in Bifidobacterium (AAP 7·7 × 105 versus lfAAP 5·0 × 104 versus probiotic 3·8 × 105 CFU/mL) and LAB (AAP 1·1 × 106 versus lfAAP 3·5 × 103 versus probiotic 1·4 × 105 CFU/mL); and reductions in Bacteroides (AAP −4·9 × 103 versus lfAAP −1·8 × 103 versus probiotic −2·5 × 103 CFU/mL) and Enterococcus (AAP −3·5 × 102 versus lfAAP −9·2 × 102 versus probiotic −3·4 × 103 CFU/mL). The authors stated that ‘post follow up, lfAAP showed highest stability of efficacy, followed by LAB, & AAP’.

Overall, the results of the included lactofermented food studies were inconsistent. However, there was a trend of within-group improvements in measured metabolic parameters. Several significant between-group improvements were reported in two studies: Kim et al.’s(Reference Kim, An and Lee53) fermented kimchi study (SBP, DBP, BFP, FBG, TC), and Salehi et al.’s(Reference Salehi, Karimpour and Imani63) ash kardeh study (SBP, FBG, TC, TG, LDL-C).

Chungkookjang, kochujang and doenjang

Soybean-based jangs fermented with Bacillus and Aspergillus species (Table 3) were used as interventions in six studies: chungkookjang(Reference Shin, Kwon and Jeong65,Reference Back, Kim and Yang67,Reference Byun, Yu and Cha77) , kochujang(Reference Han, Jeong and Ryu70,Reference Cha, Kim and Yang72) and doenjang(Reference Cha, Yang and Back71). Shin et al.(Reference Shin, Kwon and Jeong65) investigated adults with impaired FBG, whilst the other five studies(Reference Back, Kim and Yang67,Reference Han, Jeong and Ryu70Reference Cha, Kim and Yang72,Reference Byun, Yu and Cha77) were conducted in overweight/obese adults (BMI ⩾23 kg/m² and WHR ⩾0·90 (M), ⩾0·85 (F)). All studies were conducted in South Korean adults.

All six studies measured blood lipids. Back et al.(Reference Back, Kim and Yang67) showed that 12-week consumption of chungkookjang (CKJ) (26 g/d) significantly reduced ApoB compared with baseline, and in comparison with placebo (CKJ −0·2 versus placebo −0·1 µmol/L; p = 0·027; n = 60). In Cha et al.(Reference Cha, Kim and Yang72), the kochujang (KCJ) group (32 g/d) showed improvement in TG, ApoB, ApoB/ApoA1 and ApoA1 after 12 weeks, whilst the placebo group had improvements in ApoA1 and ApoB. Compared with placebo, KCJ significantly reduced TG (KCJ −0·2 versus placebo 0·1 mmol/L; p = 0·049; n = 60). In Byun et al.’s(Reference Byun, Yu and Cha77) (n = 83; 35 g/d CKJ) 12-week crossover trial, there were changes from baseline in both the CKJ female (TG, ApoA1, ApoB) and male (TC, TG, ApoA1) groups, as well as in the placebo group. In Cha et al.(Reference Cha, Yang and Back71) (n = 51), both the doenjang (9·8 g) and placebo groups had improvements in TC, LDL-C, ApoB and ApoA1 after 12 weeks. However, there was no significant difference in any lipid measures between groups in either of these studies. These overweight/obese study populations did not have any lipid abnormalities at baseline. Shin et al.(Reference Shin, Kwon and Jeong65), an 8-week, three-arm parallel trial comparing 20 g/d of CKJ or red ginseng CKJ (RGCKJ) versus a starch control in adults with impaired FBG did show significant downward trends in TC and LDL-C (CKJ, RGCKJ), and ApoB/ApoA1, but did not find any significant between-group differences in blood lipids. However, in this study, 20 g of both CKJ and RGCKJ had significant within-group improvements in FBG, and when compared with control (CKJ −1·7 versus RGCKJ −1·1 versus control 0·2 mmol/L; p < 0·05). Han et al.(Reference Han, Jeong and Ryu70) conducted a 6-week, three-arm parallel study in overweight/obese adults comparing 19 g/d high effective micro-organisms traditional KCJ (n = 19), low effective micro-organisms traditional KCJ (n = 18) and commercial KCJ (n = 17). However, it appears that they did not conduct between-group analysis; they did find that significant within-group changes only occurred in the high effective micro-organisms KCJ group (TC, LDL-C, HDL-C, TG). Cha et al.(Reference Cha, Kim and Yang72) found no significant differences in glucose parameters between KCJ and placebo groups; the remaining studies did not measure these parameters. Overall, although all studies found within-group differences in blood lipid parameters, two studies found significant between-group differences in TC or TG.

In Byun et al.(Reference Byun, Yu and Cha77), chungkookjang (CKJ) significantly reduced WC (CKJ −1·05 cm versus placebo 1·84 cm; p = 0·0067), WHR (CKJ −0·01 versus placebo 0·02; p = 0·0083) and BFP (CKJ −0·84% versus placebo 0·2%; p = 0·0488) in female participants (n = 43). After 12 weeks, Cha et al.(Reference Cha, Kim and Yang72) found that kochujang elicited a significant reduction in VFA (kochujang −4·8 cm2 versus placebo −0·4 cm2; p = 0·043). Cha et al.(Reference Cha, Yang and Back71) showed that, compared with baseline, doenjang (DE) significantly reduced BW, BFM, BFP and VFA, whilst both DE and placebo groups had improvements in TFA, SFA and WHR. There was a significant between-group difference in BW (DE −0·8 kg versus placebo −0·3 kg; p < 0·001), BFM (DE −0·7 kg versus placebo −0·4 kg; p < 0·001), VFA (DE −8·6 cm2 versus placebo −0·6 cm2; p = 0·041) and BFP (DE −0·6% versus placebo −0·4%; p = 0·007). Han et al.(Reference Han, Jeong and Ryu70) only reported significant within-group differences: high effective micro-organisms kochujang reduced WC and VFA; commercial kochujang reduced WC and WHR. This study also measured ‘beneficial, harmful and other’ faecal micro-organisms, but found no significant within-group differences. Back et al.’s(Reference Back, Kim and Yang67) 12-week administration of 26 g/d CKJ did not elicit any significant within- or between-group differences in anthropometric measures. Overall, two studies found significant reductions in certain anthropometric parameters compared to control, with another only finding significant differences in the women’s subgroup.

Despite significant differences from baseline in lipid and glucose parameters, jangs elicited few consistent between-group improvements. However, several studies showed small but significant changes in anthropometric parameters in overweight/obese participants: BW, BFM, BFP, WC, WHR and VFA.

Shiokoji, miso, touchi, amazake and kochujang

Aspergillus oryzae-fermented products (Table 4) were administered to different target populations in five studies: Touchi-extract supplemented Houji tea(Reference Fujita, Yamagami and Ohshima66) and shiokoji(Reference Nakamura, Kitagawa and Yamamoto78) in adults with mild hyperglycaemia or borderline/mild T2DM; A. oryzae-fermented KCJ in adults with mild hyperlipidaemia(Reference Lim, Jung and Choi74); brown rice amazake versus white rice amazake in adults with MetS(Reference Akamine, Millman and Uema62); and miso in adults with high normal BP or stage I hypertension(Reference Kondo, Sakuyama Tomari and Yamakawa68). South Korean adults were participants in three studies, with Japanese adults investigated in Akamine et al.(Reference Akamine, Millman and Uema62) and Lim et al.(Reference Lim, Jung and Choi74).

Nakamura et al.(Reference Nakamura, Kitagawa and Yamamoto78) administered 4 weeks of 15 g/d of shiokoji or placebo to forty-nine adults with mild hyperglycaemia (FBG 5·6–6·9 mmol/L). After data were collected, two analyses were conducted: primary analysis (n = 47), and secondary analysis (n = 30), for which participants who had finished the study were deemed ineligible due to marked changes in lifestyle (alcohol consumption/exercise) and/or HOMA >5·0 post food allocation. In the primary analysis, there was no significant difference between the groups. In the secondary analysis, shiokoji significantly reduced FBG (shiokoji −0·2 mmol/L versus placebo 0·1 mmol/L; p = 0·02) after 4 weeks. However, after 12 weeks, there was no significant difference between the groups. Blood lipids and anthropometric parameters were collected but not reported. To thirty-eight subjects with borderline/mild T2DM, Fujita et al.(Reference Fujita, Yamagami and Ohshima66) administered 0·3 g Aspergillus-fermented Touchi extract supplemented with Houji tea (TE) versus control for 12 weeks. The TE group had within-group improvements in FBG and HbA1c. Compared with the control, TE significantly improved FBG (TE −0·5 mmol/L versus control 0·2 mmol/L; p < 0·05) and HbA1c (TE −0·5% versus control 0·1%; p < 0·01). There were no significant between-group differences in anthropometric parameters or blood lipids.

In Lim et al.’s(Reference Lim, Jung and Choi74) 12-week parallel study, thirty adults with mild hyperlipidaemia (LDL-C 2·8–4·9 mmol/L or TC 5·2–6·7 mmol/L) were administered 34·5 g/d of A. oryzae- fermented kochujang (KCJ). The KCJ group had a significant reduction in TC compared with baseline, as well as compared with placebo (KCJ −0·5 versus placebo −0·2 mmol/L; p = 0·045). No significant between-group differences in anthropometric parameters were reported.

Kondo et al.(Reference Kondo, Sakuyama Tomari and Yamakawa68) studied forty adults with high-normal BP (130–139/85–89 mmHg) or untreated stage I hypertension (140–159/90–99 mm Hg), who were given a 2:1 mixture of high-ACE inhibitory MK-34-1 miso and Nenrin (common) miso, or a soy food control for 8 weeks. The miso group had significant improvements in LDL-C, night-time SBP and DBP compared with baseline, but there were no other within- or between-group differences in glucose, lipid or liver function parameters. There were significant between-group changes in BW (miso 0·636 kg versus control 0·073 kg; p < 0·05). In the whole group, as well as the stage I hypertensive subgroup, miso was found to significantly (p < 0·05) decrease night-time SBP and DBP when compared with the control; it did not significantly affect daytime blood pressure.

Akamine et al.(Reference Akamine, Millman and Uema62) compared the 4-week consumption of 350 g/d brown rice amazake (BA) to a white rice amazake (WA) control in 40 adults with MetS (BMI ⩾25 kg/m² and any 2 of: (1) TG ⩾1·7 mmol/L; (2) HDL-C ⩽1·0 mmol/L; (3) SBP ⩾130 mmHg ± DBP ⩾85 mmHg; or (4) FBG ⩾5·6 mmol/L). There were no significant within- or between-group differences in any glucose, lipid or anthropometric parameters, or short-chain fatty acid plasma concentration. The primary outcome of this study was changes in gut microbiota composition. The WA group had a significant increase in Flavonifractor plautii, but there were no other within-group differences. Comparing the two groups, after 4 weeks intervention, there were significant increases in the BA group in: the family Porphyromonadaceae (p = 0·013); the genera Parabacteroides (p = 0·011), Butyricicoccus (p = 0·012) and Sutterella (p = 0·001); and species Sutterella wadsworthensis (p = 0·001), Lactobacillales bacterium DJF B280 (p = 0·005), Firmicutes bacterium DJF VP44 (p = 0·038) and Eubacterium spp. A2 207(Reference Barcenilla, Pryde and Martin79) (p = 0·012). There were no significant differences between the groups in alpha diversity, beta diversity or taxonomy-based abundance analyses at the phylum level. To assess whether these gut microbial changes correlated with metabolic biomarkers in each subject at species level, correlation analyses were carried out. Significant correlations (correction r > 0·4 or r < −0·4; p < 0·05) were identified among the included forty-one taxa and thirteen clinical indices. Significant negative correlations were found between Sutterella wadsworthensis/blood glucose (p = 0·032); Lactobacillales bacterium DJF B280/TG (p = 0·006); and Eubacterium spp. A2 207/TG (p = 0·001). Significant positive correlations were found between Lactobacillales bacterium DJF B280/HDL-C (p = 0·037), and Eubacterium spp. A2 207/LDL-C (p = 0·046).

Overall, A. oryzae-fermented foods elicited few consistent, significant within- or between-group differences in lipid, glucose or anthropometric measurements.

Tempeh

Tempeh (Rhizopus spp.) products (Table 5) were used as interventions in three studies: Wirawanti et al.’s(Reference Wirawanti, Hardinsyah and Briawan58) tempeh drink study in hypercholesterolaemic adults; Afifah et al.’s(Reference Afifah, Nabilah and Supraba59) study in women with hyperlipidaemia; and Nadia et al.’s(Reference Nadia, Wati and Isnawati60) study of women with premenopausal obesity. All tempeh studies were conducted with Indonesian participants.

In Nadia et al.’s(Reference Nadia, Wati and Isnawati60) open, parallel study of tempeh gembus in an isoenergetic diet (30 kcal/kg of BW/d) versus isoenergetic diet alone, all forty participants were prediabetic or diabetic (range 2·6 to 5·1 mmol/L; median ∼3·2 mmol/L), with BMI between 25·1 and 47·7kg/m². At baseline, most participants in this study had low HDL-C (⩽1·0 mmol/L) and elevated TG (⩾1·7 mmol/L) levels. After 4 weeks, 150 g/d of tempeh gembus significantly improved FBG, FBI, HOMA-IR, TG, HDL-C and hs-CRP in both groups. Between-group differences were observed in FBG (tempeh −2·3 versus control −0·4 mmol/L; p = 0·00); HOMA-IR (tempeh −18·4 versus control −0·6; p = 0·00) and FBI (tempeh −14·2 µIU/L versus control −0·7 µIU/L; p = 0·00). TG (tempeh −0·1 versus control 0·03 mmol/L; p = 0·01) and hs-CRP (tempeh −1·94 mg/L versus control −1·65 mg/L; p = 0·03) and HDL-C were significantly improved by tempeh gembus consumption (tempeh 0·2 versus control 0·2 mmol/L; p = 0·00) compared with diet alone.

Afifah et al.(Reference Afifah, Nabilah and Supraba59) ran a 2-week open, parallel study in forty-five women with hyperlipidaemia (TC ⩾5·2 mmol/L and/or LDL-C ⩾2·6 mmol/L). Tempeh gembus at two different dosages (103 g T1 and 206 g T2) and the control all showed improvements in TC and LDL-C compared with baseline. T1 and T2 significantly decreased LDL-C (T1 −1·1 versus T2 −1·3 versus control −0·7 mmol/L; p = 0·035) when compared with each other and with the control (10 mg simvastatin); post-hoc analysis showed that T2 had the largest effect, significantly reducing LDL-C (means difference −0·6 mmol/L; p = 0·011) when compared with control.

Hypercholesterolaemic (TC ⩾5·2 mmol/L; n = 54) adults were provided tempeh drinks made with local (TA) or imported (TB) soybeans, or no intervention, in Wirawanti et al.’s(Reference Wirawanti, Hardinsyah and Briawan58) open, parallel study. Both TA and TB were found to significantly decrease TC after 4 weeks, and the no-intervention control group had reductions in LDL-C and HDL-C. TA and TB improved LDL-C (TA −0·1 versus TB −0·03 versus control 0·2 mmol/L; p = 0·01) compared with control; there was no significant difference between TA and TB.

Compared with controls, even over very short periods of time (i.e. 2–4 weeks), tempeh seemed to significantly improve blood lipids (LDL-C, HDL-C, TG) in hyperlipidaemic participants, and improved FBG, HOMA-IR and FBI in a prediabetic/diabetic obese female population.

Other BFFs

A variety of other BFFs (Table 6) were administered to different populations in four studies: Kim et al.(Reference Kim, Kim and Lee64) administered brown rice lees (LB, with A. oryzae, R. oryzae, LAB, yeasts) to adults with T2DM; Somanah et al.(Reference Somanah, Aruoma and Gunness61) examined the effect of fermented papaya preparation (FPP, with E. faecalis, A. oryzae) in adults with neodiabetes; Jung et al.(Reference Jung, Haddad and Kaur69) used a yeast-fermented soy powder (Q-Can) in adults with cardiovascular risk factors; and, Mohammadi et al.(Reference Mohammadi, Mazloomi and Niakousari75) used fermented garlic extract within chocolate in adults with hypertension.

Somanah et al.’s(Reference Somanah, Aruoma and Gunness61) 14-week parallel fermented papaya preparation (FPP) (6 g) versus water control study (n = 127) yielded some within-group improvements (control males and females: TC, LDL-C, HDL-C; control males: SBP, DBP; FPP (M): TG; FPP females: BMI, DBP), but no significant between-group differences in glucose, lipid or anthropometric parameters in Mauritian adults with neo diabetes (FBG 5·1–5·9 mmol/L). The only significant finding was a reduction in CRP (p = 0·018, inconsistent data provided) in the FPP group compared with the control. It should be noted here that this study was assessed as being of poor quality with a high risk of bias, due to a high number of dropouts (twenty-seven), substantial differences in group size allocation and unexplained, inconsistent numbers of participants with missing outcome data for each outcome.

Kim et al.(Reference Kim, Kim and Lee64) administered 40 g brown rice lees (LB), a by-product of Korean rice wine, for 12 weeks to thirty South Korean adults with T2DM (FBG ⩾7 mmol/L or ⩾11·1 mmol/L in OGTT), finding no significant change in any glucose parameters compared with the 45 g mixed grain (MG) control. The MG group had significant within-group improvements in TC, LDL-C, SBP, AST and ALT, whilst the LB group had improved WC. Compared with MG, LB significantly increased TC (LB 0·3 versus MG −0·2 mmol/L; p < 0·05) and LDL-C (LB 0·2 versus MG −0·3 mmol/L; p < 0·05). The authors explained the unexpected increase in these parameters as likely due to ‘relatively high-fat content’ of LB and ‘high insoluble fibre content’ of MG, resulting in ‘an increase in cholesterol levels in the LB group vs the MG group’. In the LB group, there were also significant between-group improvements in TC (LB −2·8 cm versus MG −0·9 cm; p < 0·05), as were AST (LB −4·5 IU/mL versus MG −0·1 IU/mL; p < 0·05) and ALT (LB −6·7 IU/mL versus MG 0·1 IU/mL; p < 0·05).

In a 12-week crossover trial, Jung et al.(Reference Jung, Haddad and Kaur69) compared a yeast-fermented soy powder (FS), Q-Can, with a germinated brown rice (GBR) powder control in twenty-nine American adults with cardiovascular risk factors (⩾2: smoking; SBP ⩾140/90 mmHg; LDL-C ⩾2·9 mmol/L; HDL-C ⩾1·0 mmol/L; TG ⩾1·7 mmol/L; FBG ⩾6·1 mmol/L; BMI ⩾25 kg/m²; family history of heart disease). Compared with baseline, FS significantly improved TC, LDL-C and ApoB, whilst GBR had improvements in HDL-C. Between groups, there were significant improvements in TC (FS −0·23 mmol/L versus GBR 0·14 mmol/L; p = 0·0024), LDL-C (FS −0·18 mmol/L versus GBR 0·4 mmol/L; p = 0·0317), HDL-C (FS −0·03 mmol/L versus GBR 0·09 mmol/L; p = 0·0036) and ApoA1 (FS −1·4 µmol/L versus GBR 1·4 µmol/L; p = 0·0390). There were no significant between-group differences in anthropometric measures or measured glucose parameters.

Mohammadi et al.(Reference Mohammadi, Mazloomi and Niakousari75) administered dark chocolate containing 650 mg of fermented garlic extract or a dark chocolate control to forty-four hypertensive (SBP ⩾130 mmHg ± DBP ⩾80 mmHg, or consuming antihypertensive drug) adults in a 6-week crossover trial. No significant differences were found within or between groups in lipid, anthropometric or inflammatory parameters. Glucose metabolism was not investigated.

Discussion

Our review found that almost all the studied BFFs, compared with controls, significantly improved certain lipid, glucose, anthropometric and inflammatory parameters in those with overweight/obesity, MetS or T2DM. Consumption of some BFFs led to significant beneficial changes in gut microbiota composition, with correlations to metabolic outcomes and gene expression. However, the results were inconsistent across studies, even within BFF subgroups such as fermented kimchi, and 46% of the studies were at medium to high risk of bias. The clinical relevance of these results is yet to be determined and should consider the methodological limitations of these studies.

The strength of our review lies in our robust and inclusive methodology. We were able to locate a culturally diverse range of relevant RCTs from 2001 onwards by ensuring our review had no language limits, searched from inception and included specific traditional BFF terminology from many cultures. To recognise and assess the emerging role of the gut microbiota in nutritional research, we included changes in gut microbiota composition as one of our target outcomes. To the best of our knowledge, our included studies have not been assessed in other systematic reviews or meta-analyses, and this is the first review using systematic Cochrane methodology to identify the impact of BFFs (excluding red yeast rice and vinegar) in our target population. However, our ability to perform meta-analyses, and the extrapolation of our findings, are limited by the diversity of BFFs, undeclared microbial strains and counts, dosage, intervention length and administration, and variable target conditions and study populations.

To the best of our knowledge, our systematic review is unusual in collecting data on changes in gut microbiota composition with cardiometabolic outcomes. Colonic microbiota has an established role in the development of obesity, MetS and T2DM, and fermented food consumption affects gut microbiota composition. However, only 15% (4/26) of our reviewed studies collected faecal samples for gut microbiota profiling(Reference Han, Bose and Wang57,Reference Akamine, Millman and Uema62,Reference Han, Jeong and Ryu70,Reference Tenore, Caruso and Buonomo76) , with only Han et al.(Reference Han, Bose and Wang57) and Akamine et al.(Reference Akamine, Millman and Uema62) correlating these findings to gene expression and/or metabolic outcomes. Kochujang with varying levels of effective micro-organisms did not elicit significant changes in gut microbiota composition(Reference Han, Jeong and Ryu70). In the other studies, BFF consumption generally increased the abundance of beneficial bacteria. Both fresh and fermented kimchi significantly increased Bifidobacterium (Reference Han, Bose and Wang57), but there were no between-group differences. Lactofermented Annurca apple puree, compared with a probiotic or non-fermented puree control, significantly increased beneficial Bifidobacterium and LAB, whilst reducing potential pathogens Bacteroides and Enterococcus (Reference Tenore, Caruso and Buonomo76). A similar trend was observed in Akamine et al.(Reference Akamine, Millman and Uema62), where brown rice amazake significantly increased beneficial bacteria on a family and genus level, including Sutterella wadsworthensis, Lactobacillales bacterium DJF B280, Firmicutes bacterium DJF VP44 and Eubacterium spp. A2 207, compared with white rice amazake. These findings concur with BFF-induced increases in beneficial bacteria and concurrent reductions of pathogenic bacteria during in vitro and in vivo studies(Reference Yamamoto, Suzuki and Yonezu80Reference Wu, Chu and Cheng82), as well as a limited number of human interventional trials(Reference Cheng, Shang and Lin83Reference Nielsen, Garnas and Jensen86). Additionally, both Han et al.(Reference Han, Bose and Wang57) and Akamine et al.(Reference Akamine, Millman and Uema62) correlated taxa with clinical indices, finding significant negative and positive correlations indicating the potential beneficial effects of certain bacteria on discrete metabolic outcomes. However, the included studies suffered from short intervention periods and used a variety of methods to quantify microbes. Interpretation of these results must consider the effects of short- and long-term dietary changes on gut microbiota(Reference Leeming, Johnson and Spector87), and the relative paucity of gut microbial data in such studies. Due to their important role in diet and metabolism, we suggest that all RCTs of fermented foods should collect data on the gut microbiota(Reference Shanahan, McMaster and Staudacher88), including microbial metabolite levels, and correlations with metabolic outcomes and gene expression.

In comparing our review with similar studies, we found Gille et al.’s(Reference Gille, Schmid and Walther18) 2018 review of meta-analyses of fermented foods and noncommunicable disease, and SaeidiFard et al.’s(Reference SaeidiFard, Djafarian and Shab-Bidar89) 2020 meta-analysis of fermented foods and inflammation. Notably, the studies included in these reviews did not overlap with our included publications. Gille et al.’s review stated that ‘The literature on fermented plants is characterised by a wealth of in vitro data, whose positive results are not corroborated in humans due to the absence of RCTs’(Reference Gille, Schmid and Walther18). Their section on ‘fermented foods of plant origin’ focused only on those relevant to the Swiss population (coffee, wine, beer, sauerkraut, fermented olives). SaeidiFard et al.’s(Reference SaeidiFard, Djafarian and Shab-Bidar89) meta-analyses included healthy participants, fermented dairy products and probiotic-added non-fermented foods. They found that CRP and IL-6 were not improved by fermented food intake, with only a reduction in TNF-α. Although nine of our included studies(Reference Kim, An and Lee53Reference Han, Bose and Wang57,Reference Somanah, Aruoma and Gunness61,Reference Han, Jeong and Ryu70,Reference Mohammadi, Mazloomi and Niakousari75) measured some inflammatory markers (including CRP, IL-10, IL-6, TNF-a), only one fermented versus fresh kimchi study(Reference Lee, An and Jeon56) (p = 0·048) and the fermented papaya study(Reference Somanah, Aruoma and Gunness61) (p = 0·018) observed significant reductions in CRP compared with controls. We also identified several recent meta-analyses of RCTs using vinegar or red yeast rice/monacolin/xuezhikang in adults with obesity, T2DM or dyslipidaemia. All the meta-analyses of vinegar consumption found that glycaemic control was improved in individuals with IGT or T2DM, including: FBG(Reference Cheng, Jiang and Wu42,Reference Valdes, So and Gill44) ; HbA1c(Reference Siddiqui, Assam and de Souza40,Reference Cheng, Jiang and Wu42) ; and acute glucose response/postprandial glucose and insulin(Reference Shishehbor, Mansoori and Shirani41,Reference Cherta-Murillo, Pugh and Alaraj-Alshehhi45) . Cheng et al.(Reference Cheng, Jiang and Wu42) also found statistically significant improvement in LDL-C and TC levels. However, there was substantial heterogeneity reported across studies within each meta-analysis, and there were varying numbers of RCTs assessed by each review. Like vinegar, many BFFs contain acetic acid and short chain fatty acids. Out of nineteen studies which measured glucose outcomes, our review found that only six studies(Reference Kim, An and Lee53,Reference Nadia, Wati and Isnawati60,Reference Kim, Kim and Lee64Reference Fujita, Yamagami and Ohshima66,Reference Oh, Park and Kim73,Reference Mohammadi, Mazloomi and Niakousari75) of different BFFs had significant improvements in some aspect of glycaemic control (FBG, HbA1c, FBI, 2hPPG and 2hPPI). Importantly, BFF subgroups, such as fermented kimchi, did not have the same effects on glycaemic control across all studies, which may be due to differences in study length, dosage and study populations. Meta-analyses of red yeast rice identified significant improvements in LDL-C, TC and TG compared with placebo in dyslipidaemic patients(Reference Sungthong, Yoothaekool and Promphamorn36,Reference Li, Wang and Chen37,Reference Gerards, Terlou and Yu90) . Our review also found that some lactofermented foods(Reference Kim, An and Lee53,Reference Salehi, Karimpour and Imani63,Reference Tenore, Caruso and Buonomo76) , jangs(Reference Back, Kim and Yang67,Reference Cha, Kim and Yang72,Reference Lim, Jung and Choi74) , tempeh(Reference Wirawanti, Hardinsyah and Briawan58Reference Nadia, Wati and Isnawati60), brown rice lees(Reference Kim, Kim and Lee64) and fermented soy powder(Reference Jung, Haddad and Kaur69) significantly improved lipid parameters compared with controls. Furthermore, while some anthropometric parameters (BFP, BFM, BW, WC, WHR, VFA) were significantly improved in six studies(Reference Kim, An and Lee53,Reference Kim, Kim and Lee64,Reference Kondo, Sakuyama Tomari and Yamakawa68,Reference Cha, Yang and Back71,Reference Cha, Kim and Yang72,Reference Byun, Yu and Cha77) , results were inconsistent across and within BFF subgroups. As we did not conduct meta-analyses, these comparisons with other reviews need to be made with caution.

For several reasons, caution must be exercised by clinicians and policymakers when considering our review’s findings. The findings from nutritional clinical trials may not be amenable to meta-analysis by their very nature; unlike drug trials, methodology varies considerably across studies, and dietary intake is notoriously difficult to assess and control(Reference Barnard, Willett and Ding91). For example, consider the fresh versus fermented kimchi subgroup of studies, performed by the same clinical group(Reference Kim, An and Lee53Reference Han, Bose and Wang57,Reference Oh, Park and Kim73) . Kim et al.(Reference Kim, An and Lee53) supplied a larger dose for a shorter period (300 g/d for 4 weeks) than Han et al.(Reference Han, Bose and Wang57) (180 g/d for 8 weeks) in a similar obese population. However, only Kim et al.(Reference Kim, An and Lee53) found significant between-group differences in FBG, TC, BFP, SBP and DBP post intervention. An et al.(Reference An, Lee and Jeon55) also studied a prediabetic population with MetS, administering 300 g/d over 8 weeks but, like Han et al.(Reference Han, Bose and Wang57), did not find any significant between-group differences. In relation to the role of diet in nutritional RCTs, several of the studies we assessed implemented isoenergetic diets or provided meals in addition to the BFF/placebo for the duration of the study. Although these were controlled for, these dietary changes may have confounding effects on study outcomes(Reference Welch, Antoine and Berta92). Small sample sizes may affect statistical power in some of the included studies. Three studies only recruited women(Reference Han, Bose and Wang57,Reference Afifah, Nabilah and Supraba59,Reference Nadia, Wati and Isnawati60) and two found significant differences only in the male or female subgroups(Reference Somanah, Aruoma and Gunness61,Reference Byun, Yu and Cha77) : clinically speaking, cardiovascular risk may vary between genders in those with MetS(Reference Rochlani, Pothineni and Mehta93). Ten included studies(Reference Kim, An and Lee53Reference Akamine, Millman and Uema62) were at risk of selection bias as they did not provide sufficient information regarding random sequence generation and allocation concealment. Notably, twenty-one out of twenty-six included studies were conducted in East Asian populations (South Korea, Japan, Indonesia) with one from Western Asia(Reference Salehi, Karimpour and Imani63) (Iran) using traditional BFFs from these regions, which reflects the importance of these products in their diets. The other studies from Italy(Reference Tenore, Caruso and Buonomo76), the USA(Reference Jung, Haddad and Kaur69), Iran(Reference Mohammadi, Mazloomi and Niakousari75) and Mauritius(Reference Somanah, Aruoma and Gunness61) used non-traditional fermented interventions. East Asians may be at higher risk of poor cardiometabolic outcomes at a lower BMI than Caucasians(94). This is reflected in our decision to include overweight/obese participant studies with a BMI cut-off of ⩾23 kg/m², and MetS waist circumference cut-off for East Asians as per the WHO(95) obesity guidelines (Supplementary Material 1). Our review indicates the need for consensus on consistent and robust study designs for BFFs.

For future studies of BFFs, we recommend the following: transparent sample size calculations for larger, well-powered studies; improved randomisation and allocation concealment methodology, with details provided in publications; dose ranges (as more dosage studies are needed); appropriate comparators/controls; intervention periods exceeding 12 weeks; gut microbiota data collection; and intention-to-treat analysis for missing data.

Conclusion

Our systematic review indicates that BFFs have the potential to improve metabolic outcomes in individuals with obesity, MetS and T2DM. The marginal effects of BFFs may be linked to the lack of qualified systematic studies. Based on existing results, we suggest that BFFs may be best used in conjunction with lifestyle modifications and appropriate medications. However, the paucity of studies in this area, in contrast to the diverse range of available BFFs, suggests that large-scale, long-term and well-designed RCTs with a low risk of bias are required before more definitive conclusions can be made.

Supplementary material

To view supplementary material for this article, please visit https://doi.org/10.1017/S0954422423000252

Acknowledgements

None.

Financial support

M.C. was supported by an Australian Government Research Training Program (RTP) Scholarship through the University of Melbourne, Australia. Award/grant number: N/A.

Competing interests

None.

Authorship

M.C., E.E. and K.H. conceived the study. M.C., E.E., K.H. and L.J. contributed to planning of the review and determination of its scope. M.C. and K.H. designed the protocol with input from H.B., E.E., N.L. and L.J. M.C. and H.B. developed the search strategy and executed the search, with H.B. translating the searches for multiple databases. M.C. and N.L. screened the studies, extracted data, assessed quality of evidence and performed data analysis. M.C. drafted and revised the manuscript, with editorial input from K.H., E.I.E., L.J., N.L. and H.B. All authors read and approved the final manuscript and order of authorship. A preprint of this paper was published in December 2022 on medRxiv at https://www.medrxiv.org/content/10·1101/2022·12·10·22283002v1.

References

Alberti, KGMM, Zimmet, P, Shaw, J (2006) Metabolic syndrome—a new world-wide definition. A consensus statement from the international diabetes federation. Diabet Med 23, 469480.CrossRefGoogle ScholarPubMed
Grundy, SM (2006) Metabolic syndrome: connecting and reconciling cardiovascular and diabetes worlds. J Am Coll Cardiol 47, 10931100.CrossRefGoogle ScholarPubMed
Alberti, K, Eckel, RH, Grundy, SM, et al. (2009) Harmonizing the metabolic syndrome. Circulation 120, 16401645.CrossRefGoogle ScholarPubMed
Festi, D, Schiumerini, R, Eusebi, LH, et al. (2014) Gut microbiota and metabolic syndrome. World J Gastroenterol 20, 16079.CrossRefGoogle ScholarPubMed
Ley, RE, Turnbaugh, PJ, Klein, S, et al. (2006) Human gut microbes associated with obesity. Nature 444, 10221023.CrossRefGoogle ScholarPubMed
Duncan, SH, Belenguer, A, Holtrop, G, et al. (2007) Reduced dietary intake of carbohydrates by obese subjects results in decreased concentrations of butyrate and butyrate-producing bacteria in feces. Appl Environ Microbiol 73, 10731078.CrossRefGoogle ScholarPubMed
Turnbaugh, PJ, Ridaura, VK, Faith, JJ, et al. (2009) The effect of diet on the human gut microbiome: a metagenomic analysis in humanized gnotobiotic mice. Sci Transl Med 1, 6ra14.CrossRefGoogle ScholarPubMed
Le Chatelier, E, Nielsen, T, Qin, J, et al. (2013) Richness of human gut microbiome correlates with metabolic markers. Nature 500, 541546.CrossRefGoogle ScholarPubMed
Muegge, BD, Kuczynski, J, Knights, D, et al. (2011) Diet drives convergence in gut microbiome functions across mammalian phylogeny and within humans. Science 332, 970974.CrossRefGoogle ScholarPubMed
David, LA, Maurice, CF, Carmody, RN, et al. (2014) Diet rapidly and reproducibly alters the human gut microbiome. Nature 505, 559563.CrossRefGoogle ScholarPubMed
Fabiani, R, Naldini, G, Chiavarini, M (2019) Dietary patterns and metabolic syndrome in adult subjects: a systematic review and meta-analysis. Nutrients 11, 2056.CrossRefGoogle ScholarPubMed
O’Neill, S, O’Driscoll, L (2015) Metabolic syndrome: a closer look at the growing epidemic and its associated pathologies. Obes Rev 16, 112.CrossRefGoogle Scholar
World Obesity Atlas 2022 (n.d.) World Obesity Day. https://s3-eu-west-1.amazonaws.com/wof-files/World_Obesity_Atlas_2022.pdf (accessed December 2022).Google Scholar
Saeedi, P, Petersohn, I, Salpea, P, et al. (2019) Global and regional diabetes prevalence estimates for 2019 and projections for 2030 and 2045: results from the International Diabetes Federation Diabetes Atlas, 9th edition. Diabetes Res Clin Pract 157, 107843.CrossRefGoogle Scholar
Zhang, C, Derrien, M, Levenez, F, et al. (2016) Ecological robustness of the gut microbiota in response to ingestion of transient food-borne microbes. ISME J 10, 22352245.CrossRefGoogle ScholarPubMed
Marco, ML, Heeney, D, Binda, S, et al. (2017) Health benefits of fermented foods: microbiota and beyondurrO pinin Biotechnol 44, 94102.Google ScholarPubMed
Derrien, M, van Hylckama Vlieg, JET (2015) Fate, activity, and impact of ingested bacteria within the human gut microbiota. Trends Microbiol 23, 354366.CrossRefGoogle ScholarPubMed
Gille, D, Schmid, A, Walther, B, et al. (2018) Fermented food and non-communicable chronic diseases: a review. Nutrients 10, 448.CrossRefGoogle ScholarPubMed
Pasolli, E, De Filippis, F, Mauriello, IE, et al. (2020) Large-scale genome-wide analysis links lactic acid bacteria from food with the gut microbiome. Nat Commun 11, 2610.CrossRefGoogle ScholarPubMed
Taylor, BC, Lejzerowicz, F, Poirel, M, et al. (2020) Consumption of fermented foods is associated with systematic differences in the gut microbiome and metabolome. mSystems 5, e00901e00919.Google ScholarPubMed
Wastyk, HC, Fragiadakis, GK, Perelman, D, et al. (2021) Gut microbiota-targeted diets modulate human immune status. Cell 184, 41374153.e14.CrossRefGoogle ScholarPubMed
Barengolts, E (2016) Gut microbiota, prebiotics, probiotics, and synbiotics in management of obesity and prediabetes: review of randomized controlled trials. Endocr Pract Jacksonv 22, 12241234.CrossRefGoogle ScholarPubMed
Companys, J, Pla-Pagà, L, Calderón-Pérez, L, et al. (2020) Fermented dairy products, probiotic supplementation, and cardiometabolic diseases: a systematic review and meta-analysis. Adv Nutr 11, 834863.CrossRefGoogle Scholar
Savaiano, DA, Hutkins, RW (2020) Yogurt, cultured fermented milk, and health: a systematic review. Nutr Rev 79, 599614.CrossRefGoogle Scholar
Nguyen, LA (2015) Health-promoting microbes in traditional Vietnamese fermented foods: a review. Food Sci Hum Wellness 4, 147161.Google Scholar
Hutkins, RW (2018) Introduction. In Microbiology and technology of fermented foods. 2nd ed. pp. 125. New York: Wiley-Blackwell.Google Scholar
Leech, J, Cabrera-Rubio, R, Walsh, AM, et al. (2020) Fermented-food metagenomics reveals substrate-associated differences in taxonomy and health-associated and antibiotic resistance determinants. mSystems 5, e0052220.CrossRefGoogle ScholarPubMed
Lavefve, L, Cureau, N, Rodhouse, L, et al. (2021) Microbiota profiles and dynamics in fermented plant-based products and preliminary assessment of their in vitro gut microbiota modulation. Food Front 2, 268281.CrossRefGoogle Scholar
Panghal, A, Janghu, S, Virkar, K, et al. (2018) Potential non-dairy probiotic products – a healthy approach. Food Biosci 21, 8089.CrossRefGoogle Scholar
Rodzi, NARM, Lee, LK (2021) Traditional fermented foods as vehicle of non-dairy probiotics: perspectives in South East Asia countries. Food Res Int 150, 110814.CrossRefGoogle ScholarPubMed
Luckow, T, Delahunty, C (2004) Which juice is ‘healthier’? A consumer study of probiotic non-dairy juice drinks. Food Qual Prefer, Fifth Rose Marie Pangborn Sensory Science Symposium 15, 751759.CrossRefGoogle Scholar
Silanikove, N, Leitner, G, Merin, U (2015) The interrelationships between lactose intolerance and the modern dairy industry: global perspectives in evolutional and historical backgrounds. Nutrients 7, 73127331.CrossRefGoogle ScholarPubMed
Diez-Ozaeta, I, Astiazaran, OJ (2022) Fermented foods: an update on evidence-based health benefits and future perspectives. Food Res Int 156, 111133.CrossRefGoogle ScholarPubMed
Rul, F, Béra-Maillet, C, Champomier-Vergès, MC, et al. (2022) Underlying evidence for the health benefits of fermented foods in humans. Food Funct 13, 48044824.CrossRefGoogle ScholarPubMed
Xiong, X, Wang, P, Li, X, et al. (2017) The effects of red yeast rice dietary supplement on blood pressure, lipid profile, and C-reactive protein in hypertension: a systematic review. Crit Rev Food Sci Nutr 57, 18311851.Google ScholarPubMed
Sungthong, B, Yoothaekool, C, Promphamorn, S, et al. (2020) Efficacy of red yeast rice extract on myocardial infarction patients with borderline hypercholesterolemia: a meta-analysis of randomized controlled trials. Sci Rep 10, 2769.CrossRefGoogle ScholarPubMed
Li, P, Wang, Q, Chen, K, et al. (2022) Red yeast rice for hyperlipidemia: a meta-analysis of 15 high-quality randomized controlled trials. Front Pharmacol 12, 819482.CrossRefGoogle ScholarPubMed
Xu, G, Lin, M, Dai, X, et al. (2022) Comparing the effectiveness of Chinese patent medicines containing red yeast rice on hyperlipidaemia: a network meta-analysis of randomized controlled trials. Endocrinol Diabetes Metab 5, e00314.CrossRefGoogle ScholarPubMed
Yuan, R, Yuan, Y, Wang, L, et al. (2022) Red yeast rice preparations reduce mortality, major cardiovascular adverse events, and risk factors for metabolic syndrome: a systematic review and meta−analysis. Front Pharmacol 13, 744928.CrossRefGoogle ScholarPubMed
Siddiqui, FJ, Assam, PN, de Souza, NN, et al. (2018) Diabetes control: is vinegar a promising candidate to help achieve targets? J Evid-Based Integr Med 23, 112.CrossRefGoogle Scholar
Shishehbor, F, Mansoori, A, Shirani, F (2017) Vinegar consumption can attenuate postprandial glucose and insulin responses; a systematic review and meta-analysis of clinical trials. Diabetes Res Clin Pract 127, 19.CrossRefGoogle ScholarPubMed
Cheng, LJ, Jiang, Y, Wu, VX, et al. (2020) A systematic review and meta-analysis: vinegar consumption on glycaemic control in adults with type 2 diabetes mellitus. J Adv Nurs 76, 459474.CrossRefGoogle ScholarPubMed
Launholt, TL, Kristiansen, CB, Hjorth, P (2020) Safety and side effects of apple vinegar intake and its effect on metabolic parameters and body weight: a systematic review. Eur J Nutr 59, 22732289.CrossRefGoogle ScholarPubMed
Valdes, DS, So, D, Gill, PA, et al. (2021) Effect of dietary acetic acid supplementation on plasma glucose, lipid profiles, and Body Mass Index in human adults: a systematic review and meta-analysis. J Acad Nutr Diet 121, 895914.CrossRefGoogle ScholarPubMed
Cherta-Murillo, A, Pugh, JE, Alaraj-Alshehhi, S, et al. (2022) The effects of SCFAs on glycemic control in humans: a systematic review and meta-analysis. Am J Clin Nutr 116, 335361.CrossRefGoogle ScholarPubMed
Bhide, A, Acharya, G (2015) When is it appropriate to conduct a(nother) systematic review? Acta Obstet Gynecol Scand 94, 11511152.CrossRefGoogle ScholarPubMed
Chan, M, Baxter, H, Larsen, N, et al. (2019) Impact of botanical fermented foods on metabolic biomarkers and gut microbiota in adults with metabolic syndrome and type 2 diabetes: a systematic review protocol. BMJ Open 9, e029242.CrossRefGoogle ScholarPubMed
Dickersin, K, Scherer, R, Lefebvre, C (1994) Identifying relevant studies for systematic reviews. BMJ 309, 12861291.CrossRefGoogle ScholarPubMed
Higgins, JPT, Thomas, J (2022) Cochrane Handbook for Systematic Reviews of Interventions. https://training.cochrane.org/handbook/current.Google Scholar
Higgins, JPT, Altman, DG, Gotzsche, PC, et al. (2011) The Cochrane collaboration’s tool for assessing risk of bias in randomised trials. BMJ 343, d5928d5928.CrossRefGoogle ScholarPubMed
Page, MJ, McKenzie, JE, Bossuyt, PM, et al. (2021) The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ 372, n71.CrossRefGoogle ScholarPubMed
McGuinness, LA, Higgins, JPT (2021) Risk-of-bias VISualization (robvis): an R package and Shiny web app for visualizing risk-of-bias assessments. Res Synth Methods 12, 5561.CrossRefGoogle Scholar
Kim, EK, An, S-Y, Lee, M-S, et al. (2011) Fermented kimchi reduces body weight and improves metabolic parameters in overweight and obese patients. Nutr Res 31, 436443.CrossRefGoogle ScholarPubMed
An, S-Y, Lee, MS, Choi, YJ, et al. (2012) Favorable effects of kimchi on metabolic parameters and cytokine levels in patients with type 2 diabetes mellitus. Diabetes, 72nd Scientific Sessions of the American Diabetes Association. Philadelphia, PA United States. 61, A190.Google Scholar
An, S-Y, Lee, MS, Jeon, JY, et al. (2013) Beneficial effects of fresh and fermented kimchi in prediabetic individuals. Ann Nutr Metab 63, 111119.CrossRefGoogle ScholarPubMed
Lee, MS, An, S-Y, Jeon, JY, et al. (2013) Beneficial effects of kimchi on metabolic parameters in subjects with metabolic syndrome. Diabetes, 73rd Scientific Sessions of the American Diabetes Association. Chicago, IL United States. 62, A191.Google Scholar
Han, K, Bose, S, Wang, J, et al. (2015) Contrasting effects of fresh and fermented kimchi consumption on gut microbiota composition and gene expression related to metabolic syndrome in obese Korean women. Mol Nutr Food Res 59, 10041008.CrossRefGoogle ScholarPubMed
Wirawanti, IW, Hardinsyah, H, Briawan, D, et al. (2017) Efek intervensi minuman tempe terhadap penurunan kadar low density lipoprotein. J Gizi Dan Pangan 12, 916.CrossRefGoogle Scholar
Afifah, DN, Nabilah, N, Supraba, GT, et al. (2020) The effects of tempeh gembus, an Indonesian fermented food, on lipid profiles in women with hyperlipidemia. Curr Nutr Food Sci 16, 5664.CrossRefGoogle Scholar
Nadia, FS, Wati, DA, Isnawati, M, et al. (2020) The effect of processed tempeh gembus to triglycerides levels and insulin resistance status in women with obesity. Food Res 4, 10001010.CrossRefGoogle Scholar
Somanah, J, Aruoma, OI, Gunness, TK, et al. (2012) Effects of a short-term supplementation of a fermented papaya preparation on biomarkers of diabetes mellitus in a randomized Mauritian population. Prev Med, Dietary Nutraceuticals and Age Management Medicine 54, S90S97.CrossRefGoogle Scholar
Akamine, Y, Millman, JF, Uema, T, et al. (2022) Fermented brown rice beverage distinctively modulates the gut microbiota in Okinawans with metabolic syndrome: a randomized controlled trial. Nutr Res 103, 6881.CrossRefGoogle Scholar
Salehi, SO, Karimpour, F, Imani, H, et al. (2022) Effects of an Iranian traditional fermented food consumption on blood glucose, blood pressure, and lipid profile in type 2 diabetes: a randomized controlled clinical trial. Eur J Nutr 61, 33673375.CrossRefGoogle ScholarPubMed
Kim, TH, Kim, EK, Lee, M-S, et al. (2011) Intake of brown rice lees reduces waist circumference and improves metabolic parameters in type 2 diabetes. Nutr Res 31, 131138.CrossRefGoogle Scholar
Shin, S-K, Kwon, J-H, Jeong, Y-J, et al. (2011) Supplementation of Cheonggukjang and red ginseng Cheonggukjang can improve plasma lipid profile and fasting blood glucose concentration in subjects with impaired fasting glucose. J Med Food 14, 108113.CrossRefGoogle ScholarPubMed
Fujita, H, Yamagami, T, Ohshima, K (2001) Long-term ingestion of a fermented Soybean-derived Touchi-extract with α-glucosidase inhibitory activity is safe and effective in humans with borderline and mild type-2 diabetes. J Nutr 131, 21052108.CrossRefGoogle ScholarPubMed
Back, H-I, Kim, S-R, Yang, J-A, et al. (2011) Effects of Chungkookjang supplementation on obesity and atherosclerotic indices in overweight/obese subjects: a 12-week, randomized, double-blind, placebo-controlled clinical trial. J Med Food 14, 532537.CrossRefGoogle ScholarPubMed
Kondo, H, Sakuyama Tomari, H, Yamakawa, S, et al. (2019) Long-term intake of miso soup decreases nighttime blood pressure in subjects with high-normal blood pressure or stage I hypertension. Hypertens Res 42, 17571767.CrossRefGoogle ScholarPubMed
Jung, SM, Haddad, EH, Kaur, A, et al. (2021) A non-probiotic fermented soy product reduces total and LDL cholesterol: a randomized controlled crossover trial. Nutrients 13, 535.CrossRefGoogle ScholarPubMed
Han, AL, Jeong, SJ, Ryu, MS, et al. (2022) Anti-obesity effects of traditional and commercial Kochujang in overweight and obese adults: a randomized controlled trial. Nutrients 14, 2783.CrossRefGoogle ScholarPubMed
Cha, Y-S, Yang, J-A, Back, H-I, et al. (2012) Visceral fat and body weight are reduced in overweight adults by the supplementation of Doenjang, a fermented soybean paste. Nutr Res Pract 6, 520526.CrossRefGoogle ScholarPubMed
Cha, Y-S, Kim, S-R, Yang, J-A, et al. (2013) Kochujang, fermented soybean-based red pepper paste, decreases visceral fat and improves blood lipid profiles in overweight adults. Nutr Metab 10, 24.CrossRefGoogle ScholarPubMed
Oh, M-R, Park, S-H, Kim, S-Y, et al. (2014) Postprandial glucose-lowering effects of fermented red ginseng in subjects with impaired fasting glucose or type 2 diabetes: a randomized, double-blind, placebo-controlled clinical trial. BMC Complement Altern Med 14, 237.CrossRefGoogle ScholarPubMed
Lim, J-H, Jung, E-S, Choi, E-K, et al. (2015) Supplementation with Aspergillus oryzae-fermented kochujang lowers serum cholesterol in subjects with hyperlipidemia. Clin Nutr 34, 383387.CrossRefGoogle ScholarPubMed
Mohammadi, S, Mazloomi, SM, Niakousari, M, et al. (2022) Evaluating the effects of dark chocolate formulated with micro-encapsulated fermented garlic extract on cardio-metabolic indices in hypertensive patients: a crossover, triple-blind placebo-controlled randomized clinical trial. Phytother Res 36, 17851796.CrossRefGoogle ScholarPubMed
Tenore, GC, Caruso, D, Buonomo, G, et al. (2019) Lactofermented Annurca apple puree as a functional food indicated for the control of plasma lipid and oxidative amine levels: results from a randomised clinical trial. Nutrients 11, 122.CrossRefGoogle ScholarPubMed
Byun, M-S, Yu, O-K, Cha, Y-S, et al. (2016) Korean traditional Chungkookjang improves body composition, lipid profiles and atherogenic indices in overweight/obese subjects: a double-blind, randomized, crossover, placebo-controlled clinical trial. Eur J Clin Nutr 70, 11161122.CrossRefGoogle ScholarPubMed
Nakamura, A, Kitagawa, M, Yamamoto, T, et al. (2020) Fasting blood glucose-lowering effects of salted rice koji (Shiokoji) in mildly hyperglycemic adults – a randomized, double-blind, placebo-controlled parallel-group study. Jpn Pharmacol Ther 48, 215224.Google Scholar
Barcenilla, A, Pryde, SE, Martin, JC, et al. (2000) Phylogenetic relationships of butyrate-producing bacteria from the human gut. Appl Environ Microbiol 66, 16541661.CrossRefGoogle ScholarPubMed
Yamamoto, B, Suzuki, Y, Yonezu, T, et al. (2018) Cha-Koji, comprising green tea leaves fermented with Aspergillus luchuensis var kawachii kitahara, increases regulatory T cell production in mice and humans. Biosci Biotechnol Biochem 82, 885892.CrossRefGoogle ScholarPubMed
Chan, M, Liu, D, Wu, Y, et al. (2021) Microorganisms in whole botanical fermented foods survive processing and simulated digestion to affect gut microbiota composition. Front Microbiol 12, 759708.CrossRefGoogle ScholarPubMed
Wu, T, Chu, X, Cheng, Y, et al. (2021) Modulation of gut microbiota by Lactobacillus casei fermented raspberry juice in vitro and in vivo . Foods 10, 3055.CrossRefGoogle ScholarPubMed
Cheng, I-C, Shang, H-F, Lin, T-F, et al. (2005) Effect of fermented soy milk on the intestinal bacterial ecosystem. World J Gastroenterol WJG 11, 12251227.CrossRefGoogle ScholarPubMed
Inoguchi, S, Ohashi, Y, Narai-Kanayama, A, et al. (2012) Effects of non-fermented and fermented soybean milk intake on faecal microbiota and faecal metabolites in humans. Int J Food Sci Nutr 63, 402410.CrossRefGoogle ScholarPubMed
Chiu, H-F, Chen, Y-J, Lu, Y-Y, et al. (2017) Regulatory efficacy of fermented plant extract on the intestinal microflora and lipid profile in mildly hypercholesterolemic individuals. J Food Drug Anal 25, 819827.CrossRefGoogle ScholarPubMed
Nielsen, ES, Garnas, E, Jensen, KJ, et al. (2018) Lacto-fermented sauerkraut improves symptoms in IBS patients independent of product pasteurisation – a pilot study. Food Funct 9, 53235335.CrossRefGoogle ScholarPubMed
Leeming, ER, Johnson, AJ, Spector, TD, et al. (2019) Effect of diet on the gut microbiota: rethinking intervention duration. Nutrients 11, 2862.CrossRefGoogle ScholarPubMed
Shanahan, ER, McMaster, JJ, Staudacher, HM (2021) Conducting research on diet–microbiome interactions: a review of current challenges, essential methodological principles, and recommendations for best practice in study design. J Hum Nutr Diet 34, 631644.CrossRefGoogle ScholarPubMed
SaeidiFard, N, Djafarian, K, Shab-Bidar, S (2020) Fermented foods and inflammation: a systematic review and meta-analysis of randomized controlled trials. Clin Nutr ESPEN 35, 3039.CrossRefGoogle ScholarPubMed
Gerards, MC, Terlou, RJ, Yu, H, et al. (2015) Traditional Chinese lipid-lowering agent red yeast rice results in significant LDL reduction but safety is uncertain – a systematic review and meta-analysis. Atherosclerosis 240, 415423.CrossRefGoogle ScholarPubMed
Barnard, ND, Willett, WC, Ding, EL (2017) The misuse of meta-analysis in nutrition research. JAMA 318, 14351436.CrossRefGoogle ScholarPubMed
Welch, RW, Antoine, J-M, Berta, J-L, et al. (2011) Guidelines for the design, conduct and reporting of human intervention studies to evaluate the health benefits of foods. Br J Nutr 106, S3S15.CrossRefGoogle ScholarPubMed
Rochlani, Y, Pothineni, NV, Mehta, JL (2015) Metabolic syndrome: does it differ between women and men? Cardiovasc Drugs Ther 29, 329338.CrossRefGoogle ScholarPubMed
WHO Expert Consultation (2004) Appropriate body-mass index for Asian populations and its implications for policy and intervention strategies. Lancet 363, 157163.CrossRefGoogle Scholar
WHO Consultation on Obesity (1999) Obesity: Preventing and Managing the Global Epidemic: Report of a WHO Consultation. Geneva: WHO.Google Scholar
Figure 0

Table 1. PICOS (population, intervention, comparison, outcome, study design) criteria for inclusion of studies

Figure 1

Fig. 1. PRISMA flow diagram of literature search.

Figure 2

Fig. 2. Risk-of-bias summary: domain-level judgements for each individual study.

Figure 3

Fig. 3. Risk-of-bias graph: weighted distribution of risk-of-bias judgements within each bias domain across all included studies.

Figure 4

Table 2. Characteristics of included studies – lactofermented foods

Figure 5

Table 3. Characteristics of included studies – Chungkookjang, kochujang and doenjang (Bacillus and Aspergillus species)

Figure 6

Table 4. Characteristics of included studies – Shiokoji, miso, touchi, amazake and kochujang (Aspergillus oryzae)

Figure 7

Table 5. Characteristics of included studies – tempeh products (Rhizopus species)

Figure 8

Table 6. Characteristics of included studies – others

Figure 9

Fig. 4. Schematic diagram summarising the results of the studies reviewed. Number of studies that had outcomes with significant (p < 0·05) between-group changes post intervention are given in brackets next to the body system affected.

Supplementary material: File

Chan et al. supplementary material

Chan et al. supplementary material

Download Chan et al. supplementary material(File)
File 2.4 MB