Gut microbiota composition and functionality are associated with prolactin-secreting pituitary adenoma and serum lipid levels

Abstract

Background

The changes in gut microbiota composition and their correlations with serum lipid levels in prolactin-secreting pituitary adenoma (PPA) patients remain unknown. This study investigates these alterations and associations and explores microbial markers for PPA diagnosis.

Methods

A total of 101 participants were enrolled, comprising 72 PA patients (31 with prolactin-secreting adenomas and 41 with non-functioning adenomas, i.e., PPA and NFPA groups) and 29 age and sex-matched healthy controls (HC). Utilizing 16 S rRNA gene amplicon sequencing, we examined the gut microbiota community in the PPA group and investigated its associations with serum lipid levels.

Results

Our results revealed significantly reduced microbial ecosystem richness and evenness in PPA patients compared to healthy controls. The PA group, especially PPA, exhibited substantial alterations in gut microbiota structure, including increased abundance of gram-negative pathogenic bacteria such as Desulfovibrio and Enterobacter, and decreased levels of probiotic bacteria like Bifidobacterium. We also identified significant positive correlations between PPA-enriched bacteria and serum lipid levels. A biomarker panel (including Bifidobacterium, Dorea, Blautia, Morganella, Desulfovibrio, and Enterobacter) demonstrated good performance in differentiating between PA patients and healthy controls (AUC: 0.843). Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis unveiled dysregulations in fundamental physiological pathways, particularly lipid metabolism, within the PPA group.

Conclusions

Our findings suggest that PA patients, particularly those with PPA, exhibit distinct host-microbe interactions compared to healthy controls. Notably, the intestinal flora, particularly in the PPA microenvironment, may play a role in contributing to tumor development by impacting fundamental metabolism, especially lipid metabolism.

Importance

Our study revealed that the gut microbiome is tightly linked to PPA, and the disturbed microbiome may regulate the risk of PPA in lipid metabolism. These findings, including the development of a biomarker panel, suggest the potential of intestinal flora as a diagnostic and predictive tool, emphasizing its significance as a preventive target for PPA.

Graphical Abstract

Supplementary Information

The online version contains supplementary material available at 10.1186/s12866-025-04592-2.

Introduction

Trillions of microbes, collectively referred to as intestinal flora, are found in the human gut, which function as an “external organ” influencing host physiology [1]. Their metabolites support tissue development, immune balance, and metabolic homeostasis. Their great diversity and functional potential are revealed by DNA sequencing. Obesity, heart disease, and even neurological conditions are associated with dysbiosis, or changes in the composition or activity of microorganisms [2, 3].

Pituitary adenomas (PAs) represent the most common tumors of the sellar region and are classified as either nonfunctioning pituitary adenomas (NFPAs) or hormone-secreting adenomas [4]. The most common among them are prolactin-secreting adenomas (PPAs), which account for approximately 50% of PAs. They typically exhibit slow growth, with fewer than 15% demonstrating progression after diagnosis [5–7]. The first-line treatment consists of dopamine agonists, which efficiently lower prolactin levels and induce tumor shrinkage. However, around 15% of patients exhibit resistance or intolerance [7, 8]. Consequently, research on additional targets to improve the treatment of PPA remains ongoing [9, 10].

The discovery of the brain-gut axis [11] has sparked increased research into the relationship between gut microbiota and various nervous system diseases, including Alzheimer’s disease [12] and different types of cerebral tumors [13, 14]. Meanwhile, emerging evidence suggests strong connections between gut microbiota and different types of pituitary adenomas. In this respect, Lin B. et al. [15] have highlighted profound changes in gut microbiota among individuals with GH-secreting PA, including decreased bacterial diversity, enriched abundance in Oscillibacter and Enterobacter, and functional dysbiosis. In addition, Nie D. et al. [16] discovered that gut microbiota in GH-secreting PA could regulate tumor growth by influencing the immune system and the expression of PD-L1. Moreover, Hu J. et al. [17] illustrated the significant role of gut microbiota in the invasive process of various PAs. Meanwhile, Luzardo-Ocampo I. et al. [18] revealed that prolactin could impact the composition of gut microbiota, and imbalances in prolactin levels could lead to metabolic diseases. Despite the tight relationships observed between gut microbiota, PA, and prolactin, research on changes in gut microbiota in patients with PPA remains insufficient. Therefore, there is a growing focus on comprehensively understanding the functional gut microbiota in patients with PPA.

Herein, we sought to investigate the role of gut microbiota in PPA, especially in metabolism, by analyzing the gut microbiota and clinical parameters of healthy controls (n = 29) and patients with PPA (n = 31) and NFPA (n = 41). Furthermore, we aimed to identify a microbial biomarker panel capable of distinguishing patients with PPA and NFPA from healthy individuals. In addition to identifying potential diagnostic biomarkers and therapeutic targets for disease management, this study provides a new perspective on the pathogenesis and progression of PPA. The regulation of gut microbiota holds promise as a potential adjuvant therapeutic tool for treating this patient population in the future.

Materials and methods

Participants and study design

All 101 participants provided informed consent, comprising 29 healthy controls (HC) and patients with prolactin-secreting pituitary adenomas (n = 31) and non-functioning pituitary adenomas (n = 41). Patients were newly diagnosed cases based on the criteria outlined in previous guidelines [8] by the Department of Neurosurgery at the Affiliated Hospital of Yangzhou University between January 2022 and September 2023. Details of the exclusion criteria are presented in the screening flowchart (Fig. 1). At the time of fecal specimen collection, detailed demographic and bio-clinical information were simultaneously recorded. In addition, dietary information was subsequently obtained via a structured follow-up employing the validated Mini-EAT (Mini Eating Assessment Tool) questionnaire, as described in the Supplementary Figure S4. The study received approval from the Ethics Committee of the Affiliated Hospital of Yangzhou University.

Fig. 1.

Flowchart showing derivation of study participants. *Patients may be excluded for more than one factor

Sample collection and DNA extraction

The fecal specimen collection and genomic DNA extraction were conducted following a standardized procedure. Fecal samples were collected within 6 h of hospital admission using sterile cotton swabs, promptly preserved in a dedicated solution, and stored at −80℃ for further examination. For genomic DNA extraction, we utilized the PowerMax extraction kit with the obtained DNA stored at −80℃ to maintain its integrity. The NanoDrop ND-1000 spectrophotometer from Thermo Fisher Scientific was used for assessing DNA concentration and quality.

16 S rRNA amplicon pyrosequencing

Using the specific primers (515 F and 806R), we conducted PCR amplification targeting the V4 region of the bacterial 16 S rRNA gene. To enable multiplex sequencing, sample-based barcodes were strategically integrated into TruSeq adaptors. The thermal cycling conditions followed previously established protocols [3]. Following PCR amplification, the products underwent purification using AMPure XP Beads from Omega Bio-Tek in the USA. Quantification was carried out using the PicoGreen dsDNA Assay Kit from Invitrogen in the USA. After quantification, equal amounts of the amplification products were combined, and sequencing was performed on the Illumina HiSeq4000 platform using paired-end 2 × 150 bp sequencing.

Sequencing data analysis

The initial stage of the analysis involved matching paired-end reads to their respective samples using dedicated barcodes, followed by the removal of these barcodes along with primer sequences. Subsequently, the FLASH tool was employed to merge truncated reads. Amplicon Sequence Variant (ASV) identification was performed using Vsearch (v2.22.1.), which included dereplication, quality assessment, application of the UNOISE2 algorithm for sequence denoising, elimination of chimeras [19], and alignment with ASVs using a 100% similarity threshold. To address potential discrepancies in sequencing depth, normalization was applied to the ASV sequence counts in the table, ensuring uniform sums of ASV sequences across samples and normalizing values to 1 to denote relative abundances. Additionally, a representative sequence was chosen from each ASV using default parameters. Using QIIME2 (2022.2) [20], rep-seqs and ASV table files were imported, and ASVs sequences that accounted for less than 0.001% of the total across all samples were eliminated. Taxonomic assignments for ASVs were carried out using QIIME2’s weighted taxonomic classifiers, and the resultant taxonomy was collapsed from phylum to genus through the “qiime taxa collapse” command. Alpha diversity indices (Simpson, Shannon, Chao1, PD whole tree, ace, observed species indices) were used to assess microbial diversity complexity in each sample [21]. Beta diversity analysis, utilizing UniFrac distance metrics, aimed to investigate variations in the structure of microbial communities. These structural variations were visualized through both principal coordinate analysis (PCoA) and nonmetric multidimensional scaling (NMDS) [22]. To assess statistical significance, the analysis of similarities (ANOSIM) method was applied [14]. Differentially abundant taxa were identified by employing linear discriminant analysis (LDA) in conjunction with the effect size calculation (LEfSe) [23]. For pathway enrichment analysis, the Kyoto Encyclopedia of Genes and Genomes (KEGG) and Phylogenetic Investigation of Communities by Reconstruction of Unobserved States (PICRUSt) database were utilized [24]. Furthermore, the Statistical Analysis of Metagenomic Profiles (STAMP, v2.1.3) software was used to perform in-depth analysis of the output files. To comprehensively assess the composition and function of gut microbiota, BugBase [25], a tool designed to evaluate high-level phenotypes in the microbiome, was also integrated into the analysis.

Statistical analyses

Data analyses, covering both clinical and bioinformatic components, were primarily conducted utilizing SPSS (v26.0) and R packages (v4.1.0). Data representing continuous variables were represented as mean ± standard deviation, with group differences assessed through various statistical tests, including independent t-tests, one-way analysis of variance, or Mann-Whitney U-tests. To compare microbial abundance and diversity in fecal samples among PA patients and healthy controls, the Wilcoxon rank-sum test was employed. To assess differences in UniFrac distances for making pairwise comparisons between the groups, either Student’s t-test or the Monte Carlo permutation test was utilized. Genus-level heatmaps were constructed based on the nonparametric Wilcoxon test (q < 0.1). Categorical variables underwent analysis using the chi-square test or Fisher’s exact test. Adjusted p-values, using the Benjamini-Hochberg false discovery rate (FDR) threshold below 0.05, were used to determine statistical significance in group comparisons, which consist of microbial abundance, BugBase phenotypes, correlation analyses, and KEGG pathway analyses. LEfSe results followed the standard LEfSe procedure and were not subjected to further FDR correction.

Results

Summary of clinical parameters

Table 1 provides a comprehensive overview of the clinical parameters for all participants. Importantly, no significant demographic disparities were observed between the groups (except prolactin, p < 0.001). The absence of substantial variations in demographic factors suggested that no confounding factors influenced group discrimination, establishing a foundation of comparability before the initiation of this study.

Table 1.

The baseline characteristics of all participants

	PPA	NFPA	HC	HC vs. PA	PPA vs. NFPA
	(n=31)	(n=41)	(n=29)	p value	p value
Age (Mean ± SD)	53.18±10.35	54.42±9.65	52.45±10.37	0.206	0.684
Gender				0.287	0.173
Male	8 (25.8%)	22 (53.7%)	15 (51.7%)
Female	23 (74.2%)	19 (46.3%)	14 (48.3%)
BMI	22.21±1.78	22.05±1.54	22.52±1.67	0.819	0.462
Mini-EAT Score	45.97±7.46	46.24±6.88	48.83±5.52	0.07	0.873
Smoking				0.423	0.242
Absence	16 (51.6%)	15 (36.6%)	16 (55.2%)
Presence	15 (48.4%)	26 (63.4%)	13 (44.8%)
Drinking				0.803	0.528
Never	22 (71%)	27 (65.9%)	19 (65.5%)
<1 standard drink per day	4 (12.9%)	5 (12.2%)	4 (13.8%)
≥1 standard drink per day	5 (16.1%)	9 (22%)	6 (20.7%)
Hypertension				0.746	0.649
Negative	25 (80.6%)	28 (68.3%)	22 (75.9%)
Positive	6 (19.4%)	13 (31.7%)	7 (24.1%)
Diabetes				0.138	0.956
Negative	27 (87.1%)	35 (85.4%)	27 (93.1%)
Positive	4 (12.9%)	6 (14.6%)	2 (6.9%)
Excrement regularity				0.892	0.713
Yes	28 (90.3%)	37 (90.2%)	27 (93.1%)
No	3 (9.7%)	4 (9.8%)	2 (6.9%)
Knosp grade				-	0.571
0	3 (9.6%)	6 (14.6%)	-
1	10 (32.3%)	11 (26.8%)	-
2	12 (38.7%)	18 (43.9%)
3	4 (12.9%)	5 (12.2%)
4	2 (6.5%)	1 (2.4%)
ApoA (g/L, Mean ± SD)	1.53±0.27	1.38±0.35	1.45±0.26	0.658	0.551
ApoB (g/L, Mean ± SD)	0.95±0.14	0.87±0.22	0.89±0.19	0.438	0.209
LDL (mg/dl,  Mean ± SD)	115.62±25.30	102.11±18.63	105.80±14.42	0.399	0.108
HDL (mg/dl,  Mean ± SD)	65.86±14.98	71.53±15.27	70.86±18.23	0.600	0.201
TC (mg/dl, Mean ± SD)	215.80±30.56	213.52±28.85	207.61±31.52	0.334	0.791
TG (mg/dl, Mean ± SD)	72.07±20.48	71.85±21.54	69.65±20.88	0.746	0.802
Prolactin (ng/ml, Mean ± SD)	260.82±30.50	8.83±1.51	9.51±2.10	<0.001	<0.001

HC, healthy group; PPA, prolactin-secreting pituitary adenoma; NFPA, nonfunctioning pituitary adenomas; ApoA, apolipoprotein A; ApoB, apolipoprotein B; LDL, low-density lipoprotein; HDL, high-density lipoprotein; TC, total cholesterol; TG, triglyceride; SD: Standard deviation. Mini-EAT: Mini Eating Assessment Tool

Distribution of intestinal flora in the patients with PA and the healthy controls

In examining the intestinal microbial community of individuals with pituitary adenoma, a comprehensive comparison of the relative taxon abundance among different groups was conducted. At the phylum level, all three groups displayed comparable compositions, with Firmicutes, Bacteroidetes, Proteobacteria, Actinobacteria, and Fusobacteria being the predominant five microbes (Fig. 2A-C). Similarly, at the family level, the analysis revealed that all three groups shared the top 6 microbes, including Lachnospiraceae, Tannerellaceae, Bacteroidaceae, Prevotellaceae, Ruminococcaceae, and Veillonellaceae (Fig. 2D). Furthermore, at the genus level in the healthy group, common microorganisms included Parabacteroides, Prevotella 9, Bacteroides, Roseburia, Phascolarctobacterium and Bifidobacterium (Fig. 2E). In the NFPA group, Parabacteroides, Bacteroides, Roseburia, Prevotella 9, Lachnoclostridium, and Sutterella were enriched, while in the PPA group, Parabacteroides, Bacteroides, Prevotella 9, Roseburia, Enterobacter, and Phascolarctobacterium were enriched (Fig. 2E). Supplementary Table S1 provides detailed relative abundances of these bacteria. Additionally, heatmaps representing the top 35 bacteria from all groups at the family level (Supplementary Figure S1) and the genus level (Supplementary Figure S2) were also generated.

Fig. 2.

Diverse microbial profiles observed across groups at various taxonomic levels. At the phylum level, microbial community composition was examined across all samples (A), meanwhile, the dominant bacteria (accounting more than 97% of all bacteria) in PPA, NFPA, and the HC group was also displayed in a circos plot (B). Top 10 microbial composition at the phylum level varied among PPA, NFPA, and healthy groups (C). Additionally, the distribution of microbial families (D) and genera (E) was explored among the three groups

The diversity of intestinal flora in patients with PA

Using sampling-based ASVs, alpha diversity indices were computed, revealing marked differences in microbial diversity among groups. Importantly, we found a significantly reduced gut microbial alpha diversity in the PPA group compared to both healthy subjects and the NFPA group, as indicated by various indices such as Simpson, Shannon, Chao1, ace, and observed species (Fig. 3A, B, C, E and F). Meanwhile, no significant differences were observed between the healthy group and the NFPA group (Fig. 3A-E). Beta diversity analysis shed light on structural variations in microbial communities, especially in PCoA where the gut microbiota of the PPA group clustered distinctly from that of healthy subjects (PCoA1: 9.1%, HC vs. PPA, p < 0.01, Fig. 3G). Utilizing Bray–Curtis distances, the NMDS analysis highlighted significant variations in microbial genera among the groups, demonstrating a low STRESS value of 0.1074 (Fig. 3H). A statistically significant difference between groups (R = 0.046, p = 0.018) was determined by ANOSIM analysis. However, the small R-value indicates that the proportion of variation explained by group differences is limited, suggesting relatively weak separation. Nevertheless, this result supports the existence of group-specific microbial community structures when considered alongside the clearer clustering trends observed in PCoA and NMDS analyses (Fig. 3I).

Fig. 3.

The microbial diversity of gut microbiota across three distinct groups. The α-diversity indexes, including Simpson, Shannon, Chao1, PD whole tree, ace, observed species indices (A-F) were vividly presented through boxplots. The Principal Coordinate Analysis (PCoA) visually mapped the microbial landscape, utilizing unweighted UniFrac distances (G). The axes on the plot symbolize the primary and secondary principal coordinates, capturing the highest proportion of variability within the microbial communities. Each spot on the plot represented a sample, with colors delineating group distinctions. Non-metric Multidimensional Scaling (NMDS) further delineated microbial composition variations through Bray-Curtis distances (H). Samples sharing high similarity in gut microbiome compositions are depicted as closely positioned spots on the plot. The Analysis of Similarities (ANOSIM) conducted on healthy individuals, PPA patients, and NFPA patients highlights substantially greater distinctions in inter-group comparisons compared to intra-group comparisons (I). * 0.01 < p < 0.05; ** p < 0.01;

Underlying biomarkers for patients with PPA and NFPA and healthy subjects

Despite significant changes were revealed in the intestinal flora of patients with PPA and NFPA, the predominant taxon could not be determined by the discriminant analysis alone. Hence, we conducted a comprehensive analysis at all taxonomic levels of gut microbiota to identify distinct taxa within each group by performing LEfSe analysis. 22 microbial taxa were picked out associated with PA, among these, 4 were prevalent in NFPA, 10 in PA, and 8 in healthy controls (Fig. 4A-B). Pituitary adenomas, particularly PPA, exhibited the presence of pathogenic bacteria (such as Enterobacteriaceae and Desulfovibrio), while a plethora of SCFA-producing bacteria was identified in the healthy control group (such as Bifidobacterium and Lachnoclostridium).

Fig. 4.

The cladogram reveals the dominant microbial taxa of healthy controls (HC) and patients with PPA and NFPA (A). LDA scores delineate various taxa in the gut microbiota of HC and patients with PPA and NFPA (B). Radiations, stretching from inner to outer regions, encompass a spectrum of taxonomic levels that includes phylum through genus. Nodes at each level represent species classifications. Yellow nodes denote no distinction among groups, while differently colored nodes indicate enrichment in the respective group. Only taxa with LDA values exceeding 2 are displayed

A microbial biomarker panel for discriminating patients with PPA and NFPA from HCs

In this study, various PA-associated and HC-associated gut microbes were revealed. Bifidobacterium, Dorea, Blautia, Morganella, Desulfovibrio, and Enterobacter were identified as the most prevalent and consistently detected genera among PA and HC individuals. Notably, substantial inter-individual variations were observed in the abundance of these genera, as shown in Fig. 5A-F. Further analysis evaluated these genera, both individually and collectively, as potential biomarkers for PA diagnosis. In order to mitigate potential overfitting, we performed an internal validation by randomly splitting the dataset into a training set (70%) and an independent test set (30%). The test set was used to assess the biomarker panel’s performance after it had been trained on the training set. Receiver operating characteristic (ROC) curve analysis of the combined panel on the test set yielded an AUC of 0.843 (Fig. 5G). This result points out a potential association between the identified microbial panel and the presence of PA, hence supporting its possible utility as a diagnostic tool.

Fig. 5 .

Various genera served as biomarkers for PA. The per-sample relative abundance of these biomarkers, including Bifidobacterium (A), Dorea (B), Blautia (C), Morganella (D), Desulfovibrio (E), and Enterobacter (F), was analyzed. The ROC curves (G) illustrate the discriminatory performance of each individual genus, as well as their combinations, in distinguishing between patients with PA and HC within both the training and independent test sets

Complicated associations among different bacterial communities

Intricate inter-phylum relationships were identified within the PPA cohort, encompassing 13 positive and 31 negative associations among the 10 identified phyla. Firmicutes, the dominant phylum, exhibited significant negative correlations with Bacteroidetes, Proteobacteria, and Fusobacteria. Interestingly, Fusobacteria displayed a contrasting pattern. While positively correlated with Bacteroidetes and Proteobacteria, it exhibited negative correlations with Actinobacteria and Verrucomicrobia (Figure 6).

Fig. 6.

The interrelation among various gut microbiota within the phylum level of PPA patients. Positive associations are represented in green, while negative associations are denoted in grey. The thickness of the line reflects the strength of the association (rho represents the Pearson correlation coefficient, with the value 1 indicates completely positively correlated and − 1 represents the otherwise)

Altered microbial phenotypic characteristics in patients with PPA

To reveal the mechanisms underlying behind these associations, Bugbase software was utilized to analyze the phenotypic traits of the microbes. A significant decrease in the abundance of gram-positive bacteria (p = 0.00022) and anaerobic bacteria (p = 0.00041) was observed when comparing the PPA group to the healthy group. Conversely, the abundance of gram-negative bacteria (p = 0.00009), facultative anaerobic bacteria (p = 0.00017), opportunistic pathogens (p = 0.00013), and the level of oxidative stress tolerance (p = 0.00014) displayed an opposite trend. Interestingly, no significant difference was detected in biofilm formation (p = 0.24) and aerobic bacteria (p = 0.32) (Fig. 7A–I).

Fig. 7 .

The intestinal microbial phenotypic characteristics in PA patients. The Bugbase software predicted traits were displayed, including Anaerobic (A), Aerobic (B), Facultatively Anaerobic (C), Gram-negative (D), Gram-positive (E), potentially pathogenic (F), forms biofilms (G), contains mobile elements (H) and stress tolerant (I)

Correlation analysis among various gut microbial genera and clinical indices

To elucidate the intricate relationships between gut microbiota and clinical manifestations of PPA, a comprehensive analysis of differential taxa at the genus level was conducted. Pearson correlation tests were conducted to assess potential associations between these taxa and diverse clinical indices, encompassing Knosp grade (a marker of adenoma invasiveness), tumor size, and a panel of lipid parameters (apolipoprotein A (Apoa), apolipoprotein B (Apob), total cholesterol (TC), high-density lipoprotein (HDL), low-density lipoprotein (LDL), and triglycerides (TG)). The intricate network of relationships is represented in Fig. 8. Notably, several bacteria that were abundant in the control group, including the genus Bifidobacterium, demonstrated positive correlations with HDL and negative correlations with Apoa, Apob, TC, LDL, and TG. In contrast, Fusobacterium, Desulfovibrio, and Enterobacter, which were enriched in the PPA group, exhibited the opposite correlation pattern. Furthermore, it is unsurprising that genera prominently dominant in the HC group, such as Bifidobacterium and Lachnoclostridium, were strongly negatively associated with the genera abundant in the PPA group, such as Fusobacterium, Desulfovibrio, and Enterobacter.

Fig. 8.

Correlation analysis among various gut microbial genera and clinical indices. The heatmap visually represents the intercorrelation among 35 genera enriched in PPA following Pearson correlation analysis. The color intensity directly reflects the strength of correlation between various genera. Concurrently, we employed pearson correlation analysis to investigate the associations between these genera and the clinical indices of PPA patients. The lines represent correlation, with red indicating a positive relationship and green indicating the opposite. The width of the lines signifies the degree of correlation strength, with solid lines indicating p-values < 0.05, and dashed lines for other cases

Altered functional profile of gut microbiota in PA patients

The disturbance of the microbiota balance has the potential to initiate systemic metabolic dysfunction, thereby influencing the composition of intestinal flora [26]. In the investigation of metabolic and functional shifts within gut microbial communities, individual ASVs were aligned with the built-in reference database of PICRUSt. The analysis conducted using PICRUSt databases identified the top 20 KEGG pathways that exhibited significant differential abundance between the groups. Figures 9A-B illustrate an elevation in pathways linked to lipid metabolism (especially in the biosynthesis of steroid hormone and secondary bile acid), glycan biosynthesis and metabolism, and the digestive system, contrasting with a decline in pathways related to the endocrine system within the PPA group compared to the HC group. Meanwhile, the comparison between PPA and NFPA groups yielded a similar result, as revealed in Supplementary Figure S3.

Fig. 9.

We conducted an examination of the predicted functional profiles of gut microbiota in both HC and patients with PPA. This analysis unveiled significant alterations in pathways at KEGG level 2 (A) and the top 20 significantly modified pathways at KEGG level 3 (B) using PICRUSt analysis (p < 0.05). KEGG stands for Kyoto Encyclopedia of Genes and Genomes, while PICRUSt stands for Phylogenetic Investigation of Communities by Reconstruction of Unobserved States

Discussion

Current evidence suggests that the gut harbors a multitude of microbes that produce various small molecules impacting crucial pathways associated with immune balance and neurological homeostasis, particularly within the gut-brain axis [27]). In recent years, a growing body of evidence has emphasized the pivotal role of gut microbiota and its metabolic products in various neurological conditions and diverse brain tumors [13, 14]. Understanding the intricacies of gut microbiota characteristics and regulation is increasingly crucial for diagnosing and treating individuals with different brain tumors.

Meanwhile, recent scientific investigations have suggested a significant connection between the gut microbiota and PA. Research conducted by Jin H et al. [17] has put forth evidence indicating the potential utility of the gut microbiota (such as Clostridium symblosum, Clostridium innocuum) as a biomarker for individuals with invasive pituitary adenomas. Additionally, research conducted by Ding N and colleagues [16] revealed the involvement of the gut microbiota in shaping the immune milieu of pituitary adenomas that secrete growth hormone, operating through the complex gut-brain axis. However, the existing body of research focused on PPA remains insufficient. To address this gap, our present study recruited 72 patients diagnosed with PA, further categorized into PPA group (n = 31) and NFPA group (n = 41), along with matched healthy controls (n = 29). Importantly, we sought to explore the intricate relationships between the gut microbiome and PA (especially PPA). Comparative analyses revealed a significantly diminished microbial community in the PPA group, characterized by both reduced abundance and lower diversity compared to HC. Rigorous alpha and beta diversity indices revealed distinct microbial structures within the PPA group. Further investigation delved into functional differences between PPA and NFPA groups, focusing on microbial gene functions. This comprehensive study aimed to construct a potential biomarker panel for accurate discrimination of PA, particularly PPA, from healthy individuals, and to elucidate associations between clinical parameters and the gut microbiome in PPA patients. These findings demonstrate notable alterations in the gut microbiota of PA patients while emphasizing the potential significance of investigating host-microbe interactions, particularly in the context of PPA.

Distinct alterations in the composition of gut microbiota were observed in individuals with pituitary adenomas, consistent with the literature [15, 17]. The observed reductions in α-diversity indices indicate not only a decrease in the richness of microbial communities but also an uneven distribution, potentially contributing to the initiation or progression of chronic diseases [28]. Consistently, this study unveiled the link between gut microbiota and PA. In the NFPA group, the pathogen Morganella, a gram-negative bacterium belonging to Enterobacteriaceae, exhibited a disproportionately high abundance. This may lead to the suppression of the growth of SCFA-producing bacteria, contributing to immune and intestinal dysbiosis [29]. Moreover, the genus Blautia, known to influence the development of CNS diseases by modulating inflammasome signaling pathways, thereby altering brain function, was highlighted [30]. Two additional pathogens exhibiting enrichment in the NFPA group were identified, implying an imbalanced gut microbial environment in NFPA. Furthermore, eight bacteria associated with PPA, including tumorigenic pathogens such as Enterobacter and Desulfovibrio, were detected. The Enterobacter genus, a member of the Enterobacteriaceae family known for generating pathogen-associated molecular patterns, showed significant enrichment. Numerous studies have established its close association with the development of various tumors, including colorectal and liver tumors. Furthermore, Lin B et al. [15] reported a robust positive correlation between the Enterobacter genus and the GH/IGF-1 axis in GH-secreting PAs. Moreover, the Desulfovibrio genus has the potential to promote chronic neurological inflammation through the encoding of acid and lipid profiles and modulation of brain gene expression in CNS diseases [31]. Consequently, it was observed that PPA patients exhibited a more pronounced dysbiosis in gut microbiota, indicating the relevance of gut microbiome signatures to the subtype of PA. In addition to investigating potential pathogens, we also assessed less prevalent bacteria in PA patients, such as Bifidobacterium and Dorea, which reportedly yield anti-tumor effects, primarily through the production of butyric acid [32]. Butyrate, primarily produced by anaerobic microbes, exerts a crucial impact on intestinal physiology. Additionally, it could play a vital role in regulating the life cycle of intestinal cells, preventing pathogen invasion, hindering tumor advancement, and modulating immune responses in the CNS. Moreover, as previously reported, the genus Dorea has the potential to modulate tumor progression by enhancing the immune surveillance function of CD8 + T cells [33]. Furthermore, the Bifidobacterium genus, as part of the probiotic family, may serve as a pivotal factor in maintaining the kinetic equilibrium of metabolism, neurohormones, and immunity. Therefore, a decrease in these beneficial microbes could pose a risk factor for pituitary adenomas. Furthermore, a microbial biomarker panel for PA was constructed using six dominating microbes, achieving a satisfactory predictive performance with a high AUC value. These findings collectively suggest a reciprocal relationship between a perturbed central nervous system in PA and the dysregulated gut microenvironment, emphasizing the promising prospect of utilizing gut microbiota as a diagnostic tool.

While focusing on individual bacterial taxa within the PPA gut microbiome provides valuable insights, a holistic understanding necessitates exploring the intricate network of interactions between these microbial players. We hypothesize that the synergistic and antagonistic relationships within the microbiome, rather than the isolated presence or absence of specific bacteria, may play a crucial role in the development and progression of PPA. For instance, the lipopolysaccharide derived from gram-negative bacteria [34], primarily those belonging to the Proteobacteria phylum, has the potential to stimulate the growth of facultative anaerobic pathogens by serving as an energy source for them [35]. This observed correlation implies a disturbance in the concentrations of oxygen within the gut environment, offering a reasonable explanation for the heightened prevalence of Enterobacteriaceae and the decrease in the prevalence of strictly anaerobic gut microbes [35]. Subsequently, the increased abundance of Enterobacteriaceae may trigger neutrophil transepithelial migration and result in the reduction of bacteria that produce SCFAs, such as Bifidobacterium and Lachnoclostridium [36]. Meanwhile, the decline in SCFA-producing microbes could impede their capacity to restrict Enterobacteriaceae through reducing intestinal pH [35, 37]. Hence, a vicious circle may be established, as illustrated in Fig. 6. This intricate cascade highlights the complex dynamics within the gut microbiota and their potential repercussions on intestinal equilibrium in PPA.

Meanwhile, correlation analysis among various gut microbial genera and clinical indices were performed to provide a deeper understanding about the gut microbial role in PPA. In this study, the bacteria enriched in the PPA group, including Fusobacterium, Desulfovibrio, and Enterobacter, showed positive associations with serum lipid level, which aligns with findings from previous studies [38]. Many studies demonstrated that PRL has a dose-dependent effect [39] (i.e. low concentrations under physiological conditions have a protective effect, while the high concentrations under pathological conditions do not). In this respect, Park S. et al. [40] discovered that high concentrations of PRL could contribute to insulin resistance and disturbances in glucose and lipid metabolism. Despite these known facts, the precise mechanism remains unclear. We hypothesize that the gut microbiota may be a contributing factor. Notably, a substantial portion of the bacteria enriched in PPA may be gram-negative and capable of secreting lipopolysaccharides (LPS) [41]. Several studies have reported that LPS, through various mechanisms such as activating the TLR4 pathway and impairing colonic epithelial permeability [42], may contribute to elevated concentrations of serum lipids [25, 43]. In contrast, the SCFAs-producing bacteria enriched in the healthy control group, including Bifidobacterium and Lachnoclostridium, exhibit a negative association with lipid accumulation. Numerous studies have also suggested that SCFAs could counter lipid accumulation by enhancing gut barrier function, activating regulatory T cells, and increasing leptin release [25]. As a result, these findings suggest that alterations in gut microbiota in PPA may be associated with perturbations in lipid metabolism, which could be relevant to disease pathophysiology.

In the meantime, KEGG pathway analysis was also carried out to analyze potential functional implications of these microbial alterations. Interestingly, the most significantly altered pathway in pituitary adenomas (particularly in PPA) was the increased activity in lipid metabolism, particularly in the biosynthesis of steroid hormones and secondary bile acids. As documented previously, physiological levels of steroid hormones may play a crucial role in regulating key genes essential for organ development and function. Unfortunately, an imbalance in steroid hormones could facilitate the growth of hormone-responsive tumors [44]. Additionally, secondary bile acids have been identified as potential contributors to tumor advancement by activating the IL-6/JAK1/STAT3 pathway [45]. Furthermore, imbalances were observed in the pathways related to the endocrine and digestive systems of PPA. In conclusion, the intricate associations between PPA and the gut microbiota remain complex and not fully elucidated. Simultaneously, disruptions occurring at the microbiome-metabolome interface may play a role in triggering and advancing PPA. It is reasonable to propose that this phenomenon could be initiated by metabolites and toxins associated with tumor promotion, particularly those related to lipid metabolism and inflammation. Consequently, these specific pathways represent promising targets for both diagnosing and treating PPA.

To our knowledge, this study represents a thorough investigation of alterations in both microbial composition and predicted function in PPA patients. Of particular significance is the reduction observed in SCFA-producing bacteria (like Bifidobacterium and Lachnoclostridium), an elevation in gram-negative pathogenic bacteria, and disturbances in metabolic pathways (especially in lipid metabolism). The results of this preliminary investigation highlight the potential relevance of gut microbiota for both the diagnosis and management of PPA and underscore the necessity for further validation in subsequent studies. Additionally, the exploration of animal models holds promise for unveiling potential supplementary therapeutic approaches for PPA. This may involve interventions like the targeted modulation of gut microbiota through specific drugs or even the application of fecal microbial transplant.

This study acknowledges several inherent limitations that necessitate further investigation. Firstly, the cross-sectional design restricts the ability to establish definitive causality between the observed microbial dysbiosis and PPA. Secondly, the relatively small sample size, particularly in the PPA (n = 31) and HC (n = 29) groups, is another limitation of the current investigation, which may limit the statistical power and increase the risk of Type II errors, which could result in false-negative findings. Consequently, some associations may not have been determined, and the generalizability of our findings may be constrained. Thirdly, while the study identifies a compelling link between gut dysbiosis and PPA, it is crucial to acknowledge the possibility of reverse causation. Microbial alterations could be a consequence rather than a driver of tumor progression. Moreover, although ANOSIM showed statistical significance (R = 0.046, p = 0.018), the small R-value suggests weak overall group separation and should be interpreted with caution. Additionally, rather than using actual measurements, the functional pathway results were based on PICRUSt2 predictions. Future studies will need to validate these predictions using approaches such as metabolomics (e.g., SCFAs, bile acids, LPS, or steroid hormones). Furthermore, the inherent complexity of the gut microbiome presents challenges in eliminating all potential confounding variables and biases. The absence of animal experiments testing underlying mechanisms is another significant limitation. Future studies should address these limitations through more sophisticated experimental designs.

Conclusions

In our investigation, we observed a decreased diversity in the microbiota of individuals with PPA. Simultaneously, we identified a set of microbial markers for PPA, including Bifidobacterium, Dorea, Blautia, Morganella, Desulfovibrio, and Enterobacter, which indicated potential utility in distinguishing PPA patients from controls. Most bacteria enriched in PPA demonstrated strong associations with serum lipid levels. According to KEGG pathway analysis, metabolic pathways—particularly those pertaining to lipid metabolism—are the main cause of the altered gut microbiota in PPA. Consequently, we propose that exploring a novel therapeutic target based on microbiota manipulation holds promise for the treatment of PPA and warrants further investigation.

Abbreviations

GM

Gut Microbiota

PA

Pituitary Adenoma

PPA

Prolactin-Secreting Pituitary Adenoma

NFPA

Nonfunctioning Pituitary Adenomas

CNS

Central Nervous System

HC

Healthy Controls

KEGG

Kyoto Encyclopedia of Genes and Genomes

ASV

Amplicon Sequence Variant

ROC

receiver operating characteristic curve

AUC

area under curve

Apoa

Apolipoprotein A

Apob

Apolipoprotein B

PCoA

principal coordinate analysis

NMDS

non-metric multidimensional scaling

BMI

body mass index

CI

confidence interval

TC

Total Cholesterol

HDL

High-density Lipoprotein

LDL

Low-density Lipoprotein

TG

Triglycerides

LDA

Linear Discriminant Analysis

Authors’ contributions

HX J, W Z and AJ P conceived and designed this project. JY H, F Y, HE F, XL Z and T Q collected the samples and performed the data analysis. JHX, JH W, JY H, YL P, YW W, P Z and W Z conducted the experiments and wrote the manuscript. All authors read and approved this manuscript. All authors reviewed the results and approved the final version of the manuscript.

Funding

This work was supported by Postgraduate Practical Innovation Program of Jiangsu Province (SJCX23_2030), Program of Jiangsu Commission of Health (No. M2022068) and the Science and Technology Planning Project of Yangzhou City (No. YZ2023153).

Data availability

The datasets generated and/or analysed during the current study are available in the NCBI repository (accession number: PRJNA807001).

Declarations

Ethics approval and consent to participate

This study was conducted in accordance with the ethical principles outlined in the Declaration of Helsinki. Each participant provided informed consent, and we obtained ethical approval from the Ethics Committee of Affiliated Hospital of Yangzhou University (2024-YKL09-K06). Clinical trial number: not applicable (this study only involves observational data analysis and does not qualify as a clinical trial).

Consent for publication

Not Applicable.

Competing interests

The authors declare no competing interests.