Introduction

Vitamin B6 is a water-soluble vitamin with several chemically related forms including pyridoxine, pyridoxal, pyridoxamine, and their respective phosphorylated derivatives. Pyridoxal 5’-phosphate (PLP), the main biologically active form, serves as a co-factor for more than 140 distinct biochemical reaction related to amino acid metabolism, neurotransmitter synthesis, heme production, carbohydrate metabolism and so on [1]. Meat, fish and fowl are good sources of vitamin B6. Bacteria residing in the intestine can also synthesize vitamin B6 through deoxy 5-xylulose (DXP)-dependent or DXP-independent pathway [2,3,4]. Vitamin B6 is absorbed mainly in the jejunum and ileum. Decreased levels of vitamin B6 are frequently found in patients with systemic inflammation, liver disease and rheumatoid arthritis and some women with type 1 diabetes [5,6,7].

Crohn’s disease (CD) is a chronic, recurrent inflammatory disease that mainly affect the gastrointestinal tract [8]. Malnutrition and micronutrient deficiencies are common in CD because of aggravated intestinal losses, decreased oral intake, frequently small bowel involvement, diminished absorptive capacity and aberrant hypermetabolism status [9].

Some previous studies have reported significant decrease of vitamin B6 in CD patients compared to healthy control [2, 10, 11]. These studies were carried out early when biological agents were not be widely used for CD therapy. We speculated that the vitamin B6 deficiency rate of CD patients remained high while newer treatments have become widespread. Our present study aims to estimate the prevalence of vitamin B6 abnormalities in patients with CD, to identify risk factors associated with an abnormal B6 concentration and to explore the association between vitamin B6 status and intestinal microbiota.

Methods

Subjects

All CD patients with detection of vitamin B6 who were evaluated in the Department of Gastroenterology, the Shanghai Tenth People’s Hospital of Tongji University (Shanghai, China) between July 1, 2018 to August 1, 2020 were studied retrospectively (n = 364). We used Crohn’s Disease Activity Index (CDAI) to assess the severity of disease in CD patients. If patients were previously identified as vitamin B6 deficiency or were supplementing vitamin B6 at the time of initial visit, they were excluded (n = 4). Fecal samples were collected from another 20 CD patients between April 1, 2021 and August 31, 2021. Controls were included from patients evaluated at our center with a diagnosis of UC (n = 55). Approval for this study was obtained through the Institutional Review Board for Clinical Research of the Shanghai Tenth People’s Hospital of Tongji University.

Vitamin B6 detection

Serum vitamin B6 concentrations of all patients were collected from the computerized laboratory data. If more than one vitamin B6 concentration was available, we only chose the initial concentration. For the primary analysis, vitamin B6 concentrations <4.9 ng/mL were considered abnormal according to the laboratory standards of our hospital.

Covariates

Covariates of interest were selected beforehand. They were determined by retrospective review of patients’ medical records. These included age, sex, disease duration, disease location, disease behavior, existence of perianal disease, existence of extraintestinal manifestations and intestinal surgery history. Current usage of medical therapy including 5-aminosalicylate, immunosuppressors, corticosteroids, biologicals, probiotics and enteral nutrition was also considered. Other laboratory indices collected at the time of vitamin B6 measurement included erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP).

DNA extraction and 16S rRNA gene sequencing

Microbial genome DNA was extracted from fecal samples for amplification based on V4–V5 regions of 16S rRNA gene. Sequencing was performed on an Illumina MiSeq platform according to the standard protocols (Shanghai BIOZERON Co., Ltd).

Processing of sequencing data

Raw fastq files were first demultiplexed using in-house perl scripts according to the barcode sequences information for each sample. Passed sequences were clustered into operational taxonomic units (OTUs) at 100% similarity (identical) using the Deblur denoising algorithm [12]. The phylogenetic affiliation of each 16S rRNA gene sequence was analyzed by uclust algorithm (http://www.drive5.com/usearch/manual/uclust_algo.html) against the silva (SSU138.1) 16S rRNA database using confidence threshold of 80% [13].

Alpha- and beta-diversity analyses

The rarefaction analysis based on Mothur v.1.21.1 was conducted to reveal the alpha diversity indices [14]. Q2‐diversity in Qiime2 was used to calculate beta-diversity metrics [15]. Permutational multivariate analysis of variance (PERMANOVA) was used to compare the beta-diversity differentiations between CD patients with normal vitamin B6 and abnormal vitamin B6 [16, 17].

Differential microbiota enrichment analysis

Differential enrichment between CD patients with normal vitamin B6 and abnormal vitamin B6 was analyzed by linear discrimination analysis effect size (LEfSe) [18]. We considered linear discriminant analysis (LDA) absolute values > 2 and P values < 0.05 as statistically significant. Differential family or genus that showed the highest correlation with vitamin B6 level were identified as biomarkers to differentiate two CD groups. The discriminatory power of candidate biomarkers was assessed by receiver operating characteristic (ROC) curves and area under the ROC curve (AUC) using R software [19]. Spearman correlation analysis was also performed between genera enriched in abnormal vitamin B6 group. Then we selected genera with strong correlations (r absolute values > 0.5) to construct a correlation network using Cytoscape V.3.8.2 [20].

Metabolic function prediction of the microbial genes

Phylogenetic Investigation of Communities by Reconstruction of Unobserved States 2 (PICRUSt2) program was used to predict metabolic functions based on Kyoto Encyclopedia of Genes and Genomes (KEGG) database and MetaCyc database [21,22,23]. LEfSe was used to calculate differential metabolism-related enzymes, KEGG orthology (KO) and metabolic pathways between two CD groups.

Statistical analysis

Summary statistics were calculated for comparison of CD patients and UC controls. The prevalence of abnormal vitamin B6 and 95% confidence intervals (CI) were determined. Chi-squared test was used to comparing the prevalence of abnormal vitamin B6 between patients with CD and UC. Continuous variables are presented as median (interquartile range [IQR]) and between-group comparisons were made using Wilcoxon rank-sum test. Covariates of interest were compared between CD patients with abnormal vitamin B6 concentrations and those with normal concentrations. Unadjusted odds ratios (ORs) were estimated and statistical significance was determined using univariable logistic regression analysis. For multivariable logistic regression analysis, “Forward: Conditional” method of SPSS was used and adjusted ORs and 95% CI were estimated. Wilcoxon rank-sum test was used to compare alpha diversity between two CD groups. All statistical analyses were performed by SPSS 20.0 software (SPSS Inc., Chicago, IL, United States) or R software. The plots were drawn by Prism GraphPad 7 (GraphPad Software; La Jolla, CA, USA), Cytoscape V.3.8.2 and R software. Two tailed P < 0.05 was considered statistically significant.

Results

Patient characteristics

Three hundred and sixty CD patients were included in the study. Compared to UC patients, the prevalence of a vitamin B6 abnormality was significantly higher in CD [CD: 38.3% (95% CI 33.3–43.3%) versus UC: 20.0% (95% CI 9.4–30.6%), P = 0.01] (Fig. 1A). However, the median vitamin B6 concentration between the two groups showed no significant difference (CD: 6.26 ng/mL versus UC: 6.55 ng/mL, P = 0.312) (Fig. 1B). Baseline characteristics of CD and UC patients are described in Table 1.

Fig. 1: The prevalence of vitamin B6 abnormity and the concentrations of vitamin B6.
figure 1

A The prevalence of vitamin B6 abnormity in CD and UC. B The serum vitamin B6 concentrations in CD and UC.

Full size image
Table 1 Clinical and Demographic Characteristics of CD and UC patients.
Full size table

Correlation between clinical characteristics and vitamin B6 levels

In CD patients, there was no significant correlation between vitamin B6 and CDAI (r = 0.019, P = 0.721), CRP (r = 0.03, P = 0.571), ESR (r = 0.03, P = 0.616), hemoglobin (r = −0.012, P = 0.817), or albumin (r = 0.028, P = 0.594) (Fig. S1A–E). However, there was a weak negative correlation between vitamin B6 and BMI (r = −0.109, P = 0.038) (Fig. S1F). Besides, we compared critical categorical characteristics between CD patients with abnormal and normal vitamin B6 (Fig. S2). We found that only the distribution of lesion location showed significantly different.

The median vitamin B6 level for CD patients in remission was 6.54 ng/mL, showing no significant difference compared with patients in a mildly active disease state (5.94 ng/mL) and patients in a moderately/severe active disease state (5.78 ng/mL) (Fig. 2A). Besides, the level of vitamin B6 was significantly lower in patients with colonic lesion (L2) (3.89 ng/mL) than those with ileal or upper gastrointestinal lesion (L1/L4) (6.57 ng/mL, P = 0.009) and those with ileocolonic lesion (L3) (6.54 ng/mL, P = 0.067) (Fig. 2B). In term of disease behavior, the vitamin B6 level showed no significant difference (non-stricturing/penetrating CD: 6.25 ng/mL, stricturing CD: 6.19 ng/mL, penetrating CD: 5.66 ng/mL, stricturing/penetrating CD: 7.3 ng/mL) (Fig. 2C). CD patients with perianal disease (5.75 ng/mL) had similar vitamin B6 level with patients without perianal disease (6.54 ng/mL) (Fig. 2D). There was also no significant difference between patients without surgery history and patients experiencing surgery, regardless of surgical location (no surgery: 6.53 ng/mL, ileal resection: 4.78 ng/mL, ileocecal resection 6.12 ng/mL, colonic resection: 5.43 ng/mL) (Fig. 2E). In addition, compared with patients without extraintestinal manifestations (6.12 ng/mL), the vitamin B6 level in patients with extraintestinal manifestations (6.76 ng/mL) presented no statistical difference (Fig. 2F).

Fig. 2: Correlation between clinical characteristics and vitamin B6 levels.
figure 2

A Correlation between serum vitamin B6 levels and disease activity. B Correlation between serum vitamin B6 levels and disease location. C Correlation between serum vitamin B6 levels and disease behavior. D Correlation between serum vitamin B6 levels and perianal lesions. E Correlation between serum vitamin B6 levels and intestinal surgery history. F Correlation between serum vitamin B6 levels and extraintestinal manifestations. **＜0.01.

Full size image

Evaluation of factors associated with vitamin B6 abnormality in CD patients

On univariate analysis, the results showed that disease location was significantly associated with vitamin B6 status. Compared to patients with L2, patients with L1/L4 (OR 0.265, 95% CI 0.130–0.539) and L3 (OR 0.35, 95% CI 0.182–0.670) were less likely to have an abnormal vitamin B6. Besides, existence of extraintestinal manifestations (OR 0.488, 95% CI 0.250–0.951) and usage of immunosuppressor (OR 0.546, 95% CI 0.324–0.920) were associated with normal vitamin B6 (Table 2). Univariate analyses revealed no significant differences in the prevalence of an abnormal B6 concentration based on gender, age, disease duration, previous intestinal surgery, disease activity and so on.

Table 2 Univariate Associations Between Covariates and Abnormal B6 Concentration in Patients with Crohn’s Disease (n = 360).
Full size table

Furthermore, variables with P value < 0.20 (disease location, extraintestinal manifestations, surgery history, usage of immunosuppressor, usage of biologicals, usage of corticosteroids) were included in a multivariable logistic regression analysis. We found that L1/L4 (OR 0.211, 95% CI 0.099–0.449), L3 (OR 0.378, 95% CI 0.192–0.745), extraintestinal manifestations (OR 0.42, 95% CI 0.199–0.814), ileal resection (OR 2.777, 95% CI 1.208–6.384), and usage of immunosuppressor (OR 0.462, 95% CI 0.264–0.808) were independent risk or protective factors for the occurrence of vitamin B6 abnormality (Table 3).

Table 3 Multivariate Analysis of Factors Associated with Abnormal B6 Concentration in Patients with Crohn’s Disease (n = 360).
Full size table

The same analysis was also performed for the enrolled UC patients. However, the results showed that none of the factors we were interested in were significantly associated with vitamin B6 status of UC patients (Table S1).

Microbial diversity and differential enrichment analysis

We further explored the change of gut microbiota associated with abnormal vitamin B6 in CD patients (Table S2). Firstly, we found alpha diversity (richness, Shannon index, Simpson index, Chao index, and ACE index) had no statistical difference in the two CD groups (Fig. 3A–E). As for beta diversity, we calculated Bray-Curtis dispersion, Jaccard, unweighted UniFrac, and weighted UniFrac distance metrics. Among them, unweighted UniFrac (P = 0.003) showed a statistical difference between two groups (Table 4, Fig. 3F). Therefore, the overall structure of gut microbiome also changed when CD patients had an abnormal vitamin B6 status.

Fig. 3: Microbial diversity and differential enrichment analysis.
figure 3

A–E Comparation of alpha diversity (richness, Shannon index, Simpson index, Chao index and ACE index) between two CD groups. F Principal coordinates analysis showing cecal microbiota b-diversity based on the unweighted Bray-Curtis distances at the OTU level.

Full size image
Table 4 Differences of beta diversity indexes between CD patients with normal B6 and abnormal B6.
Full size table

Recently, a study assessed the genomes of 256 common human-gut bacteria for the presence of vitamin B6 biosynthesis pathways and found that the majority of Actinobacteria, Bacteroidetes, and Proteobacteria had the ability to synthesize vitamin B6 [24]. We compared the relative abundance of these three phyla between two CD groups (Fig. S3A–C). The relative abundance of Bacteroidetes showed a decreasing tendency in CD patients with abnormal vitamin B6, although the difference was not statistically significant. Furthermore, by applying the LEfSe algorithm, we performed differential enrichment analysis between the two CD groups in the levels of family and genus, respectively. In family level, we obtained 19 differential families that were all enriched in patients with abnormal vitamin B6 (Fig. S4A, B). Correlation analysis between the 19 families and vitamin B6 level was also performed using Spearman correlation coefficient (Table S3). Gemmatimonadaceae showed the highest correlation with vitamin B6 (r = −0.746, P = 1.6e–4). ROC curve implied that the abundance of Gemmatimonadaceae could potentially be applied to differentiate vitamin B6 levels in our samples (AUC = 0.995) (Fig. S4C). In terms of genus level, among all the 19 differential genera obtained, Clostridioides showed higher abundances in patients with normal vitamin B6, while the other 18 genera had higher abundances in patients with abnormal vitamin B6 (Fig. 4A, B). After ruling out unidentified genera, RB41 showed the highest correlation with vitamin B6 (r = −0.607, P = 4.6e-3) (Table S4). It attained good diagnostic accuracy to distinguish abnormal vitamin B6 and normal vitamin B6 (AUC = 0.92) (Fig. 4C). Besides, we also performed correlation analysis between genera enriched in patients with abnormal vitamin B6 (Fig. S5A). Then we constructed bacteria interaction network based on the strongly correlated genera (Spearman correlation coefficient >0.5) (Fig. S5B).

Fig. 4: Enrichment of genera in CD patients.
figure 4

A, B Differential enrichment of genera in CD patients with abnormal and normal vitamin B6. C Prediction of vitamin B6 status by RB41.

Full size image

Functional analysis of metabolic pathways

PICRUSt2 was used to predict microbiota metabolic functions. We found that some enzymes and metabolic pathways between the two groups were significantly different. Among them, 23 enzymes and 19 KOs were highly enriched in patients with normal vitamin B6, while 25 enzymes and 1 KO were highly enriched in patients with abnormal vitamin B6 (Figs. 5, and S6). In terms of differential metabolic pathways, we used both KEGG and metacyc database to annotate metabolic pathways. In metacyc database, 10 metabolic pathways [such as incomplete reductive TCA cycle, pentose phosphate pathway and so on] were highly enriched in patients with normal vitamin B6 and 7 pathways (such as peptidoglycan biosynthesis, starch degradation and so on) were highly enriched in patients with abnormal vitamin B6 (Fig. 6A). In KEGG database, 4 metabolic pathways were highly enriched in patients with normal vitamin B6, mainly associated with carbohydrate metabolism, phosphonate and phosphinate metabolism and lysosome function; 7 pathways were highly enriched in patients with abnormal vitamin B6, associated with nicotinate and nicotinamide metabolism, arachidonic acid metabolism, polyketide sugar unit biosynthesis and mRNA surveillance pathway and so on (Fig. 6B).

Fig. 5
figure 5

Differential enrichment of enzymes in CD patients with abnormal and normal vitamin B6.

Full size image
Fig. 6: The analysis of differential enrichment of pathway.
figure 6

A Differential enrichment of pathways in metacyc database in CD patients with abnormal and normal vitamin B6. B Differential enrichment of pathways in KEGG database in CD patients with abnormal and normal vitamin B6.

Full size image

Discussion

Inflammatory bowel disease (IBD) is regarded as a risk factor for malnutrition and micronutrient deficiencies. In present study, we analyzed the prevalence and associated impact factors of vitamin B6 abnormality in CD patients. The incidence rate of abnormal vitamin B6 assessed at our tertiary medical center was significant and approached 40%, which is similar to a prior study evaluating the vitamin B6 deficiency of CD patients in Canada (45.7%) [10]. Besides, in our study vitamin B6 abnormity was more common in CD patients than in UC patients. This is also consistent with previous studies and may reflect the obvious heterogeneity between CD and UC [10].

Persistent inflammation may contribute to the micronutrient deficiencies [25]. A prior study showed that IBD patients with active disease had more frequent vitamin B6 abnormity than patients with quiescent disease [2]. However, in our study we did not observe the difference of vitamin B6 status between active patients and patients in remission. This discrepancy may be caused by the small sample size (61 patients included) of the previous study, which introduced α error in hypothesis test.

After univariable and multivariable logistic regression analysis, we found that disease location (Montreal classification) was independently associated with the risk of abnormal vitamin B6 concentrations. Intuitively patients with small intestinal lesions were more likely to have vitamin B6 abnormity due to impaired absorption, but in fact patients with L2 were prone to manifest as abnormal vitamin B6. Accordingly, the vitamin B6 levels in L2 patients were lower than those in L1/L4 and L3 patients. Besides, prior ileal resection was an independent risk factor for vitamin B6 abnormity. Therefore, the integrity of small intestine rather than the distribution of lesions may be more important for the maintenance of normal vitamin B6 concentrations. In addition, current usage of immunosuppressors was an independent protective factor, further supporting that inflammation is a contributor of vitamin B6 abnormity in CD patients [25]. Surprisingly, compared to patients with extraintestinal manifestations, those without extraintestinal manifestations were more likely to have an abnormal vitamin B6. The mechanism underlying this phenomenon remains to be further explored.

Given that some intestinal bacteria can synthesize vitamin B6 and dysbiosis is a typical characteristic of CD patients, we speculate that vitamin B6 abnormity may be associated with specific alteration of intestinal microbiota [26, 27]. β diversity analysis based on the unweighted UniFrac distance metrics demonstrated that the overall microbiota structure was significantly different in samples from patients with abnormal vitamin B6 compared to samples from patients with normal vitamin B6. We also investigated which bacterial taxa were distinct between the two groups in levels of family and genus. In the level of family, Lactobacillaceae was significantly enriched in patients with abnormal vitamin B6. Some species in Lactobacillaceae, such as Lactobacillus casei and Lactobacillus delcruebkii, require specific forms of vitamin B6 for growth [28]. Overgrowth of these bacteria consumes more vitamin B6 and interferes with normal absorption by host, which may partially explain the abnormity of vitamin B6 in these patients. Among these enriched families, Gemmatimonadaceae had the strongest negative correlation with vitamin B6 level. The alteration of Gemmatimonadaceae in IBD is rarely observed. Whether Gemmatimonadaceae can affect vitamin B6 absorption remains to be further studied. Several genera also showed difference between the two groups. Butyrate, a kind of short chain fatty acids, is produced by a small number of Faecalibacterium [29, 30]. Butyrate plays a critical role in maintaining intestinal mucosal homeostasis by providing energy for epithelial cells and regulating different immune cells [31,32,33,34]. On the other hand, Excessive accumulation of butyrate may impair proliferation of intestinal stem cells (ISCs) and inhibit wound repair after epithelial injury [35, 36]. The loss of Faecalibacterium prausnitzii has been observed in IBD patients [29]. However, Faecalibacterium was found to be increased in patients with abnormal vitamin B6 compared to patients with normal B6. Among the genera enriched in patients with abnormal vitamin B6 exist many pairs of strong correlations, which implies that these genera have complicated interaction, collectively participating in the development of CD.

We also perform metabolic function prediction. We did not find the alteration of vitamin B6 metabolism pathway (including pyridoxal 5’-phosphate biosynthesis I and superpathway of pyridoxal 5’-phosphate biosynthesis and salvage) (Fig. S7). However, pentose phosphate pathway was down-regulated in the abnormal vitamin B6 group (Fig. 6A). In the DXP-dependent pathway, the raw material, erythrose 4-phosphate (E4P) is derived from the pentose phosphate pathway [37]. In addition, we found that some pathways related to energy supply and basic biological function, such as TCA cycle, was significantly decreased in patients with abnormal vitamin B6.

There were some limitations in our study. CD patients in our center are usually sicker or require more surgical interventions, which may overestimate the prevalence of vitamin B6 abnormity. Besides, our sample size for flora research was small. In the future, we need more samples to support our findings.

In conclusion, our results indicate that abnormal vitamin B6 concentrations are common in CD patients. Disease location, existence of extraintestinal manifestations, history of intestinal surgery and usage of immunosuppressor are independent impact factors for vitamin B6 abnormity. We recommend routine screening for vitamin B6 abnormity as part of CD treatment. Besides, intestinal microbiome structure alteration and taxonomic variations may partially explain the occurrence of vitamin B6 abnormity.