Randomized study of Lacticaseibacillus fermented milk in Indonesian elderly houses: Impact on gut microbiota and gut environment

Abstract

BACKGROUND

Health maintenance in elderly houses includes management of the gut microbiota and the environment. Lacticaseibacillus paracasei Shirota (LcS) is a probiotic strain that positively affects the human gut. However, the evidence of its effects on the Indonesian population remains limited.

AIM

To investigate the effect of LcS-fermented milk on the gut microbiota and environment of Indonesian elderly houses.

METHODS

This double-blind, randomized, placebo-controlled trial involved 112 participants from Indonesian elderly houses, spanning a 2-week baseline and 24-week treatment. Participants were randomly assigned to probiotic or placebo groups, consuming fermented milk with or without LcS (> 6.5 × 10⁹ colony-forming units). Fecal samples were collected every three months. Gut microbiota analysis was performed using 16S rRNA gene sequencing and reverse transcription quantitative polymerase chain reaction, while gut environment was assessed by measuring fecal organic acids, amino acid metabolites, and stool frequency.

RESULTS

Analyses of 16S rRNA gene sequence data at the 3-month period revealed increased Bifidobacterium and Succinivibrio and decreased Rikenellaceae RC9 gut group in the probiotic group. These shifts were associated with significant differences in β-diversity metrics. The change in Bifidobacterium was confirmed by reverse transcription quantitative polymerase chain reaction, demonstrating higher abundance in the probiotic group than in the placebo group (8.5 ± 1.1 vs 8.0 ± 1.1, log₁₀ bacterial cells/g; P = 0.044). At 6-month period, the differences in Succinivibrio and Rikenellaceae RC9 gut group persisted. The probiotic group showed higher butyrate levels than the placebo group at the 6-month period (5.04 ± 3.11 vs 3.95 ± 2.89, μmol/g; P = 0.048). The effect on amino acid metabolites and stool frequency was not significant.

CONCLUSION

Daily intake of LcS positively affects the gut microbiota and environment of people living in Indonesian elderly houses.

Key Words: Lacticaseibacillus paracasei Shirota; Fermented milk; Intestinal microbiota; Elderly; Probiotics

Core Tip: The daily consumption of fermented milk containing Lacticaseibacillus paracasei Shirota significantly altered the gut microbiota among individuals in Indonesian elderly houses. Notably, these changes included increases in Bifidobacterium and Succinivibrio and a decrease in the Rikenellaceae RC9 gut group. These modifications are associated with an improvement in the gut environment, as indicated by elevated butyrate levels.

INTRODUCTION

The increase in life expectancy is driving a substantial demographic shift toward an aging population worldwide. Indonesia, ranked the fourth most populous country, is not exempt from this phenomenon. The proportion of the elderly population (aged 60 years and older) is expected to increase to 11% by 2035[1]. The elderly population is growing at an annual rate of 4.7%, which exceeds the general population growth rate of 2.9%[2]. Simultaneously, the country’s socioeconomic and environmental development has resulted in changes to traditional community structures, prompting the elderly to relocate from rural to urban areas, where they often live independently. Given these demographic and lifestyle dynamics, the significance of elderly facilities (or houses) is increasing, playing a crucial role in preserving the health and well-being of the aging population.

The gut microbiome and environment are crucial factors affecting the health of elderly individuals. Unhealthy aging has been associated with a decline in gut microbial diversity, characterized by a reduction in beneficial bacteria (e.g., Bifidobacterium) and an increase in opportunistic and/or pro-inflammatory bacteria[3,4]. Such shifts affect microbial consortium metabolism, leading to decreased production of organic acids, including acetate and butyrate, excessive production of putrefactive metabolites, such as ammonia and phenols, and compromised immune function[3,4]. These alterations are driven by the complex interplay between the genetic, environmental, and lifestyle factors associated with aging, ultimately contributing to the loss of systemic homeostasis and functional decline.

Moreover, residents of elderly facilities face an increased risk of infection and colonization by multidrug-resistant organisms in the intestine[5]. Addressing these concerns involves regulating the microbiota surrounding the elderly, including that of individuals who regularly interact with them, and preserving the microbiota of the residents themselves. Haran et al[6] demonstrated that residents in close proximity share similar species and strains. This finding underscores the importance of preserving the gut microbiota and environment of elderly residents and individuals in proximity, such as facility staff, to mitigate the risk of horizontal transmission or infection by health-threatening strains within the facility.

Administration of probiotics is a promising method for sustaining the health of residents in elderly houses by modulating their gut microbiota and the environment. According to the International Scientific Association for Probiotics and Prebiotics, probiotics are live microorganisms that, when administrated in adequate amounts, confer a health benefit on the host[7]. Systematic reviews have shown the potential benefits of probiotics in improving gut health and mitigating age-related issues, such as constipation, common cold incidence/duration, and other infections, although some conflicting results exist owing to strain and study design variations[8,9]. The Lactobacillus casei strain Shirota, currently reclassified as Lacticaseibacillus paracasei Shirota (LcS), is a well-known probiotic that has a longstanding presence in the global market[10]. Clinical trials on the impact of LcS-fermented milk in nursing homes have shown positive outcomes, including increased levels of beneficial bacteria such as bifidobacteria and lactobacilli, and a reduction in potentially harmful bacteria such as Clostridioides difficile. These changes are associated with increased organic acid production, decreased intestinal pH (indicating gut environment improvement), and reduced risk of health complications in elderly residents, such as shorter duration of fever caused by norovirus infections and improved bowel movements[11,12]. Notably, Nagata et al[12] demonstrated that the positive impacts of LcS on the gut microbiota and environment extend not only to elderly residents but also to staff in the facility.

This indicates that LcS-fermented milk is beneficial for maintaining optimal gut conditions for individuals in elderly houses, thereby contributing to their overall health. However, the effects of LcS in the Indonesian population have not been adequately studied. Therefore, the present study aimed to investigate the effects of daily LcS-fermented milk intake on the gut microbiota and environment of individuals in Indonesian elderly houses, comprising residents and staff within a randomized controlled setting.

MATERIALS AND METHODS

Trial design

This was a multicenter, double-blind, placebo-controlled, parallel-group superiority study with balanced randomization (1:1) for the two arms.

Participants

Residents and staff of three elderly houses in Denpasar, Singaraja, and Tabanan, Bali, Indonesia, were studied. Bali is ranked third in Indonesia for its proportion of elderly individuals (13.5%) in 2022, followed by DI Yogyakarta (16.7%) and East Java (13.7%)[2]. Previous studies have reported that residents of elderly houses in Bali exhibit a less favorable gut microbiome than those in Yogyakarta, characterized by a higher abundance of potentially detrimental bacteria, such as Clostridium perfringens and Enterococcus[13]. The present study included individuals who provided informed consent and were confirmed to be physically and mentally healthy by doctors with no issues with regular meals. The exclusion criteria were excessive alcohol intake (> 28 glasses/week for men and > 21 glasses/week for women); narcotic/psychotropic drug use; recent antibiotic/laxative use (within 2 weeks prior to the study); recent consumption of probiotics, prebiotics, or fermented dairy foods (within 2 weeks); irritable bowel syndrome (diagnosed according to the Rome IV criteria); history of gastrointestinal surgery; pregnancy; and allergies to the study product. Additionally, individuals who could not avoid consuming fermented dairy products, probiotics, or prebiotics during the study period were excluded.

Interventions

Participants were randomized to receive either a fermented milk drink containing > 6.5 × 10⁹ colony-forming units LcS/65 mL (Yakult^®) or an acidified milk placebo matched in nutritional values and organoleptic characteristics. Following a 2-week baseline period, the participants consumed one bottle of their assigned product daily at breakfast for 24 weeks (6 months). The bacterial count used in this study reflected the composition of a commercially available product, which was designed to be consumed in one bottle per day. This dosing approach enables the evaluation of the intervention under conditions that closely simulate real-world use, offering practical insights into its applicability and effectiveness in the elderly community. During the study, the fermented milk samples were handled according to the manufacturer’s instructions to maintain the stability of the minimum viable cell count during the study. The products were stored in a refrigerator at 4-10 °C. Temperature was monitored using a thermohygrometer (TFA Dostmann, Germany). The consumption of the assigned product was monitored by collecting empty bottles.

Outcomes

The primary endpoint was the impact of the study product on the gut microbiome, which was assessed through fecal samples collected at the end of the 2-week baseline period and 3 and 6 months post-treatment (all with a ± 1-day window). The samples were subjected to 16S rRNA gene sequencing and reverse transcription quantitative polymerase chain reaction (RT-qPCR) to identify robust microbiome alterations. The secondary endpoint was the evaluation of gut environmental indicators through the measurement of organic and amino acid metabolite levels and fecal pH in the collected samples. In addition, stool frequency was assessed throughout the study.

Fecal sample collection and transportation

The protocol for fecal sample collection differed between the residents and staff. Elderly facility residents collected a portion of their stool onto trail paper (Eiken Chemical, Japan) and transferred it to a sterile 100 mL container using a disposable spoon. Subsequently, the filled container was delivered to the investigator and stationed at a pre-established clinical center within the facility. The investigator then aliquoted the sample into three sampling tubes labeled A, B, and C (cat. no. 80.734.001, Sarstedt, Germany). Tube A was prefilled with 2 mL of RNAlater (Thermo Fisher Scientific, Waltham, MA, United States) and zirconia beads. Immediately after addition of the stool, tube A was vigorously mixed using a vortex mixer. Staff residing outside the facilities received a collection kit packaged in an insulated box (cat. no. KC-3; Karux, Japan) containing two ice packs (cat. No. 160240; Yakult Shoji, Tokyo, Japan). They were instructed to collect their stool directly into three pre-labeled tubes (A, B, and C). Following sample collection, the staff delivered the samples to the investigator at their designated clinical center within 1 hour, maintaining cool conditions during transport. The samples collected in the evening were acceptable for delivery the following morning. Tube A was then delivered via refrigerated shipping to the rental P2 laboratory of Narita Animal Science Laboratory, Ltd. (Chiba, Japan) for subsequent nucleic acid extraction. After extraction, the samples were sent to the Yakult Central Institute (Tokyo, Japan) for microbiome analysis. Tubes B and C were delivered to the Prodia Clinical Laboratory/Prodia Occupational Health Institute, Jakarta, Indonesia, in a frozen state for organic acid and amino acid metabolite analyses.

Gut microbiota analysis

16S rRNA gene sequencing and taxonomic assignment were performed as previously described[14]. Fecal DNA was extracted by bead-beating in phenol[15]. Subsequently, variable regions 1 and 2 (V1 and V2) of the 16S rRNA gene were amplified using the primers 27Fmod2-MiSeq and 338R-MiSeq (Supplementary Table 1). Amplicons were sequenced on a MiSeq System using a paired-end 2 × 250 bp cycle run and the MiSeq Reagent Kit v2 (Illumina, CA, United States). The Quantitative Insights into Microbial Ecology 2 (QIIME2) software package (version 2018.8)[16] was used to analyze the amplicon sequence reads. DADA2 was employed to filter low-quality sequences, denoise, concatenate 16S rRNA reads, and remove potential chimeric sequences[17]. The resulting features were clustered into phylotypes (clustered features) using q2-vsearch for open reference clustering against the Living Tree Database (LPTs132_SSU)[18]. The taxonomic assignment for each phylotype was achieved using the QIIME feature classifier against the SILVA database (Release 138)[19] with a minimum bootstrap threshold of 50%. Alpha diversity (the number of OTUs observed, Chao1, Shannon index, and Faith’s phylogenetic diversity) and β-diversity (Bray-Curtis, Jaccard, Weighted UniFrac, Unweighted UniFrac, and Aitchison distance) were estimated for 6500 randomly selected sequences.

16S and 23S rRNA-targeted RT-qPCR was performed using the Yakult Intestinal Flora-Scan system (YIF-SCAN^®)[20]. Fecal RNA was isolated as previously described[11,12]. A standard curve was generated using the threshold cycle (C_t), the cycle number at which the threshold fluorescence was reached, and the corresponding cell count, which was determined microscopically using 4,6-diamidino-2-phenylindole (Vector Laboratories, Burlingame, CA, United States) staining for the dilution series of standard strains described elsewhere[11,12]. To determine the bacteria present in the samples, three serial dilutions of the extracted RNA samples were used for RT-qPCR. C_t values in the linear range of the assay were applied to the standard curve generated in the same experiment to obtain the corresponding bacterial cell count for each nucleic acid sample, which was then converted to the number of bacteria per sample. The specificity of the RT-qPCR assay using group-, genus-, or species-specific primers was determined as previously described[11,12]. The primers used in this study and their annealing temperatures are listed in Supplementary Table 2.

Organic acid and pH analysis

Organic acids, including acetate, formate, propionate, isobutyrate, butyrate, isovalerate, valerate, isocaproate, caproate, heptanoate, lactate, and succinate, were analyzed following established protocols with minor modifications[21-24]. Briefly, a 200 mg fecal sample was suspended in 1 mL of ultrapure water, homogenized with a vortex, and sonicated for 20 minutes. The homogenate was centrifuged at 10000 × g for 5 minutes, and the resulting clear brownish supernatant was transferred to separate tubes for analysis. For the organic acids except succinate and lactate, 300 μL of ultrapure water containing the internal standard (2,2-dimethyl butyric acid; Merck, Germany) was added to 100 μL of fecal homogenate supernatant. Then, 100 μL of the mixture was combined with 600 μL of a diluent containing isopropanol (Merck) and 1.5 M hydrochloric acid (Merck) and transferred to gas chromatography (GC) vials. For succinate and lactate, 75 μL of 3 M hydrochloric acid in methanol was added to 25 μL of fecal homogenate supernatant. Subsequently, 400 μL of methanol (Merck) was added to the 100 μL derivatized mixture and transferred to GC vials. External standards were prepared to create calibration curves. Samples were injected into a 7890 Gas Chromatography System coupled with a 5977 mass-selective detector equipped with an electron impact source (Agilent Technologies, CA, United States). Organic acids were separated using a Nukol column (film thickness: 30 m × 0.25 mm × 0.25 μm; Supelco, PA, United States). The GC oven temperature was increased from 60 °C to 150 °C at a rate of 10 °C/minute. The mass-selective detector was operated in scan mode, acquiring data within a mass range of 40-150 m/z. The fecal pH was measured by inserting the probe of a pH meter (Eutech Instruments, Singapore) into the fecal sample.

Amino acid metabolite analysis

Indole, p-cresol, phenol, and ammonia levels were quantified by distillation, followed by GC with flame ionization detection (Agilent Technologies, CA, United States) for indole, p-cresol, and phenol, and spectrophotometry for ammonia. Briefly, fecal samples (0.5 g) were homogenized with 0.5 g of phosphate buffer, and Milli-Q water (ultrapure/type 1 water) up to 5 g was added. The homogenate was transferred to a distillation apparatus and distilled at 110 °C for 180 minutes. In measuring indole, p-cresol, and phenol, the distillate was collected in a 4 mL test tube containing 500 μL of methanol. Subsequently, the collected distillate was diluted to 5 mL with Milli-Q water, and 1 mL of the diluted solution was transferred to a 2 mL vial for GC with flame ionization detection. For ammonia measurement, the distillate was collected in a 4 mL test tube containing 500 μL of 12.04 N hydrochloric acid. Using Milli-Q water, the collected distillate was diluted to 5 mL, followed by 20-fold dilution. Subsequently, 5 mL of the 20-fold diluted sample was mixed with 0.2 mL of phenol solution (mixture of liquefied phenol (≥ 89%) and 95% v/v ethyl alcohol, in 1:8.009 ratio, respectively), 0.2 mL of 0.5% w/v sodium nitroprusside solution, and 0.5 mL of oxidizing solution (alkaline citrate solution and sodium hypochlorite mixture, in 4:1 ratio, respectively) in a 12 mL test tube. The test tube was covered and incubated for 1 hour at 22-27 °C (or room temperature). Absorbance was measured at 640 nm using a spectrophotometer (Agilent, China).

Stool frequency

The participants reported their daily stool frequency and compliance with the study protocol throughout the baseline and intervention periods using a self-completed diary.

Sample size

The sample size was determined to replicate the findings of a previous study by Nagata et al[12], which included all elderly residents and staff in an elderly care facility. We hypothesized that administering probiotics to the entire community might improve the gut microbiome, such as, partially by reducing the transmission of pathogenic bacteria. Accordingly, we planned to recruit all eligible residents and staff from participating facilities, resulting in a total of 112 participants. The rationale for this sample size was further supported by a previous intervention study, which reported mean lactobacilli counts of 6.20 ± 0.82 Log₁₀ bacterial cells/g in the treatment group and 5.73 ± 0.66 Log₁₀ bacterial cells/g in the control group following four weeks of commercial probiotic consumption. Using a two-sided test with a 5% significance level and 85% power, the required sample size was calculated to be 45 participants per group. After accounting for a 20% dropout rate, the adjusted sample size was 56 participants per group (112 participants in total). This calculation ensured a balanced sample size sufficient to yield reliable and generalizable results.

Randomization

A statistician from Prodia, the CRO (Jakarta, Indonesia), who was independent of participant recruitment, generated a random allocation sequence using block randomization with a block size of four, stratified by participant type (resident/staff) and study site (Denpasar/Singaraja/Tabanan). PT Yakult Indonesia Persada (Jakarta, Indonesia) concealed the allocation sequence and dispensed blinded investigational products.

Blinding

To maintain participant and investigator blinding, the treatment allocation was concealed using a randomization code. The investigational products were labeled with unique codes (“SBMC” and “SBMD”), one corresponding to the placebo and the other to LcS fermented milk. However, the labels did not disclose which code represented which intervention, ensuring that the participants and site personnel could not identify their assigned treatment based on appearance or taste. Furthermore, fecal tubes were labeled with unique identifiers corresponding to the randomization code to ensure blinded fecal sample analysis.

Statistical analysis

This study employed an intention-to-treat approach, incorporating all participants regardless of loss to follow-up, noncompliance, or withdrawal. Differential abundance analysis on genus-level microbiome data was conducted at each time point using ALDEx2, ANCOM, and ANCOM-BC, with adjusting for participant type (resident/staff) and study site (Denpasar/Singaraja/Tabanan) effects[25-29]. The top and bottom 20 bacteria were identified using the effect size (ALDEx2), centered log ratio mean difference (ANCOM), and log-fold change (ANCOM-BC), and their significance testing information was reported. This methodological approach adheres to the recommendation of employing multiple methods to interpret differential abundance data in microbial communities[30]. Genera were considered differentially abundant if they showed significance in at least one analysis with consistent ordering in all analyses. Significance testing for global changes in community composition (β-diversity) was performed using permutational multivariate analysis of variance based on Bray-Curtis, Jaccard, Weighted UniFrac, Unweighted UniFrac, and Aitchison distance metrics. The analysis was adjusted for participant type and study site using the Adonis function in the vegan package[31]. Between-group differences for continuous outcomes were evaluated using the Student’s t-test after log or square root transformation, if necessary, for data normalization. Alternatively, the nonparametric Mann-Whitney U test was used for non-normalized data. Categorical variables were compared using the Fisher’s exact test. Mixed linear (or logistic) models were constructed based on the framework of Twisk et al[32], adjusting for baseline values, participant type, and study site with the appropriate link function applied. A subgroup analysis was conducted to assess the differences in treatment responses between residents and staff by evaluating the interactions between participant type and treatment. P < 0.05 was considered statistically significant. All analyses were performed using R software (version 4.1.1; The R Foundation for Statistical Computing, Vienna, Austria).

RESULTS

Participant flow and recruitment

Figure 1 shows a participant flow diagram. Eligible participants were recruited from three elderly houses in Bali, Indonesia in September 2018. Overall, 112 participants were randomized using a stratified design according to participant type (resident/staff) and facility location (Denpasar/Singaraja/Tabanan). This resulted in 56 participants (33 residents and 23 staff) allocated to the placebo group and 56 participants (34 residents and 22 staff) allocated to the probiotic group. Following a 2-week baseline period in October 2018, participants received daily investigational products for 6 months (October 2018 to March 2019). Two participants in the placebo group withdrew before receiving their assigned drinks because of death and missing baseline fecal samples. During the intervention period, four participants in the placebo group and seven in the probiotic group were lost to follow-up. Reasons for withdrawal included unavoidable circumstances such as death unrelated to the study products, hospitalization, relocation from the facility, and noncompliance with the protocol (missed stool sample collections or refusal to consume the assigned drinks). An intention-to-treat analysis was performed on participants who received the allocated interventions, including 54 participants in the placebo group and 56 in the probiotic group.

Figure 1 Flow diagram representing the study.

Baseline data

Table 1 shows the demographic and baseline characteristics of the participants. The baseline data were well balanced between the groups, indicating effective randomization.

Table 1 Baseline characteristics of the participants, n (%).

	Placebo (n = 54)	Probiotic (n = 56)
Participant type, resident/staff	31 (57.4)/23 (42.6)	34 (60.7)/22 (39.3)
Age, median (IQR), year	65.0 (47.0-61.7)	65.5 (50.8-61.7)
Height, mean ± SD, cm	153.8 ± 10.6	155.0 ± 9.8
Weight, median (IQR), kg	53.9 (41.6-64.6)	52.4 (41.8-66.6)
BMI, median (IQR), kg/m2	22.6 (19.0-25.7)	22.3 (18.6-26.1)
Sex, male/female	15 (27.8)/39 (72.2)	18 (32.1)/38 (67.9)
Nutrient intake, mean ± SD, day
Energy, mean ± SD, kcal	893.30 ± 154.37	901.39 ± 154.51
Water, mean ± SD, g	696.86 ± 134.93	735.78 ± 154.77
Protein, mean ± SD, g	42.28 ± 10.11	41.60 ± 10.31
Fat, mean ± SD, g	38.50 ± 9.03	39.40 ± 10.44
Carbohydrate, mean ± SD, g	95.23 ± 20.55	96.00 ± 15.83
Dietary fiber, mean ± SD, g	5.25 ± 1.93	5.55 ± 1.86
Alcohol, mean ± SD, g	0.00 ± 0.00	0.00 ± 0.01
Polyunsaturated fatty acid, mean ± SD, g	6.87 ± 2.00	7.37 ± 2.56
Cholesterol, mean ± SD, mg	190.50 ± 85.76	202.28 ± 92.92
Vitamin A, mean ± SD, μg	563.72 ± 344.67	536.09 ± 273.61
Carotene, mean ± SD, mg	0.08 ± 0.11	0.09 ± 0.12
Vitamin E, mean ± SD, mg	2.75 ± 0.83	2.81 ± 0.86
Vitamin B1, mean ± SD, mg	0.45 ± 0.15	0.44 ± 0.14
Vitamin B2, mean ± SD, mg	0.52 ± 0.16	0.54 ± 0.15
Vitamin B6, mean ± SD, mg	0.72 ± 0.20	0.74 ± 0.22
Folic acid, mean ± SD, μg	109.34 ± 38.93	110.76 ± 35.88
Vitamin C, mean ± SD, mg	30.02 ± 24.76	30.48 ± 25.23
Sodium, mean ± SD, mg	392.78 ± 375.76	312.71 ± 287.12
Potassium, mean ± SD, mg	1030.62 ± 998.85	919.93 ± 267.01
Calcium, mean ± SD, mg	211.76 ± 96.06	221.41 ± 90.79
Magnesium, mean ± SD, mg	167.40 ± 107.23	159.73 ± 36.73
Phosphorus, mean ± SD, mg	483.79 ± 132.22	469.54 ± 106.26
Iron, mean ± SD, mg	6.02 ± 2.99	5.74 ± 1.50
Zinc, mean ± SD, mg	4.22 ± 0.95	4.35 ± 0.99

IQR: Interquartile range; BMI: Body mass index.

Effect of probiotic drink on the gut microbiota

The intervention showed no significant effect on α-diversity metrics throughout the study (Supplementary Table 3). Conversely, several β-diversity measures, including Aitchison distance and Jaccard dissimilarity index, revealed significant alterations in the overall microbial community following 3 months of probiotic drinking (P = 0.045 and 0.035, respectively) (Supplementary Table 4). However, these effects were no longer detected at 6-month period.

Differential abundance analysis using three methods (i.e., ALDEx2, ANCOM, and ANCOM-BC) revealed consistent shifts in the microbial composition at the genus level (Figure 2). At 3-month period, all methods identified a significant increase in Bifidobacterium in the probiotic group compared to that in the placebo group, with the largest effect size (Figure 2A). Additionally, the methods consistently indicated that Succinivibrio was the second-most induced genus and that the Rikenellaceae RC9 gut group was the most suppressed, both reaching significance in ANCOM-BC. At 6-month period, these changes in Succinivibrio and Rikenellaceae RC9 gut group persisted (Figure 2B).

Figure 2 The top and bottom 20 differentially abundant bacterial genera at three months and six months, compared to the placebo group, identified by three methods (ALDEx2, ANCOM, and ANCOM-BC) with their respective metrics. A: Three months; B: Six months. The number next to the ANCOM bar represents the W value.

The transient induction of Bifidobacterium at 3 months in the probiotic group compared to the placebo group was further corroborated by RT-qPCR using the YIF-SCAN^® system (8.5 ± 1.1 vs 8.0 ± 1.1, log₁₀ bacterial cells/g, P = 0.044) (Table 2)[20]. Additionally, the analysis revealed an exponential increase in the abundance of Lacticaseibacillus in the probiotic group at three and six months, reaching 10⁶ bacterial cells/g. This is possibly due to the presence of “live” LcS derived from the probiotic intervention, with intact RNA molecules detectable by RT-qPCR. Notably, studies in several countries have shown that LcS in Yakult^® can remain viable at > 10⁶ bacterial cells/g in human stool[33-36]. Another study showed that 12 weeks of LcS consumption in the elderly increased the abundance of the genus Lacticaseibacillus in the probiotic group compared to that at baseline and in the placebo group[37]. The study findings indicate that LcS from the probiotic drink significantly impacts the gut microbiota after three months, as evidenced by increases in Bifidobacterium and Succinivibrio and a decrease in the Rikenellaceae RC9 gut group, leading to significant changes in several β-diversity metrics. Although the increase in Bifidobacterium appeared transient, alterations in the other two genera persisted after six months.

Table 2 Changes in microbiota measured by reverse transcription quantitative polymerase chain reaction.

	Placebo						Probiotic
	Baseline		3 months		6 months		Baseline		3 months		6 months
Total bacteria	9.3 ± 1.5	98%	9.9 ± 0.8	98%	9.8 ± 1.4	98%	9.8 ± 0.9	100%	9.9 ± 1.0	98%	9.8 ± 1.0	100%
Bacillota
Clostridium coccoides group	8.8 ± 1.3	89%	9.2 ± 0.9	96%	9.3 ± 0.9	94%	9.0 ± 1.1	100%	9.4 ± 0.7	94%	9.1 ± 1.0	98%
Clostridium leptum subgroup	8.6 ± 1.4	96%	9.2 ± 0.9	94%	9.4 ± 0.7	94%	9.1 ± 1.0	100%	9.1 ± 1.0	98%	9.2 ± 1.0	100%
Clostridium perfringens	6.3 ± 1.5	91%	6.0 ± 1.3	88%	5.5 ± 1.2	92%	6.4 ± 1.4	98%	5.9 ± 1.2	92%	5.5 ± 1.1	94%
Clostridioides difficile	6.0 ± 1.7	6%	6.9 ± 0.9	6%	4.9 ± 1.9	6%	3.9 ± 0.4	5%	< 2.3	0%	< 2.3	0%
Total lactobacilli1	6.4 ± 1.7	93%	7.0 ± 1.3	96%	7.1 ± 1.6	94%	6.7 ± 1.6	98%	7.1 ± 1.3	98%	7.2 ± 1.2	98%
Lactobacillus2	5.4 ± 1.5	76%	5.6 ± 1.4	90%	5.8 ± 1.5	88%	5.7 ± 1.6	88%	5.8 ± 1.5	81%	5.5 ± 1.5	92%
Levilactobacillus brevis2	3.3 ± 0.6	24%	3.5 ± 0.6	37%	3.5 ± 1.2	38%	3.4 ± 0.8	38%	4.0 ± 1.0	23%	3.3 ± 0.4	29%
Lacticaseibacillus2	4.6 ± 1.4	19%	5.2 ± 1.4	31%	4.2 ± 1.0	36%	4.9 ± 1.3	32%	6.1 ± 1.2a	94%b	6.3 ± 0.8b	88%b
Limosilactobacillus fermentum2	5.6 ± 1.0	56%	5.3 ± 0.7	71%	4.9 ± 0.9	56%	5.4 ± 0.7	71%	5.5 ± 0.9	71%	4.9 ± 0.6	65%
Fructilactobacillus fructivorans2	3.1	2%	3.1 ± 0.4	6%	2.7	2%	< 2.3	0%	3.0 ± 0.2	6%	2.8 ± 0.4	6%
Lactiplantibacillus2	4.1 ± 1.1	85%	4.4 ± 1.2	92%	4.6 ± 1.0	94%	4.2 ± 1.0	91%	4.4 ± 1.2	83%	4.6 ± 1.1	98%
Limosilactobacillus except L. fermentum2	4.7 ± 1.3	80%	5.1 ± 1.2	90%	4.7 ± 1.3	94%	5.2 ± 1.3	82%	5.3 ± 1.2	87%	4.9 ± 1.1	90%
Liquorilactobacillus and Ligilactobacillus2	6.8 ± 2.1	35%	7.3 ± 1.6	48%	7.7 ± 1.5	48%	7.1 ± 2.0	36%	6.9 ± 2.0	58%	7.3 ± 1.7	53%
Latilactobacillus2	4.7 ± 0.4	4%	4.6 ± 0.4	15%	4.0 ± 0.8	16%	4.3 ± 0.8	9%	4.0 ± 0.9	12%	3.9 ± 0.8	16%
Enterococcus	6.8 ± 1.5	78%	7.5 ± 0.8	85%	7.5 ± 1.0	86%	6.9 ± 1.1	79%	7.1 ± 1.1c	81%	7.2 ± 1.2	90%
Streptococcus	5.1 ± 1.1	83%	5.0 ± 0.8	96%	5.3 ± 0.8	92%	5.1 ± 0.9	96%	5.0 ± 0.8	94%	5.3 ± 0.9	98%
Staphylococcus	7.4 ± 1.9	70%	8.0 ± 1.0	77%	8.2 ± 0.9	72%	8.1 ± 1.0	80%	8.1 ± 1.2	71%	8.0 ± 1.2	84%
Bacteroidota
Bacteroides fragilis group	7.6 ± 0.9	80%	8.0 ± 0.9	92%	8.2 ± 0.8	94%	7.7 ± 1.0	93%	8.0 ± 0.9	94%	8.2 ± 1.0	98%
Prevotella	8.3 ± 1.7	80%	8.9 ± 1.3	88%	8.9 ± 1.4	74%	8.8 ± 1.2	80%	9.1 ± 1.5	83%	9.1 ± 1.1	78%
Actinomycetota
Atopobium cluster	8.7 ± 1.2	94%	8.9 ± 0.8	96%	8.8 ± 0.9	96%	9.1 ± 0.8	100%	8.9 ± 0.8	98%	8.8 ± 0.8	100%
Bifidobacterium	8.1 ± 1.1	81%	8.0 ± 1.1	94%	8.2 ± 1.1	90%	8.3 ± 1.1	98%	8.5 ± 1.1a	94%	8.1 ± 0.9	98%
Pseudomonadota
Enterobacteriaceae	7.7 ± 1.4	89%	8.2 ± 1.0	96%	8.2 ± 0.9	94%	7.7 ± 1.2	100%	8.1 ± 1.0	96%	8.0 ± 1.0	98%
Pseudomonas	4.3 ± 1.0	30%	3.9 ± 0.9	50%	4.1 ± 0.7	76%	3.8 ± 0.7	27%	3.9 ± 0.9	42%	4.0 ± 0.5	78%

^aP < 0.05 vs placebo.

^bP < 0.01 vs placebo.

^cP < 0.10 vs placebo.

¹Total lactobacilli refers to the sum of bacteria denoted by the bacterial groups.

²The bacterial groups (e.g., Lactobacillus, Levilactobacillus brevis, etc.).

The results are expressed as mean ± SD (log₁₀ cells/g of feces) and detection rate (%).

Effect of probiotic drink on the gut environment

Stool butyrate concentration significantly increased after six months in the probiotic group compared to that in the placebo group (5.04 ± 3.11 vs 3.95 ± 2.89 μmol/g, P = 0.048) (Table 3). The probiotic drinks did not significantly affect the levels of other organic acid species, pH, and amino acid metabolites. Nonetheless, mixed linear effects modeling adjusted for baseline values, participant type, and study site revealed potential trends (Supplementary Table 5). The model demonstrated a transient increase in acetate and total organic acid levels three months after the administration of the probiotic drink (P = 0.061 and P = 0.069, respectively) and a lower stool pH (P = 0.053). Furthermore, ammonia levels tended to decrease and indole levels tended to increase at six months (P = 0.068 and P = 0.071, respectively). However, further investigations are required to confirm these trends.

Table 3 Changes in stool organic acids and amino acid metabolites.

	Placebo			Probiotic
	Baseline	3 months	6 months	Baseline	3 months	6 months
Organic acids, μmol/g
Acetate, mean ± SD	13.75 ± 4.82	15.88 ± 8.01	14.63 ± 5.54	14.24 ± 5.49	17.97 ± 8.21	15.81 ± 4.85
Formate, mean ± SD	1.91 ± 0.40	1.71 ± 0.30	2.44 ± 0.24	1.90 ± 0.38	1.72 ± 0.17	2.43 ± 0.21
Propionate, mean ± SD	4.44 ± 2.43	5.56 ± 3.51	5.43 ± 3.81	4.68 ± 2.66	6.46 ± 4.09	5.82 ± 3.27
Isobutyrate, mean ± SD	0.41 ± 0.23	0.50 ± 0.39	0.48 ± 0.28	0.48 ± 0.32	0.51 ± 0.42	0.52 ± 0.27
Butyrate, mean ± SD	3.56 ± 2.23	4.32 ± 3.11	3.95 ± 2.89	4.29 ± 3.03	5.21 ± 3.05b	5.04 ± 3.11a
Isovalerate, mean ± SD	0.36 ± 0.24	0.45 ± 0.35	0.42 ± 0.27	0.44 ± 0.34	0.48 ± 0.39	0.43 ± 0.24
Valerate, mean ± SD	0.65 ± 0.41	0.63 ± 0.47	0.57 ± 0.35	0.73 ± 0.44	0.68 ± 0.39	0.67 ± 0.37
Isocaproate, median (IQR)	0.01 (0.01-0.03)	0.02 (0.01-0.03)	0.01 (0.00-0.01)	0.01 (0.01-0.02)	0.02 (0.01-0.04)	0.01 (0.01-0.01)
Caproate, median (IQR)	0.07 (0.04-0.14)	0.06 (0.01-0.23)	0.05 (0.03-0.12)	0.05 (0.03-0.16)	0.06 (0.02-0.11)	0.05 (0.03-0.17)
Heptanoate, median (IQR)	0.01 (0.01-0.03)	0.01 (0.00-0.10)	0.01 (0.00-0.01)	0.02 (0.01-0.03)	0.00 (0.00-0.06)	0.01 (0.00-0.02)
Lactate, median (IQR)	0.03 (0.02-0.05)	0.03 (0.02-0.04)	0.03 (0.02-0.04)	0.03 (0.01-0.04)	0.02 (0.02-0.03)	0.02 (0.02-0.04)
Succinate, median (IQR)	0.01 (0.01-0.02)	0.02 (0.01-0.04)	0.01 (0.01-0.03)	0.01 (0.01-0.03)	0.03 (0.01-0.06)	0.02 (0.01-0.03)
Total organic acids, mean ± SD	25.31 ± 9.08	29.38 ± 13.83	28.11 ± 11.81	27.00 ± 9.54	33.33 ± 13.86	30.95 ± 10.11
pH
pH, median (IQR)	6.12 (5.86-6.30)	5.99 (5.80-6.35)	6.06 (5.81-6.39)	6.08 (5.75-6.41)	5.89 (5.60-6.38)	5.93 (5.75-6.18)b
Amino acid metabolites, μg/g
Ammonia, median (IQR)	383.7 (245.4-479.5)	467.6 (361.5-568.1)	2015.4 (651.5-3263.7)	358.9 (260.8-506.0)	470.2 (374.6-585.3)	1401.8 (335.8-3116.6)
Indole, median (IQR)	15.92 (9.19-30.40)	13.46 (8.14-22.35)	3.82 (2.78-10.28)	11.97 (7.29-23.70)	14.70 (11.50-23.87)	5.94 (2.95-12.01)
Phenol, median (IQR)	5.24 (4.35-8.67)	5.74 (1.91-14.24)	3.71 (1.96-5.46)	7.02 (2.87-15.46)	3.65 (1.66-10.27)	2.31 (1.40-4.07)
p-cresol, median (IQR)	16.36 (7.42-26.30)	13.03 (8.54-29.98)	6.94 (4.33-11.47)	12.45 (9.46-26.09)	12.44 (7.94-25.91)	8.09 (5.64-12.06)

^aP < 0.05 vs placebo.

^bP < 0.10 vs placebo.

IQR: Interquartile range.

Effect of probiotic drink on stool frequency

Probiotic consumption did not significantly influence stool frequency during the intervention period, despite exhibiting higher frequencies in the probiotic group than in the placebo group for most months. Notably, all participants reported near-optimal bowel movement frequency before the start of the intervention (Table 4).

Table 4 Changes in stool frequency, mean ± SD.

Months	Stool frequency (times/week)		Stool frequency (days/week)
	Placebo	Probiotic	Placebo	Probiotic
Baseline	6.24 ± 2.61	6.20 ± 2.14	5.59 ± 1.62	5.69 ± 1.63
1	6.05 ± 1.72	6.07 ± 1.43	5.85 ± 1.49	5.89 ± 1.33
2	6.19 ± 1.62	6.27 ± 1.37	5.98 ± 1.38	6.08 ± 1.22
3	6.25 ± 1.47	6.21 ± 1.32	6.02 ± 1.29	6.08 ± 1.26
4	6.18 ± 1.49	6.50 ± 1.14	6.03 ± 1.40	6.27 ± 1.00
5	6.32 ± 1.40	6.56 ± 1.05	6.10 ± 1.27	6.32 ± 0.87
6	6.50 ± 1.25	6.78 ± 1.02	6.29 ± 1.15	6.51 ± 0.77

Subgroup analysis

Subgroup analyses were conducted to evaluate potential variations in probiotic efficacy according to the participant type (residents vs staff). Therefore, the interaction effects between treatment group and participant type were assessed for all outcome measures, followed by least squares (LS) mean difference calculations within each participant type. Significant interactions were observed in the RT-qPCR outcomes (Table 5). Bifidobacterium levels exhibited significant interaction effects at three (P = 0.031) and six months (P = 0.044). Notably, the 95% confidence interval of the LS mean difference at 3 months was positive for residents and overlapped with zero for staff, indicating a greater effect of the probiotic drink on increasing Bifidobacterium levels in residents than in staff. Additionally, a significant interaction was identified for Enterococcus levels at six months (P = 0.003). The mean differences in LS at three and six months indicated a notable reduction in Enterococcus abundance in the staff population following probiotic consumption.

Table 5 Subgroup analyses on reverse transcription quantitative polymerase chain reaction outcomes.

	Participant types	3 months		6 months
		LS mean difference (95%CI)	P value1	LS mean difference (95%CI)	P value1
Total bacteria	Resident	0.12 (-0.33 to 0.57)	0.488	-0.05 (-0.68 to 0.59)	0.752
	Staff	-0.12 (-0.65 to 0.41)		0.11 (-0.62 to 0.83)
Bacillota
Clostridium coccoides group	Resident	0.30 (-0.14 to 0.74)	0.279	0.08 (-0.42 to 0.58)	0.068
	Staff	-0.07 (-0.60 to 0.45)		-0.62 (-1.20 to -0.05)
Clostridium leptum subgroup	Resident	0.13 (-0.35 to 0.61)	0.198	0.06 (-0.42 to 0.54)	0.159
	Staff	-0.36 (-0.93 to 0.22)		-0.47 (-1.03 to 0.09)
Clostridium perfringens	Resident	0.07 (-0.62 to 0.75)	0.451	0.22 (-0.41 to 0.86)	0.491
	Staff	-0.34 (-1.14 to 0.46)		-0.12 (-0.87 to 0.64)
Total lactobacilli	Resident	0.37 (-0.31 to 1.06)	0.273	0.51 (-0.22 to 1.25)	0.067
	Staff	-0.21 (-1.01 to 0.59)		-0.55 (-1.43 to 0.32)
Enterococcus	Resident	-0.13 (-0.70 to 0.43)	0.144	0.32 (-0.26 to 0.90)	0.003b
	Staff	-0.77 (-1.42 to -0.12)		-1.04 (-1.70 to -0.37)
Streptococcus	Resident	-0.06 (-0.47 to 0.36)	0.964	0.08 (-0.37 to 0.53)	0.698
	Staff	-0.07 (-0.56 to 0.42)		-0.06 (-0.57 to 0.46)
Staphylococcus	Resident	0.17 (-0.52 to 0.86)	0.966	-0.01 (-0.65 to 0.62)	0.340
	Staff	0.19 (-0.57 to 0.94)		-0.48 (-1.21 to 0.26)
Bacteroidota
Bacteroides fragilis group	Resident	0.04 (-0.42 to 0.50)	0.639	0.10 (-0.35 to 0.55)	0.167
	Staff	-0.13 (-0.68 to 0.43)		-0.38 (-0.90 to 0.14)
Prevotella	Resident	0.44 (-0.29 to 1.16)	0.206	0.15 (-0.65 to 0.96)	0.911
	Staff	-0.28 (-1.13 to 0.57)		0.22 (-0.61 to 1.04)
Actinomycetota
Atopobium cluster	Resident	0.00 (-0.41 to 0.40)	0.904	-0.05 (-0.49 to 0.38)	0.969
	Staff	-0.04 (-0.53 to 0.44)		-0.07 (-0.57 to 0.44)
Bifidobacterium	Resident	0.84 (0.26 to 1.42)	0.031a	0.25 (-0.26 to 0.75)	0.044a
	Staff	-0.14 (-0.81 to 0.53)		-0.54 (-1.11 to 0.04)
Pseudomonadota
Enterobacteriaceae	Resident	-0.03 (-0.54 to 0.49)	0.390	0.10 (-0.4 to 0.61)	0.102
	Staff	-0.37 (-0.97 to 0.23)		-0.54 (-1.12 to 0.05)
Pseudomonas	Resident	-0.19 (-0.74 to 0.36)	0.721	-0.21 (-0.56 to 0.15)	0.865
	Staff	-0.39 (-1.35 to 0.58)		-0.16 (-0.62 to 0.31)

¹P values for the interaction between treatment and participant type.

^aP < 0.05.

^bP < 0.01.

LS: Least squares; CI: Confidence interval.

Adverse events

Table 6 summarizes the Unified Medical Language System/Medical Dictionary for Regulatory Activities-coded adverse events, focusing on gastrointestinal disorders. No significant differences were found in overall adverse events between the groups (Table 6). Bloating had the highest odds ratio compared to the placebo group (odds ratio = 5.00; P = 0.496), whereas respiratory, thoracic, and mediastinal disorders had the lowest odds ratio (odds ratio = 0.23; P = 0.202).

Table 6 Number and percentage of participants with adverse events, n (%).

	Placebo	Probiotic	Odds ratio	P value
Any adverse event	22 (40.7)	21 (38.9)	0.87	0.845
Gastrointestinal disorders	8 (14.8)	6 (11.1)	0.69	0.577
Diarrhea	2 (3.7)	2 (3.7)	0.96	1.000
Stomatitis	1 (1.9)	0 (0)	0.32	0.491
Stomachache	2 (3.7)	1 (1.9)	0.47	0.615
Toothache	1 (1.9)	1 (1.9)	0.96	1.000
Pain of epigastrium	1 (1.9)	2 (3.7)	1.96	0.514
Bloated	0 (0)	2 (3.7)	5.00	0.496
Gastric pain	0 (0)	1 (1.9)	2.95	1.000
Dyspepsia syndrome	1 (1.9)	0 (0)	0.32	0.491
Lower abdominal pain	1 (1.9)	0 (0)	0.32	0.491
Nausea	0 (0)	1 (1.9)	2.95	1.000
Nervous system disorders	5 (9.3)	8 (14.8)	1.63	0.557
Musculoskeletal and connective tissue disorders	5 (9.3)	6 (11.1)	1.18	1.000
General disorders and administration site conditions	4 (7.4)	5 (9.3)	1.23	0.523
Respiratory, thoracic, and mediastinal disorders	4 (7.4)	1 (1.9)	0.23	0.202
Infections and infestations	2 (3.7)	2 (3.7)	0.96	1.000
Skin and subcutaneous tissue disorders	2 (3.7)	1 (1.9)	0.47	0.615
Eye disorders	1 (1.9)	1 (1.9)	0.96	1.000
Metabolism and nutrition disorders	0 (0)	1 (1.9)	2.95	1.000
Vascular disorders	0 (0)	1 (1.9)	2.95	1.000
Psychiatric disorders	1 (1.9)	0 (0)	0.32	0.491

DISCUSSION

This study investigated the impact of daily consumption of fermented milk containing > 6.5 × 10⁹ colony-forming units LcS on the gut microbiota and metabolite profile of individuals in elderly houses, including residents and staff, over a 6-month period. After three months, significant gut microbiota changes were observed, characterized by an increase in Bifidobacterium and Succinivibrio and a decrease in the Rikenellaceae RC9 gut group. These alterations coincided with a shift in β-diversity metrics, indicating a change in overall microbial community. Although limited changes were detected in the metabolite profile at this time point, increased total organic acids and acetate and decreased fecal pH were observed. This was followed by a remarkable induction of butyrate production at six months.

The metabolite profile alterations were consistent with changes in the microbial community. Succinivibrio has been reported to exhibit a strong correlation with butyrate production[38,39]. It is a plant polysaccharide-fermenting bacterium that produces succinate and acetate, typically enriched in the Prevotella-type gut microbial community, which is an enterotype prevalent in the Indonesian population[13,40,41]. The observed increase in this genus in our relatively older study group (median age: 65 years) demonstrated that LcS consumption may have shifted the putrefactive-prone gut microbial metabolism toward a more saccharolytic one, potentially leading to higher organic acid production and lower pH. Additionally, Bifidobacterium, which produces acetate from mono- or polysaccharides, may have contributed to the increase in organic acids and lower fecal pH at three months and subsequent butyrate induction at six months. Intriguingly, the acetate produced by Bifidobacterium can be converted to butyrate by other colonic bacteria through cross-feeding interactions[42-45]. Thus, the early modifications in bacterial communities and metabolism at three months may have gradually created an environment favoring cross-feeding interactions that ultimately led to increased butyrate production at six months. Future metatranscriptomic analyses could further elucidate these hypotheses by providing deeper insights into microbial gene expression patterns.

These observed shifts in the gut microbiome have been associated with positive effects on gut and systemic health. Bifidobacterium have been linked to improved outcomes due to their immunomodulatory properties and acetate production[46-49]. Acetate fuels the colonocytes, maintains a robust gut barrier, and prevents pathogen infiltration[50]. Similarly, butyrate is a critical energy source for intestinal cells, reduces inflammation, supports immune function, preserves gut barrier integrity, and may lower the risk of colorectal cancer[50]. Furthermore, a recent study utilizing Mendelian randomization identified Rikenellaceae RC9 gut group as a detrimental bacterium that contributes to ischemic stroke[51]. This bacterium has also been implicated in Parkinson’s disease through its role in modulating the gut branched-chain and aromatic amino acid ratios, potentially influencing immune responses[52]. This collective evidence indicates that LcS has beneficial effects on the gut microbiota and environment, leading to improved human health.

Our findings corroborate those of previous studies. Notably, a randomized, double-blind, placebo-controlled trial in Japan observed similar increases in Bifidobacterium and organic acids following six months of LcS-fermented milk consumption by elderly facility residents and staff[12]. Moreover, this study linked Bifidobacterium changes to elevated stool acetate levels and decreased stool pH, mirroring our results. However, the current study noted these alterations only at three months, whereas the previous study sustained them for six months. These discrepancies in responsiveness may stem from differences in administered bacterial counts or participant characteristics. Specifically, the optimal stool frequency observed in our participants indicated a potentially more favorable baseline gut environment than that of participants in the previous study. This favorable environment may explain the diminished overall responsiveness and selective Bifidobacterium increase observed in the older resident subgroup.

A significant finding in this study was the decrease in Enterococcus in the staff subgroup after LcS consumption. Enterococci pose a growing concern in long-term care facilities owing to their propensity to cause antibiotic-resistant hospital infections[53,54]. These bacteria acquire and disseminate resistance by the horizontal transfer of mobile genetic elements, mediating vancomycin resistance transfer from enterococci to methicillin-resistant Staphylococcus aureus[54]. Notably, antibiotic-resistant Enterococcus can spread directly through patient contact or indirectly through transient carriage on the hands of healthcare personnel, contaminated surfaces, and equipment. The observed reduction in Enterococcus raises the possibility of using LcS as a preventive measure to mitigate the horizontal transmission of antibiotic-resistant Enterococcus strains in elderly care facilities. Further studies that directly investigate the impact on antibiotic resistance genes are warranted to confirm this potential benefit.

This study has several limitations. First, bacterial counts from RT-qPCR analysis were unexpectedly lower than those reported in a previous preliminary study[13]. For example, the prevalence of the Clostridium coccoides group decreased from 100% in a preliminary study conducted in Yogyakarta and Bali, to 89% in the current study. Similarly, the mean total bacterial count decreased from 10¹⁰ cells/g to 10⁹ cells/g. This discrepancy is likely due to RNA degradation at the various stages of fecal sample handling, including collection, storage, transport, and extraction. Because this degradation occurred randomly across samples, it may not have significantly impacted the statistical comparisons between the study groups, although the absolute bacterial counts could be biased. Notably, DNA degradation was less of a concern, as indicated by the consistent bacterial counts obtained by qPCR (data not shown). The agreement between 16S rRNA gene sequencing and RT-qPCR results further supports the robustness of our findings. Additionally, the ammonia measurements in some samples may have been biased because of the bumping reaction during the heating step. Another limitation of this study was that it could not establish a cause-and-effect relationship between the altered gut microbiota, observed gut environment changes, and the health outcomes of the participants. However, a previous study conducted in an elderly facility in Japan found that LcS administration reduced fever duration during a mass norovirus gastroenteritis outbreak, possibly due to an improved gut environment and immune-enhancing effects[11]. This specific effect could not be evaluated in our study because of the small number of participants experiencing similar health issues; however, the observed lower odds ratio for respiratory, thoracic, and mediastinal disorders shows a possible link between LcS-mediated improvements in the gut environment and immune modulation. Future trials in Indonesian populations focusing on clinically relevant outcomes may further elucidate this potential benefit.

CONCLUSION

Daily consumption of LcS-fermented milk significantly increased Bifidobacterium and Succinivibrio levels and decreased the Rikenellaceae RC9 gut group in the guts of individuals in Indonesian elderly houses. These alterations were accompanied by favorable modifications in stool metabolite markers, particularly increased butyrate levels, indicating an improved gut environment. These findings reveal that regular consumption of LcS-fermented milk could benefit the gut microbiota and overall gut health of the elderly population. Further trials are warranted to assess their effects on clinically relevant health outcomes.

ACKNOWLEDGEMENTS

We would like to thank Dina Aulia Nurfiana and Fathyah Hanum P for their assistance in analyzing food record data, Devin Varian Wiryohanjoyo and Bedri Sekar Nurmadhani for their helpful assistance during participant screening, and Monika Dwi Adkhayati and Suharman for their helpful assistance during the sample collection period of this study.