Osteoporosis is a metabolic bone disorder characterised by skeletal fragility and fractures even with minor trauma. It affects both sexes, but women are more prone to developing osteoporosis than men. Postmenopausal osteoporosis is the most common type of osteoporosis, caused by the loss of oestrogen after menopause. Roughly 50 % of the postmenopausal women suffer from osteoporosis.
Bone tissue is subjected to continuous cycles of modelling and remodelling where homoeostasis is maintained by a balance between bone formation by osteoblasts and bone resorption by osteoclasts. However, this balance is disrupted after menopause with oestrogen deficiency favouring bone resorption over bone formation. Macroscopically, bone can be classified into two types: cortical bone and trabecular bone. Cortical bone is compact, dense and solid bone, while trabecular bone is a lace-like structure of interconnected trabecular plates and bars surrounding marrow-filled cavities(Reference Brandi1). In osteoporotic conditions, there is a disruption of trabecular continuity due to trabecular perforation, leading to increased bone fragility. In addition, thinning and increased porosity of the cortices occur, with the conversion of plate-like trabeculae into thinner rod-like structures(Reference Kalpakcioglu, Morshed and Engelke2).
Current treatments for osteoporosis aim to improve bone quality and strength by increasing bone formation through anabolic drugs or decreasing bone resorption by antiresorptive agents. Teriparatide and abaloparatide are the only two FDA (Food and Drug Administration)-approved anabolic agents for the treatment of osteoporosis. Both have been shown to reduce the incidence of vertebral and non-vertebral fractures(Reference Neer, Arnaud and Zanchetta3,Reference Miller, Hattersley and Riis4) . Antiresorptive agents include oestrogen, bisphosphonates (e.g. alendronate), selective oestrogen receptor modulators (e.g. raloxifene), human monoclonal antibody against RANKL (denosumab) and strontium ranelate (SR). However, these treatments can cause side effects such as joint and muscle pain, heartburn and urinary tract infections. Additionally, they may have lengthy treatment duration and high cost, which can limit their use.
Recent research on the influence of gut microbiota on a person’s health is providing exciting new insights into the crosstalk between the homoeostasis of bone metabolism and the intestinal flora(Reference Behera, Ison and Tyagi5–Reference Ilesanmi-Oyelere, Roy and Kruger10) and could help in developing new treatment strategies. Probiotic gut bacteria, such as Lactobacillus and Bifidobacterium, have been shown to promote the absorption of minerals such as Ca, Mg and P and thus increase bone mineral density (BMD)(Reference Rodrigues, Castro and Rodrigues11). Probiotics are live micro-organisms that confer health benefits on the host when consumed in adequate amounts(Reference Howarth and Wang12). Lactobacillus, Bifidobacterium, Saccharomyces and Clostridium are few examples of probiotics. The gut microbiome also plays a vital role in the synthesis of vitamin B and K and metabolising bile acids(Reference Clarke, Stilling and Kennedy13), which play a key role in bone health and Ca absorption(Reference Villa, Diaz and Pizziolo14,Reference van Wijngaarden, Doets and Szczecińska15) . These studies point towards the idea that manipulation of the microbiome or its metabolites by the consumption of probiotics may improve bone health and thus prevent or treat osteoporosis. Probiotics function by manipulating the intestinal microbiota and stimulating the proliferation and differentiation of epithelial cells, which can lead to a stronger immune system. They have also been shown to have inhibitory effects on osteoclastic bone resorption(Reference Ohlsson, Engdahl and Fåk16,Reference McCabe, Irwin and Schaefer17) and properties of osteoblastic bone formation(Reference Collins, Rios-Arce and Schepper18,Reference Behera, Ison and Voor19) , which is why probiotics are now being considered as an alternative osteoporosis treatment.
There have been only ten clinical studies investigating the effect of probiotics on bone health, with only five focusing on postmenopausal loss of bone. Among these studies, only one reported a significant improvement in bone mass content following probiotic supplementation, while the others showed smaller effects or no change at all. Several preclinical studies have explored the potential of probiotics in improving bone health, with most conducted on bone loss associated with diseases like diabetes or in the presence of certain drugs. Only few studies have specifically focused on postmenopausal bone loss. Additionally, the results of these studies are contradictory, and the exact mechanisms by which probiotics improve bone health are not well understood. While some studies suggest that probiotics may increase Ca absorption or have a direct effect on bone formation, others propose that they may inhibit bone resorption. Thus, a meta-analysis could help to synthesise the available evidence and determine the overall effect of probiotics on postmenopausal osteoporotic bones. However, there is currently insufficient data from human studies to conduct a meta-analysis, so our analysis was conducted on preclinical studies instead. We also aimed to check which of the two genera of probiotics, Lactobacillus or Bifidobacterium, could have a more positive effect on bone health. Finally, we evaluated whether probiotics exert their positive effect on bone by decreasing bone resorption or increasing bone formation.
Materials and methods
The protocol of this meta-analysis was registered in the systematic review trials registry ‘PROSPERO’ with the registration number – CRD42023445290.
Search strategy
To identify the studies that assessed the effect of probiotics on bone, we conducted a literature search on three electronic databases: PubMed, Google Scholar and Web of Science, until June 2023. We used various combinations of the keywords ‘probiotics’, ‘bone’ and ‘osteoporosis’ in our electronic search. The search was not restricted by language. The process of study identification and selection is represented as a flow chart in Fig. 1.
Fig. 1. Flow chart showing the process of study identification and selection.
Inclusion and exclusion
Inclusion and exclusion criteria were defined, and studies were screened and selected strictly according to that. Inclusion criteria were (i) original and full-length research articles, (ii) studies on primary osteoporosis of the long bone and vertebra, (iii) studies on laboratory animals, (iv) studies where ovariectomy was used to mimic postmenopausal osteoporotic conditions, (v) studies where animals were treated with single probiotic species or a mixture of them and (vi) studies published in English language. Exclusion criteria were (i) review articles/meta-analytical reviews/systematic reviews, (ii) clinical reports and/or trials, (iii) books, (iv) studies on osteoporosis of the jaw bone, (v) studies on secondary osteoporosis, (vi) studies where animals were treated with GMO (Genetically Modified Organism) species or cell-free culture supernatant, (vii) studies on knockout models, (viii) articles where in vitro effects were studied and (ix) studies that failed to provide the required information. There were no restrictions regarding species and duration of probiotic treatment.
Data extraction
Literature was screened independently by the authors, and disagreements, if any, were resolved by discussion. WebPlotDigitizer-4 software was used to extract data in numeric form from bar graphs of the selected research articles. Data were also noted down directly from the tables of the articles. The data were presented in a Microsoft Excel spreadsheet where we recorded PubMed Identifier (PMID) and authors of the study, species and strains used, age and sample size of the control and treated groups, probiotic species used for treatment, treatment method along with its duration and the bone parameters that were measured, that is, bone mineral density (BMD), bone volume fraction (BV/TV), bone formation rate (BFR), serum osteocalcin (OCN), serum C-terminal telopeptide (CTX-1), serum alkaline phosphatase (ALP), serum calcium (Ca) and osteoclast surface by bone surface (Oc.S/B.S).
Outcome assessment
The studies were categorised into two groups: probiotic and control group. BMD and BV/TV data were primarily categorised into three groups based on the type of bone: femur, tibia or vertebra. Under BMD, wherever applicable, each group was further divided into three based on the region considered, that is, total, trabecular or cortical bone. Further, analysis was done on BFR, bone formation markers (like serum osteocalcin, serum ALP and serum Ca) and bone resorption markers (serum CTX-1). Effect of probiotics on Oc.S/B.S was also analysed.
Quantitative data analysis
Pooled data analysis was conducted using Review Manager 5.4 software (RevMan) and Jeffreys’s Amazing Statistics Program 0.16.3.0 (JASP). Studies where more than one species of probiotic group was involved were split to include only one probiotic group per analysis. The effect size chosen was standardised mean difference (SMD) also known as Cohen’s d. Heterogeneity index (I2 statistic) was used for assessing heterogeneity across studies, and P-values of < 0·1 were considered to be statistically significant due to the low stringency of this test(Reference Higgins, Chandler and Cumpston20). I2 value of < 25 % was considered as low heterogeneity, 50 % as moderate heterogeneity and > 75 % as high heterogeneity(Reference Cochran21). Based on the level of heterogeneity, the summary/pooled effect size was calculated either using random effects model or fixed effects model in RevMan(Reference Vesterinen, Sena and Egan22). In addition to this, data obtained from RevMan, that is, effect size and CI of the individual studies, were entered in Microsoft Excel, standard error of effect size was calculated and csv files were created. These were loaded in JASP, and classical meta-analysis was performed to obtain pooled effect sizes again. Here also, based on heterogeneity, fixed or random effects model was used. The method for meta-analysis used in the case of random effects model was restricted maximum likelihood method.
Assessment of risk of bias
Risk of bias was assessed using the Systematic Review Centre for Laboratory Animal Experimentation (SYRCLE) bias risk tool(Reference Hooijmans, Rovers and de Vries23). This tool consists of ten entries covering various biases like selection bias, performance bias, outcome bias, detection bias, attrition bias, reporting bias and other bias. For each entry, studies were categorised as high, low or unclear risk of bias by both the authors independently taking the help of signalling questions provided in the tool for assisting judgement. Disagreements, if any, were resolved by discussion.
Publication bias analysis
Publication bias analysis was carried out qualitatively on the basis of funnel plot asymmetry and quantitatively on the basis of Egger’s intercept test using the JASP program. In the case of publication bias, the unbiased estimates were computed using Duval and Tweedie’s Trim and Fill method using the JASP program.
Results
Design of the study and parameters measured
Literature search revealed that the effect of probiotics on postmenopausal bone health has been studied majorly on animal models, that is, mouse and rat. In contrast, there were only ten studies conducted on human subjects. After the removal of duplicates, we found a total of 135 hits in our initial search which included fifty-five research articles and the rest were clinical trials, books, review articles, systematic reviews and meta-analytical reviews. After screening the title and abstract of the articles according to our inclusion/exclusion criteria, we selected twenty-two research articles that were relevant to our meta-analytical study. Of these, eleven studies each were conducted on mice and rats. All the eleven studies on rats used the strain Sprague Dawley(Reference Narva, Rissanen and Halleen24–Reference Lee, Kim and Kim34). A total of 238 ovariectomised rats were used; 95 in control and 143 in experimental group. Among the eleven studies on mice, four studies used Balb/c mice(Reference Britton, Irwin and Quach35–Reference Wallimann, Hildebrand and Groeger38), one used Institute of Cancer Research (ICR) mice(Reference Yang, Lin and Li39), five used C57Bl/6 mice(Reference Ohlsson, Engdahl and Fåk16,Reference Yu, Hang and Sun40–Reference Sapra, Shokeen and Porwal43) and one used ddy mouse strain(Reference Yamada, Park and Node44). A total of 215 ovariectomised mice were used, of which 88 were in control group and 127 in experimental group. Fifteen studies used Lactobacillus as the probiotic of choice(Reference Ohlsson, Engdahl and Fåk16,Reference Narva, Rissanen and Halleen24–Reference Lim, Song and Kim29,Reference Lee, Jung and Park33–Reference Dar, Shukla and Mishra37,Reference Yang, Lin and Li39–Reference Chiang and Pan41) , three used Bifidobacterium (Reference Parvaneh, Ebrahimi and Sabran31,Reference Wallimann, Hildebrand and Groeger38,Reference Sapra, Shokeen and Porwal43) , and Bacillus (Reference Dar, Pal and Shukla32) and Saccharomyces (Reference Yamada, Park and Node44) were used by one study each. In one study, treatment groups included Lactobacillus, Bacillus or Bifidobacterium supplementation(Reference Montazeri-Najafabady, Ghasemi and Dabbaghmanesh30), and in another study, treatment groups included Lactobacillus, Bifidobacterium or a mixture of Lactobacillus and Bifidobacterium (Reference Kim, Kim and Han42). Probiotic supplementation to the animals was carried out in three different ways. In six studies, animals were treated with the probiotic in the form of its fermented product/extract(Reference Narva, Rissanen and Halleen24,Reference Shim, Kim and Ha26–Reference Shim, Kim and Ha28,Reference Lee, Kim and Kim34,Reference Chiang and Pan41) . In one study, diet of the animals was supplemented with the probiotic(Reference Yamada, Park and Node44). In the remaining fifteen studies, probiotics were orally gavaged to the animals(Reference Ohlsson, Engdahl and Fåk16,Reference Parvaneh, Karimi and Jamaluddin25,Reference Lim, Song and Kim29–Reference Lee, Jung and Park33,Reference Britton, Irwin and Quach35–Reference Yu, Hang and Sun40,Reference Kim, Kim and Han42,Reference Sapra, Shokeen and Porwal43) . The duration of the treatment varied from study to study, ranging from 4 weeks to 3 months. A flow diagram for the search, screening, eligibility and inclusion/exclusion of the studies is given in Fig. 1. Table 1 provides the general characteristics of the included studies, while Table 2 shows the parameters measured in each study for the meta-analysis.
Table 1. Characteristics of the included studies that reported impact of probiotic consumption on bone health in animal models
OVX, ovariectomised; CFU, Colony Forming Unit; VPP, Valyl-Prolyl-Proline; HRT, Hwangryun-haedok-tang; KFRI, Korea Food Research Institute; CMC, carboxymethylcellulose.
Table 2. Bone-related parameters reported by the studies included in our meta-analysis
BMD, bone mineral density; BV/TV, bone volume fractions; BFR, bone formation rate; OCN, Osteocalcin; ALP, alkaline phosphatase; Ca, calcium; CTX-1, C-terminal telopeptide; Ob.S/B.S osteoblast surface by bone surface; Oc.S/B.S, osteoclast surface by bone surface.
Outcomes
Probiotics significantly increase bone mineral density of ovariectomised animals
To understand the effect of probiotics on the BMD, we analysed data from all the studies done on probiotics in the ovariectomy model irrespective of the type of bacteria. There were seventeen studies that had reported effect of probiotics on BMD. The studies were divided into five groups based on the type of bone and region included: femur total BMD(Reference Parvaneh, Karimi and Jamaluddin25,Reference Montazeri-Najafabady, Ghasemi and Dabbaghmanesh30) , femur trabecular BMD(Reference Ohlsson, Engdahl and Fåk16,Reference Shim, Kim and Ha26–Reference Lim, Song and Kim29,Reference Parvaneh, Ebrahimi and Sabran31,Reference Dar, Pal and Shukla32,Reference Lee, Kim and Kim34,Reference Britton, Irwin and Quach35,Reference Dar, Shukla and Mishra37,Reference Yang, Lin and Li39–Reference Chiang and Pan41) , tibia trabecular BMD(Reference Narva, Rissanen and Halleen24,Reference Dar, Pal and Shukla32,Reference Dar, Shukla and Mishra37) , tibia cortical BMD(Reference Narva, Rissanen and Halleen24,Reference Dar, Pal and Shukla32,Reference Dar, Shukla and Mishra37) and vertebral BMD(Reference Montazeri-Najafabady, Ghasemi and Dabbaghmanesh30,Reference Dar, Pal and Shukla32,Reference Britton, Irwin and Quach35,Reference Dar, Shukla and Mishra37,Reference Sapra, Shokeen and Porwal43) . A pooled analysis was conducted for all groups except tibia cortical BMD using a random effects model since there was high heterogeneity. Pooled analysis revealed that probiotic treatment favours an increase in BMD significantly at all the sites, with the highest effect seen in femur total BMD (SMD = 5·24, 95 % CI (3·04, 7·43); P < 0·001). The result is summarised in Fig. 2 and Table 3.
Fig. 2. Forest plot depicting pooled effect analysis on BMD of (a) femur total (b) femur trabecular (c) tibia trabecular (d) vertebra and (e) tibia cortical.
Table 3. Summary of results of pooled analysis conducted on BMD and BV/TV data
BMD, bone mineral density; BV/TV, bone volume fractions; SMD, standardised mean difference.
Probiotics significantly improve bone volume fractions in the bones of ovariectomised animals
Next, we investigated whether probiotics are able to improve bone health by increasing the bone volume in ovariectomised animals. Out of twenty-two selected studies, sixteen reported the effects of probiotics on bone volume. Again, we grouped the studies based on the type of bone: femur BV/TV(Reference Ohlsson, Engdahl and Fåk16,Reference Parvaneh, Karimi and Jamaluddin25–Reference Shim, Kim and Ha28,Reference Parvaneh, Ebrahimi and Sabran31,Reference Dar, Pal and Shukla32,Reference Britton, Irwin and Quach35–Reference Dar, Shukla and Mishra37,Reference Yang, Lin and Li39–Reference Chiang and Pan41,Reference Sapra, Shokeen and Porwal43) , tibia BV/TV(Reference Narva, Rissanen and Halleen24,Reference Sapra, Dar and Bhardwaj36,Reference Dar, Shukla and Mishra37,Reference Sapra, Shokeen and Porwal43) and vertebra BV/TV(Reference Britton, Irwin and Quach35–Reference Dar, Shukla and Mishra37,Reference Sapra, Shokeen and Porwal43,Reference Yamada, Park and Node44) . We conducted a pooled effect analysis for the three groups using a random effects model and found that probiotic treatment significantly increased bone volume compared with the control (Fig. 3).
Fig. 3. Forest plot depicting pooled effect analysis on BV/TV of (a) femur (b) tibia and (c) vertebra.
Table 3 shows the summary of the results of the pooled analysis conducted on BV/TV data. For femur BV/TV, the model yielded a SMD of 1·399 with a 95 % CI of (0·83, 1·96) and P-value of < 0·001. The I2 statistic showed a high degree of heterogeneity (75·98 %, P-value < 0·001). For tibia BV/TV, the model yielded an SMD of 2·76 with a 95 % CI of (0·43, 5·09) and P-value of 0·02. The I2 statistic showed a very high degree of heterogeneity (90·7 %, P-value < 0·001). For vertebral BV/TV, the model yielded an SMD of 2·26 with a 95 % CI of (0·75, 3·77) and P-value of 0·003. The I2 statistic showed a high degree of heterogeneity (85·76 %, P-value = 0·001).
In summary, our findings demonstrate that probiotics significantly improve BV/TV in ovariectomised animals, with the most significant effects observed in the tibia and vertebra.
Probiotics show potential to increase serum bone formation markers
To further investigate the effect of probiotic treatment on bone formation, we examined the data from twenty-two selected articles that reported serum bone formation markers levels. Of these, six studies reported on serum osteocalcin(Reference Ohlsson, Engdahl and Fåk16,Reference Parvaneh, Karimi and Jamaluddin25,Reference Parvaneh, Ebrahimi and Sabran31,Reference Lee, Jung and Park33,Reference Lee, Kim and Kim34,Reference Kim, Kim and Han42) , seven studies on serum Ca(Reference Ohlsson, Engdahl and Fåk16,Reference Parvaneh, Karimi and Jamaluddin25,Reference Lee, Kim and Lim27,Reference Montazeri-Najafabady, Ghasemi and Dabbaghmanesh30,Reference Parvaneh, Ebrahimi and Sabran31,Reference Chiang and Pan41,Reference Kim, Kim and Han42) and six studies on serum ALP(Reference Lee, Kim and Lim27,Reference Lim, Song and Kim29,Reference Montazeri-Najafabady, Ghasemi and Dabbaghmanesh30,Reference Lee, Kim and Kim34,Reference Yu, Hang and Sun40,Reference Chiang and Pan41) . Pooled effect analysis was carried out using random effects model due to sufficient heterogeneity among studies. Results showed that probiotic treatment favoured a trend towards increasing all three serum bone formation markers (Fig. 4). However, the results were not statistically significant. A summary of the results is depicted in Table 4.
Fig. 4. Forest plot depicting pooled effect analysis on serum bone formation markers: (a) serum osteocalcin, (b) serum calcium, (c) serum ALP, and serum bone resorption marker: (d) serum CTX-1.
Table 4. Summary of the results of pooled analysis conducted on serum bone markers and histomorphometric parameters of bone turnover data
SMD, standardised mean difference; ALP, alkaline phosphatase; CTX-1, C-terminal telopeptide; BFR, bone formation rate; Oc.S/B.S, osteoclast surface by bone surface.
Probiotics show potential to decrease bone resorption marker – serum C-terminal telopeptide levels
Of the twenty-two selected articles, we identified five studies that reported on serum CTX-1 levels after probiotic treatment(Reference Ohlsson, Engdahl and Fåk16,Reference Parvaneh, Karimi and Jamaluddin25,Reference Parvaneh, Ebrahimi and Sabran31,Reference Lee, Kim and Kim34,Reference Wallimann, Hildebrand and Groeger38) . Two of these were split to make two studies each(Reference Ohlsson, Engdahl and Fåk16,Reference Lee, Kim and Kim34) , since they showed the effect of two different probiotic groups on CTX-1 levels. While five studies showed a decrease in serum CTX-1 levels(Reference Parvaneh, Karimi and Jamaluddin25,Reference Parvaneh, Ebrahimi and Sabran31,Reference Lee, Kim and Kim34,Reference Wallimann, Hildebrand and Groeger38) in the probiotic-treated group, two studies showed an increase(Reference Ohlsson, Engdahl and Fåk16). Random effects model analysis revealed that overall, probiotic treatment tended to decrease serum CTX-1 levels, but the effect was not significant (Fig. 4, Table 4), possibly due to high heterogeneity (I2 = 93 %).
Probiotic treatment reduced bone formation rate, but not significantly
Next, we aimed to investigate if treatment with probiotics had any effect on the BFR. We found only two studies(Reference Ohlsson, Engdahl and Fåk16,Reference Britton, Irwin and Quach35) among the twenty-two selected studies that reported the impact of probiotics on BFR. The heterogeneity between these two studies was sufficiently low, and thus we conducted a meta-analysis using fixed effects model. Pooled effect analysis revealed a reduction in BFR upon probiotic treatment (SMD = −0·66). However, the effect was not statistically significant (CI (–1·34, 0·02); P = 0·056) (Fig. 5, Table 4).
Fig. 5. Forest plot depicting pooled effect analysis on bone histomorphometric parameters: (a) BFR and (b) Oc.S/B.S.
Probiotic treatment shows a trend towards reduced osteoclast surface/bone surface, but not significantly
To investigate the effects of probiotics on bone-forming and bone-resorbing cells, specifically osteoblasts and osteoclasts, we conducted a literature search and identified two studies that reported the impact of probiotics on osteoblast surface (Ob.S/B.S)(Reference Parvaneh, Karimi and Jamaluddin25,Reference Parvaneh, Ebrahimi and Sabran31) and three studies that reported the impact on Oc.S/B.S(Reference Narva, Rissanen and Halleen24,Reference Parvaneh, Karimi and Jamaluddin25,Reference Parvaneh, Ebrahimi and Sabran31) . However, due to high heterogeneity between the two studies, a meta-analysis could not be performed on Ob.S/B.S data(Reference Ryan45). Among the three studies reporting Oc.S/B.S data, two studies showed a decrease in osteoclast surface upon probiotic treatment(Reference Parvaneh, Karimi and Jamaluddin25,Reference Parvaneh, Ebrahimi and Sabran31) , while one showed an increase(Reference Narva, Rissanen and Halleen24). Pooled analysis results using random effects model showed that probiotic treatment favoured a decrease in Oc.S/B.S, but the effect was not significant (Fig. 5). A summary of the results is depicted in Table 4.
Comparative analysis of Lactobacillus and Bifidobacterium probiotics on bone health
From all the twenty-two articles selected for this meta-analysis, the majority of studies used either Lactobacillus or Bifidobacterium, which are also the major genera of gut microbiota. We therefore aimed to determine which of these probiotic genera was more effective in improving bone health. Thus, we segregated the studies based on these two probiotic genera and carried out meta-analysis. The parameters that we took into consideration were BMD, BV/TV and bone formation marker, that is, serum Ca.
Bone mineral density;
We found three articles reporting on vertebral BMD among those using Lactobacillus as the probiotic(Reference Montazeri-Najafabady, Ghasemi and Dabbaghmanesh30,Reference Britton, Irwin and Quach35,Reference Dar, Shukla and Mishra37) and two articles using Bifidobacterium as the probiotic(Reference Montazeri-Najafabady, Ghasemi and Dabbaghmanesh30,Reference Sapra, Shokeen and Porwal43) . Unfortunately, due to a lack of sufficient studies or homogeneity, we were unable to include BMD of other regions such as femur trabecular, femur cortical and tibia trabecular. As usual, articles involving more than one species were split such that each study included only one species. Pooled effect analysis for BMD studies on Lactobacillus was carried out using random effects model and that on Bifidobacterium was carried out using fixed effects model. In both the cases, pooled analysis results showed that probiotics increased the BMD over control (Fig. 6). However, the effect size was larger for Lactobacillus (SMD = 2·643) compared with Bifidobacterium (SMD = 1·38). Results are summarised in Table 5.
Fig. 6. Forest plot depicting comparative analysis of (a) vertebral BMD- using Lactobacillus (b) vertebral BMD- using Bifidobacterium (c) femur BV/TV- using Lactobacillus (d) femur BV/TV- using Bifidobacterium (e) serum calcium- using Lactobacillus and (f) serum calcium- using Bifidobacterium.
Table 5. Comparative analysis of Lactobacillus and Bifidobacterium
BMD, bone mineral density; SMD, standardised mean difference; BV/TV, bone volume fractions.
Bone volume fractions
Femur BV/TV was reported by ten articles using Lactobacillus (Reference Ohlsson, Engdahl and Fåk16,Reference Parvaneh, Karimi and Jamaluddin25–Reference Shim, Kim and Ha28,Reference Britton, Irwin and Quach35–Reference Dar, Shukla and Mishra37,Reference Yang, Lin and Li39,Reference Chiang and Pan41) and by two articles using Bifidobacterium as the probiotic(Reference Parvaneh, Ebrahimi and Sabran31,Reference Sapra, Shokeen and Porwal43) . Analysis on BV/TV of tibia and vertebra could not be conducted due to the insufficient number of studies using Bifidobacterium. Random effects model was used to carry out analysis on Lactobacillus group and fixed effects model was used on Bifidobacterium group. Similar to that of BMD data, here too pooled analysis results showed that both Lactobacillus- and Bifidobacterium-treated groups favoured an increase in BV/TV over control (Fig. 6). However, the effect size for Lactobacillus was larger than that for Bifidobacterium. Results are summarised in Table 5.
Bone formation marker
Among the previously selected articles, serum osteocalcin was reported in five articles using Lactobacillus species(Reference Ohlsson, Engdahl and Fåk16,Reference Parvaneh, Karimi and Jamaluddin25,Reference Lee, Jung and Park33,Reference Lee, Kim and Kim34,Reference Kim, Kim and Han42) and two articles using Bifidobacterium species(Reference Parvaneh, Ebrahimi and Sabran31,Reference Kim, Kim and Han42) . However, due to high heterogeneity, meta-analysis could not be performed on Bifidobacterium. Similarly, serum ALP was reported in six articles using Lactobacillus (Reference Lee, Kim and Lim27,Reference Lim, Song and Kim29,Reference Montazeri-Najafabady, Ghasemi and Dabbaghmanesh30,Reference Lee, Kim and Kim34,Reference Yu, Hang and Sun40,Reference Chiang and Pan41) and only one article using Bifidobacterium (Reference Montazeri-Najafabady, Ghasemi and Dabbaghmanesh30), preventing a comparison of the two species.
Serum Ca was reported in six articles using Lactobacillus species(Reference Ohlsson, Engdahl and Fåk16,Reference Parvaneh, Karimi and Jamaluddin25,Reference Lee, Kim and Lim27,Reference Montazeri-Najafabady, Ghasemi and Dabbaghmanesh30,Reference Chiang and Pan41,Reference Kim, Kim and Han42) . These were split into separate studies for each species, resulting in a total of ten studies. Of these, six showed an increase while four showed a decrease in serum Ca upon treatment. In contrast, all three studies using Bifidobacterium (Reference Montazeri-Najafabady, Ghasemi and Dabbaghmanesh30,Reference Parvaneh, Ebrahimi and Sabran31,Reference Kim, Kim and Han42) reported an increase in serum Ca after treatment. Pooled effect analysis was carried out on both of them using random effects model, and the results showed that both the probiotics had increased serum Ca over control (Fig. 6). However, the results were not statistically significant. A summary of the results is presented in Table 5.
Assessment of risk of bias
The twenty-two included studies were assessed for their risk of bias using the SYRCLE tool (online Supplementary Fig. 4). Briefly, none of the studies fulfilled all the ten criteria required for low risk of bias. Only one study accurately described the random sequence generation method used, and thus, the selection bias in the randomisation entry for all the other studies was judged as ‘unclear risk’. The majority of the studies had similar baseline characteristics between control and experimental groups. The risk of bias was unclear for all the studies regarding allocation concealment, random housing and random outcome assessment. Blinding of outcome assessment was mentioned in only two studies and thus allocated a low-risk label. All the studies were allocated a low-risk label for attrition and reporting bias.
Publication bias
Publication bias was carried out qualitatively based on the funnel plot test and quantitatively based on Egger’s test. We found significant publication bias for femur total BMD (P < 0·001), femur trabecular BMD (P < 0·001), vertebral BMD (P < 0·001), BV/TV (femur, tibia and vertebral) (P < 0·001), serum Ca (P < 0·001) and Oc.S/B.S (P = 0·001). We used Trim and Fill method to compute unbiased estimates for the above. However, the estimates for femur total BMD, vertebral BMD, femur BV/TV, vertebral BV/TV, serum Ca and Oc.S/B.S did not change after using Trim and Fill method. Adjusted estimates, using Trim and Fill method, were obtained for femur trabecular BMD (summary estimate: 1·14, 95 % CI 0·73, 1·55) and tibia BV/TV (summary estimate: 2·0, 95 % CI −0·37, 4·38). Publication bias was absent for tibia trabecular BMD, tibia cortical BMD, serum osteocalcin, serum CTX-1 and serum ALP. Funnel plots and P-value for Egger’s test conducted are provided in the online Supplementary figures.
Discussion
Postmenopausal osteoporosis is a serious health issue in women, and it is a major public health concern worldwide. In recent years, the use of probiotics has gained popularity as a complementary therapy for various disorders. The regulation of bone homoeostasis by probiotics is proposed to be achieved through their immunomodulatory ability, which is mediated by the production of SCFA or the direct management of intestinal permeability(Reference Li, Pi and Li46). SCFA have been shown to stimulate the expansion of regulatory T cells(Reference Davie47,Reference Sanford, Zhang and Williams48) that, along with CD8 + T cells, modulate the production of Wnt10b(Reference Tyagi, Yu and Darby49). Wnt10b acts on stromal cells and osteoblasts to promote bone formation(Reference Bodine, Zhao and Kharode50). In addition to their role in regulating regulatory T cells, probiotics also help maintain the integrity of intestinal epithelial cells(Reference Ohland and Macnaughton51), which is critical for separating commensal bacteria from mucosal immune cells. In osteoporosis, the integrity of intestinal epithelial cells is often compromised(Reference Guan, Luo and Liu52), leading to the expansion of Th17 cells that produce osteoclastogenic inflammatory cytokines(Reference Li, Rao and Cheng53). These cytokines promote the formation of bone-resorbing osteoclasts, leading to bone loss. Probiotics have been shown to improve epithelial barrier function and restrict the expansion of Th17 cells(Reference Tanabe54), thereby preventing the development of osteoclastogenic inflammation and bone loss.
In this meta-analysis, we aimed to investigate the effect of probiotics on postmenopausal bone health by analysing the available data from both animal and human studies. However, most of the available research are carried out on animal models, and the human studies, although conducted, are inadequate for meta-analysis. Therefore, we performed our meta-analysis with the studies in the preclinical models of postmenopausal osteoporosis. Ovariectomised rodents are a well-established and preferred experimental model for investigating postmenopausal osteoporosis, as it closely mimics the disease’s characteristics. This preclinical model exhibits significantly reduced BMD and BV/TV in the femur, tibia and vertebra, as compared with control animals (sham-operated). Furthermore, ovariectomised animals demonstrate increased bone turnover, which can be evaluated by analysing bone cell activity through histology or by measuring serum biochemical markers of bone formation and resorption(Reference Turner55).
BMD is a measure of the amount of minerals contained in a certain volume of bone. About 50 % of trabecular and 30 % of cortical bone is lost by women during the course of their lifetime, about half of which is lost during the first 10 years after menopause(Reference Finkelstein, Brockwell and Mehta56,Reference Riggs and Melton57) . Our meta-analysis found that probiotics significantly increase BMD in ovariectomised animals at all sites except tibia cortical. The highest effect was observed in femur total BMD.
To address concerns about the potential for errors in BMD measurements due to high variability(Reference Nguyen, Sambrook and Eisman58), we also examined the more reliable micro-CT data performed in animal models, specifically, BV/TV. BV/TV is defined as the volume of mineralised bone per unit volume of the sample and is a preferred measurement of bone quality, as it provides a more accurate representation of the actual bone volume. The pooled effect analysis using a random effects model showed that probiotic treatment significantly increases bone volume compared with the control in all the studied skeletal sites. The most significant effects were observed in the tibia and vertebra.
The currently available treatments for postmenopausal osteoporosis can be broadly classified into two categories: antiresorptives and anabolics. While antiresorptives inhibit bone resorption, anabolics increase bone formation, thereby enhancing bone mass(Reference Tu, Lie and Wan59). To gain further insight into the mechanism of action of probiotics, we investigated their impact on the hallmarks of bone formation and resorption. To analyse the effect of probiotic treatment on bone formation, we examined data that reported serum bone formation marker levels like serum osteocalcin, Ca and ALP. Osteocalcin is a protein produced by osteoblasts during bone formation(Reference Singer and Eyre60,Reference Ivaska, Hentunen and Vääräniemi61) , Ca is an essential component of bone mineralisation(Reference Walker, Hall and Hurst62) and ALP is an enzyme that plays a vital role in bone mineralisation(Reference Harris63), about 50 % of the total ALP activity in serum of normal healthy adults arises from bone(Reference Green, Anstiss and Fishman64). A pooled effect analysis using a random effects model showed that probiotic treatment tended to increase all three serum bone formation markers, but the effects were not statistically significant.
Similarly, data were analysed to evaluate the effect of probiotics on bone resorption by analysing serum CTX-1 levels. CTX of fibrillar collagens such as collagen type I and type II are peptide fragments that are produced during bone resorption, and hence serum CTX-1 represents an important marker of the same(Reference Risteli and Risteli65). The findings indicated that, while not statistically significant, probiotics treatment had a suppressing effect on bone resorption.
We further analysed data to evaluate the effect of probiotics on BFR, which is the amount of mineralised bone formed per unit time per unit volume of bone surface. BFR slows down in osteoporosis, and decreased BFR can lead to bone fragility and increased fractures(Reference Malluche, Davenport and Lima66–Reference Arlot, Delmas and Chappard68). Pooled effect analysis on the BFR data reported in two studies using fixed effects model revealed a decrease in BFR after probiotic treatment which was not statistically significant. However, accuracy of the result obtained might be affected due to involvement of only two studies.
Increased bone resorption by osteoclasts and inability of osteoblasts to make up for this bone loss leads to weakening of bone in osteoporosis(Reference Udagawa, Takahashi and Jimi69–Reference Xiong, Piemontese and Onal71). Finally, we investigated whether probiotic treatment could decrease the osteoclast surface and identified three studies that reported Oc.S/B.S data. Pooled effect analysis showed a decrease in osteoclast surface over bone surface upon probiotic treatment. This was again statistically not significant.
Together, these results revealed that probiotic treatment has a positive effect on bone mass in postmenopausal animal models of osteoporosis, evidenced by increases in BMD and BV/TV in the femur, tibia and vertebrae. Additionally, probiotic treatment displayed a tendency towards increased serum bone formation markers and decreased bone resorption markers. It should be noted, however, that the limited number of studies reporting serum bone turnover markers and the high heterogeneity among the studies mean that these results should be interpreted with caution. Further research is necessary to validate the effects of probiotics on serum bone turnover markers.
Although the primary aim of this meta-analytical study was to assess the overall effect of probiotic supplementation on osteoporotic bone, we also checked if a particular probiotic genus produced a better overall effect over the other. The majority of studies selected for this meta-analysis used either Lactobacillus sp. or Bifidobacterium sp., which are the major genera of gut microbiota(Reference Tap, Mondot and Levenez72). The unavailability of sufficient data limited our analysis to only BMD, BV/TV and serum Ca. The results of the analysis showed that both Lactobacillus and Bifidobacterium probiotics were effective in improving bone health, as seen in the increase in BMD and BV/TV over control. However, the effect size was larger for Lactobacillus compared with Bifidobacterium, indicating that Lactobacillus may be more effective in improving bone health. It is important to note that the number of studies reporting on Bifidobacterium is limited, and further studies are needed to confirm these findings.
With respect to serum Ca, both probiotics showed an increase, although not statistically significant. Notably, in studies using Bifidobacterium, all three reported an increase, while only six out of ten studies using Lactobacillus reported an increase. This suggests Bifidobacterium may be more effective in raising serum Ca levels. However, due to significant study heterogeneity, further research is required for confirmation.
Overall, our study has several strengths, including the exclusive focus on preclinical models of postmenopausal osteoporosis, which allowed us to obtain a comprehensive understanding of the effect of probiotics on postmenopausal bone loss. Moreover, we investigated multiple outcomes related to bone health, including BMD, BV/TV, BFR and serum bone turnover markers to improve our understanding of the effect of probiotics on osteoporotic bone. However, we acknowledge several limitations that should be considered. Firstly, most of the studies included in our analysis were conducted on animal models, which may limit the generalisability of our findings to humans. Secondly, the duration of treatment varied widely among the studies, which could have influenced our results. Thirdly, the heterogeneity of the studies included in our analysis was high, which could have affected the accuracy of our results. Finally, the number of studies reporting serum bone formation/resorption markers was limited, and our results should be interpreted with caution.
Future studies should focus on conducting randomised controlled trials on humans to further investigate the effect of probiotics on postmenopausal bone health. These studies should also aim to standardise the duration and dosage of probiotic treatment to obtain more accurate and reliable results. Additionally, future studies should aim to identify the specific strains of probiotics that are most effective in improving bone health and the mechanisms by which they exert their effects. This information would be useful in the development of probiotic-based therapies for the prevention and treatment of osteoporosis.
In conclusion, our meta-analysis suggests that probiotics have the potential to improve postmenopausal bone health in the preclinical models of the disease. Probiotic supplementation could be a simple and safe strategy for preventing or delaying the onset of osteoporosis in women after menopause. However, more studies are needed to confirm these findings and to identify the specific strains of probiotics and mechanisms involved in the observed effects.
Acknowledgements
This study was supported by funding from the Science and Engineering Research Board (SERB), Government of India (KS). Research fellowship from the Indian Council of Medical Research-Junior Research Fellowship (SB) is also acknowledged.
S. B.: Investigation, Data curation and Writing – original draft preparation. K. S.: Writing – review and editing, Supervision, and Study design.
Shibani Bose and Kunal Sharan declare that they have no conflict of interest.
The data that support the findings of this study are available from the corresponding author (KS) upon reasonable request.
Supplementary material
For supplementary material/s referred to in this article, please visit https://doi.org/10.1017/S0007114523002362
