Introduction

Population aging is a global issue, and the speed of population aging is much faster than in the past. According to the United Nations, the number of people aged over 60 is estimated at 1.2 billion by 2025 [1]. In China, the population over 60 years old is estimated at 300 million by 2025 [2]. This demographic shift brings major challenges to the health and social systems of all countries. Frailty is an aging-associated syndrome characterized by significant vulnerability to stressors due to physiological decline across multiple organ systems, elevating the risk of adverse outcomes, such as falls, disability, hospitalization and even death [3]. Frailty is associated with aging, malnutrition, sarcopenia, concurrent multiple diseases, sleep disorders, anxiety and depression [4,5,6].

Gut microbiota is defined as the community of microbes (including bacteria, archaea, and eukarya) inhabiting the GI tract which has a complicated and mutually beneficial relationship with the host [7]. The number of gut microorganisms is estimated to over 1014 [8]. Gut microbiota plays an important role in the metabolism and immune function of the host. Human gut microbiota composition is affected by multiple factors, such as age, race, disease, and medication intake [9,10,11]. Intestinal microorganisms can hydrolyze complex plant polysaccharides to promote the host’s metabolic capacity and produce short-chain fatty acids, which can improve the energy metabolism of colon cells [12]. Some short-chain fatty acids even have anti-inflammatory effects [13]. Human gut microbiota composition is relatively stable during adulthood but undergoes the age-related reduction in biodiversity [9, 14,15,16]. Changes in the composition of the gut microbiota have been reported to be related with several clinical conditions, such as obesity, diabetes, nonalcoholic fatty liver disease, allergic diseases, diabetes, gastrointestinal diseases, autoimmune diseases, and even cancer [17, 18]. Verdi et al. have shown that frailty is associated with gut microbiota in the TwinsUK British twin cohort aging over 40 years [19]. Theou et al. have reported that a prebiotic intervention can reduce frailty levels in non-demented nursing home elder over 65 years, suggesting that gut microbiota has an effect on the frailty levels in the elderly [20]. Although the change in the composition of gut microbiota has been reported in elderly [21] and in frail individuals [22]; however, studies on gut microbiota in frail elderly are still limited [23, 24], especially in late elderly (aged ≥ 75 years). The purpose of this study was to characterize the gut microbiota in late elderly patients with frailty.

Methods

Participants and grouping

From September 2017 to February 2018, 27 elderly patients hospitalized in the Department of Geriatrics, First Affiliated Hospital of Sun Yat-sen University were enrolled. Admission reasons, dietary regimen during hospital stay, and pharmacologic treatments of the enrolled patients have been summarized in Table 1. Frail status of the patient was evaluated by Rockwood Clinical Frailty Scale [25]. The exclusion criteria were as follows: acute infectious diseases or end-stage disease; patients receiving antibiotics, prebiotics treatment or other drugs that regulate intestinal flora in the previous month; receiving gastrointestinal surgery within 6 months. This study was conducted in accordance with the Declaration of Helsinki and approved by the institutional review board of our hospital. Written informed consent was obtained from the patient.

Table 1 Admission reasons, dietary regimen during hospital stay, and pharmacologic treatments of the patients
Full size table

Frailty assessment

Frail status of the patient was evaluated by modified Rockwood frailty questionnaire [26], which consisted of 50 health defects items, mainly including: disease history: hypertension, diabetes, stroke, depression, degenerative diseases, family history of cognitive impairment, etc.; self-care ability: ability to dress, bath, toilet independently, etc.; mental health: emotions problems, anxiety or depression, loneliness, etc.; auxiliary examinations: albumin, BMI, etc. Each item represented a health defect issue. The patient having the defect of the item was recorded as “1”, otherwise it was recorded as “0”. The frailty index was calculated as follows: FI = cumulative health defects/total defects (50). Patients with frailty index < 0.25 were grouped into the frailty group (n = 15), while those with frailty index ≥ 0.25 were grouped into the control group (n = 12) [26].

16S rRNA sequencing

To analyze the gut microbiota diversity, fecal samples of the participants were collected for 16S rRNA sequencing. Fresh fecal samples were collected within 3 days after enrollment. After collection, fresh stool samples were stored at − 80 °C refrigerator or liquid nitrogen tanks within 4 h. The genomic bacterial DNA was extracted from fecal samples using the freeze–thaw method. The 16S rDNA was PCR-amplified and then sequenced on the MiSeq system (Illumina, USA). The primer sequence used in the PCR were F: 5′GTGCCAGCMGCCGCGGTAA3′ and R:5′ GGACTACHVGGGTWTCTAAT3′.

Operational taxonomic unit (OTU) clustering

The sequence reads with a similarity greater than 97% were identified and clustered into an OTU using the Usearch software [27]. The Ribosomal Database Project (RDP) classifier Bayesian algorithm was applied to classify the OTUs clusters, and the community composition was analyzed in each taxonomic rank: domain, kingdom, phylum, class, order, family, genus, and species.

Rank-abundance curve

Within a single sample, the number of sequences included in every single OTU was counted, and the OTUs were sorted by the order of richness. Then the percentage of the sequences number in each OTU was plotted on the rank-abundance curve, which can be used to explain the species abundance and uniformity.

Rarefaction curve

Rarefaction curves were created by repeatedly randomly re-sampling the whole population multiple times and then plotting the average number of species found in each sample. It can be used to compare the species richness between samples, and it can also be used to indicate whether the sampling size of the sample is reasonable.

Intra-sample diversity analysis (alpha-diversity)

The single sample diversity analysis (alpha diversity) can reflect the richness and diversity of microbial communities, including a series of statistical analysis indices to estimate the species abundance and diversity of communities. The community richness indexes included Chao1, Ace, and SOBs, while the community diversity indexes were Shannon and Simpson. Sequencing depth was evaluated by coverage.

Statistical analysis

Continuous variables were presented as mean ± standard deviation (SD) while categorical data were expressed as number and percentage (%). Student’s independent t test was used to compare the differences between groups. If normality of continuous variables was not assumed, non-parametric analysis Mann–Whitney U test would be used instead. Chi square test and Fisher’s exact test (if any expected value lower than 5 was observed) were used for categorical data. Principal co-ordinates analysis (PCoA) and non-metric multidimensional scaling (NMDS) were used to analyze the beta diversity. Hierarchical and K means clustering methods were used to examine the grouping results from diversity index. A p value lower than 0.05 would be recognized as significant. All analyses were performed using IBM SPSS Version 20 (SPSS Statistics V20, IBM Corporation, Somers, New York). Bioinformatics results were analyzed using R software (version 3.5.2).

Results

Patient’s clinical characteristics and laboratory index

A total of 27 patients (17 males and 10 females, mean age 81.63 ± 7.90) were enrolled and divided into frailty group (frailty index < 0.25, n = 15) and control group (frailty index ≥ 0.25, n = 12). Patient’s demographic and clinical characteristics were summarized in Table 2. Frailty index was significantly higher in the frailty group than in the control group (0.35 ± 0.09 vs. 0.14 ± 0.07; p < 0.001). However, there was no significant difference in other outcomes, including all demographic and clinical characteristics, and all blood biochemical indexes (all p > 0.05, Table 2), These results indicated the comparability between the two groups.

Table 2 Patient’s demographic and clinical characteristics
Full size table

Operational taxonomic unit (OTU) clustering and community richness

To characterize gut microbiota, fecal samples were collected for 16S rRNA-amplicon sequencing. The 16S rRNA sequencing reads were clustered using the Usearch software with a 97% sequence similarity threshold. A total of 2501 OTUs were detected (range 187–794). Figure 1 showed the rank-abundance curve of each sample. As the number of OTUs increased, the relative abundance decreased. Most samples had an abundance between 200 and 600.

Fig. 1
figure 1

The rank-abundance curve of per sample

Full size image

The Venn diagram showed the number of common microbial taxonomies between the frailty and control groups (Fig. 2). There were 1282 OTUs overlapping between the two groups, while 621 and 598 OTUs were exclusively belonging to the control and frailty groups, respectively.

Fig. 2
figure 2

The Venn diagram of OTUs distribution of the control and frailty groups

Full size image

Alpha diversity analyses

Rarefaction curve indicates the results of observed species corresponding to random sampling sequences per sample (Fig. 3). In all samples, a flat trend was observed as the number of sequences increased, indicating that maximum sequencing level had been reached and the sequencing results were reliable.

Fig. 3
figure 3

The rarefaction curve of random sequences per sample and their corresponding number of observed species

Full size image

For alpha diversity analysis, we calculated the Shannon index and Simpson index. The Shannon index of frailty and control groups were 4.39 ± 1.34 and 4.63 ± 0.65, respectively (p = 0.722, Fig. 4a); while the Simpson index of frailty and control groups were 0.84 ± 0.13 and 0.89 ± 0.04, respectively (p = 0.887, Fig. 4b). These results suggested that there was no statistical difference in alpha diversity of microbiota community between the two groups.

Fig. 4
figure 4

The Shannon index (a) and Simpson index (b) of the frailty group (red line) and control group (blue line)

Full size image

Beta diversity analyses

For beta diversity analysis, principal co-ordinates analysis (PCoA) and non-metric multidimensional scaling (NMDS) were performed. Although a major overlapping was observed in the PCoA plot, the PC2 axis shows a relatively larger distribution in the frailty group than the control group (Fig. 5a). Likewise, in the NMDS plot, the frailty group had a relatively larger distributed area than the control group. These results suggested that the frailty group had a larger beta diversity than the control group.

Fig. 5
figure 5

The distributed results of PCoA (a) and NMDS (b)

Full size image

Gut microbiota diversity analysis in the phylum and genus level

The diversity of intestinal microbe between the frailty and control groups were analyzed on phylum (Fig. 6a) and genus (Fig. 6b). The top-10 taxonomy categories and abundances of gut microbiota were similar between the two groups.

Fig. 6
figure 6

The percentages of gut microbiota diversity at phylum (a) and genus (b) levels

Full size image

The top-10 abundant phyla were Firmicutes, Bacteroidetes, Proteobacteria, Actinobacteria, Verrucomicrobia, Fusobacteria, Acidobacteria, Chloroflexi, Patescibacteria, and Epsilonbacteraeota. The top-10 abundant genera included Bac Bacteroides, Lac Lachnospiraceae UCG-008, Lac Blautia, Rum Faecalibacterium, Lac Lachnoclostridium, Str Streptococcus, Lac Roseburia, Tan Parabacteroides, Rik Alistipes, and Bif Bifidobacterium.

The community richness at family and genus levels were compared between the frailty and control groups. As shown in Table 3, there were 4 and 17 significant the gut microbiota composition at family and genus levels, respectively (p < 0.05), suggesting that microbial diversity and distribution are different between the frailty and control groups.

Table 3 Significant results of community richness differences between groups
Full size table

Clustering analyses

Hierarchical clustering and K-means clustering methods were used to investigate the grouping by microbial diversity indexes, including Shannon index, Simpson index, chao1, ACE, and goods coverage. As indicated in Fig. 7, the hierarchical clustering correctness rates of the frailty and control groups were 60% and 72.7%, respectively.

Fig. 7
figure 7

The dendrogram of hierarchical clustering results, label A and B are the frailty and control groups, respectively

Full size image

The K means clustering correctness rates of the frailty and control groups were 55.6% and 75%, respectively. These results suggested that the frailty and control groups indeed possessed different gut microbiota diversity, and the diversity indexes could be used to discriminate the frailty and non-frailty elderly.

Discussion

In this study, we investigated the gut microbiota of the frail elderly. The results showed that there were 1903 and 1880 OTUs identified in the control and frailty groups, respectively, with 1282 OTUs overlap between the two groups. The alpha diversity of microbiota community was similar between the two groups, whereas the frailty group had larger beta diversity than the control group. The top-10 taxonomy categories and the abundances of gut microbiota between two groups were mainly identical. As for the gut microbiota composition, however, 4 families and 17 genera were significantly different between the two groups. Taken together, these results suggested that frailty can affect gut microbiota diversity and compositions in late elderly hospitalized patients.

The gut microbiota composition and function are stable in adulthood up to 65–70 years, and then the inter-individual variability would be increased with the decline of biodiversity and a tendency to dysbiosis, leading to significant disturbance in host physiology [28]. Animal studies have demonstrated that dysbiosis of gut microbiota can elevate intestinal permeability [29], which promote endotoxin and other microbial products [e.g., lipopolysaccharide (LPS)] entering the circulation. Circulating LPS induces skeletal muscle inflammation and insulin resistance [30], which impairs both the metabolic and contractile characteristics of skeletal muscle. The relevant systemic inflammation has been demonstrated in the pathophysiology of sarcopenia patients [31]. These findings suggest the possibility of a “gut–muscle axis”, namely gut microbiota act as the mediator of the effects of nutrition on muscle cells. This phenomenon leads to a decrease in muscle strength and quality, further enhances insulin resistance, and ultimately leads to sarcopenia and physical frailty [32, 33].

Accumulating evidence has revealed the correlation between gut microbiota and frailty [34,35,36], as well as the composition of gut microbiota in the elderly [37,38,39], but little attention has been paid to the gut microbiota in frail elderly. In a small pilot study, van Tongeren et al. report that the abundance of Bacteroides/Prevotella and the Faecalibacterium prausnitzii is significantly reduced in frailer elderly, while the abundance of Enterobacteriaceae which can cause opportunistic infections is significantly increased in frailer individuals [23]. Jackson et al. have reported that frailty is negatively associated with the alpha diversity of gut microbiota. Species more abundant in the frail elderly included Eubacterium dolichum and Eggerthella lenta, while Faecalibacterium prausnitzii is less abundant in frail elderly [24]. F. prausnitzii is a key butyrate producer [35, 36] and has an anti-inflammatory effect on the gut [40]. The abundance of F. prausnitzii is decreased in patients with inflammatory bowel disease [41]. Likewise, Haran et al. have demonstrated that the nursing home elder with increasing frailty has lower abundances of butyrate-producing bacteria [42]. Nevertheless, the composition of the gut microbiota is greatly affected by diet, environment, and ethnicity [9,10,11], and may exhibit varying patterns in different countries and geographic regions.

The dominant gut microbial phyla in human are Firmicutes, Bacteroidetes, Actinobacteria, Proteobacteria, in which Firmicutes and Bacteroidetes representing 90% of gut microbiota [43]. Supporting this notion, our results demonstrated that the common dominant phyla between the two groups were Firmicutes, Bacteroidetes, Proteobacteria, and Actinobacteria, while the common dominant genera included Bac Bacteroides, Lac Lachnospiraceae UCG-008, Lac Blautia, and Rum Faecalibacterium. Of note, there was no significant difference in the top-10 taxonomy categories and the abundances of gut microbiota at both phylum and genus levels between the two groups. According to the Shannon index and the Simpson index, we found that the alpha diversity of gut microbiota was similar between the frailty group and the control group, which is inconsistent with Jackson et al.’s report [24]. This discrepancy may be attributed to the small sample size of this study. While in the beta diversity analysis, the frailty group had a relatively larger distribution area in both PCoA and NMDS plots as compared with the control group. One possible explanation might be that the various etiologies of frailty contributed to a larger difference in the beta diversity among the 15 frail elderly patients.

Comparing the gut microbiota at the genus level between the two groups, we found that Gemella, Lachnoanaerobaculum, Eubacterium[U1]_ruminantium_group, Azospira, Tyzzerella, Cloacibacterium, and EU455341_g genera were significantly less abundant in the frailty group, while the Acetanaerobacterium, Catenibacterium, [Ruminococcus]_torques_group, DQ801572_g, Ruminococcaceae_UCG-011, Prevotella_9, Olsenella, EF434341_g, KF843164_g, and Pseudoxanthomonas were more abundant in the frailty group as compared with the frailty group. Family Moraxellaceae was more abundant in the frailty group. These results indicated that the composition of gut microbiota shows different patterns in different countries and geographic regions. Of them, genera Prevotella, Olsenella, Catenibacterium, and Tyzzerella are reported to be associated with a lifetime risk of cardiovascular disease [44]. On the other hand, Olsenella bacteria are involved in endodontic infections in humans [45]. The pathogenic role of these microbiomes may be involved in the development of frailty in the elderly. However, it is unknown whether these relationships are the causes or consequences of frailty, and the underlying mechanism remains to be further investigated.

Reduction in microbiota-related metabolic capacity, such as lower short-chain fatty acids (SCFAs) levels may be associated with age-related frailty [46]. SCFAs mainly include acetic acid, propionic acid, and butyric acid, which stimulate the secretion of mucin by inducing hypoxia-inducible factor (HIF) in the intestinal epithelial cells, thereby strengthening the intestinal mucosal barrier. Meanwhile, SCFAs promote IL-8 release, accelerating repair of the intestinal mucosal barrier and modulate inflammation, oxidative stress [47]. High frail individuals are reported to have a decrease in the abundance of Eubacteria, Faecalibacterium, and Lactobacilli [48]. We also found that Eubacteria was less abundant in the frailty elderly (p = 0.023); however, there was no significant difference in Lactobacilli and Faecalibacterium between the two groups. Eubacterium is considered as protective colonic bacteria by producing SCFAs (especially butyrate) [49] and modulating the immune system [50]. Thereby, a decrease in Eubacterium abundance may negatively affect gut health or induce systemic effects. The Eubacterium decreased in the frail elderly may support the protective effect of SCFAs on human guts. However, the detailed mechanism is needed to be further investigated. Our result showed that genus Prevotella was more abundant in the frail group, which is in line with van Tongeren et al.’s report [23]. For other bacterial genera with an abundant change in our study, their potential correlations with the development of frailty remain to be elucidated.

Several limitations of this study should be pointed out. First, the sample size of this study is small, and the findings of this study should be validated in a large study. In addition, hospitalization may have an impact on gut microbiota composition due to exposure to nosocomial pathogens, disruption of usual dietary and lifestyle habits, presence of acute diseases and need for acute treatments. Nevertheless, this study was conducted in the elderly health care ward of our hospital, and all patients were admitted to a single ward. In addition, the patients with acute infectious diseases, antibiotics or probiotics were excluded. Thus, the impact of hospitalization could be minimized. Moreover, we cannot determine whether frailty is the cause, consequence or bystander of the changes in gut microbiota diversity and composition. All these studies should be addressed in the future study.

In summary, our study demonstrated that frailty can change the abundance and composition of gut microbiota in Chinse frail late elderly hospitalized patients, further highlighting an association between the gut microbiota alterations and development of frailty. Our finding may provide a theoretical basis for further study on the pathological mechanism for the development of frailty.