Characteristics of Gut Microbiota in Patients with GH-Secreting Pituitary Adenoma

Abstract

Prior study has demonstrated that gut microbiota at the genus level is significantly altered in patients with growth hormone (GH)-secreting pituitary adenoma (GHPA). Yet, no studies exist describing the state of gut microbiota at species level in GHPA. We performed a study using 16S rRNA amplicon sequencing in a cohort of patients with GH-secreting pituitary adenoma (GHPA, n = 28) and healthy controls (n = 67). Among them, 9 patients and 10 healthy controls were randomly chosen and enrolled in metagenomics shotgun sequencing, generating 280,426,512 reads after aligning to NCBI GenBank DataBase to acquire taxa information at the species level. Weighted UniFrac analysis revealed that microbial diversity was notably decreased in patients with GHPA, consistent with a previous study. With 16S rRNA sequencing, after correction for false-discovery rate (FDR), rank-sum test at the genus level revealed that the relative abundance of Oscillibacter and Enterobacter was remarkably increased in patients and Blautia and Romboutsia genera predominated in the controls, augmented by additional LEfSe (linear discriminant analysis effect size) analysis. As for further comparison at the species level with metagenomics sequencing, rank-sum test together with LEfSe analysis confirmed the enrichment of Alistipes shahii and Odoribacter splanchnicus in the patient group. Notably, LEfSe analysis with metagenomics also demonstrated that Enterobacter sp. DC1 and Enterobacter sp. 940 PEND, derived from Enterobacter, were both significantly enriched in patients. Functional analysis showed that amino acid metabolism pathway was remarkably enriched in GHPA, while carbohydrate metabolism pathway was notably enriched in controls. Further, significant positive correlations were observed between Enterobacter and baseline insulin-like growth factor 1 (IGF-1), indicating that Enterobacter may be strongly associated with GH/IGF-1 axis in GHPA. Our data extend our insight into the GHPA microbiome, which may shed further light on GHPA pathogenesis and facilitate the exploration of novel therapeutic targets based on microbiota manipulation. IMPORTANCE Dysbiosis of gut microbiota is associated not only with intestinal disorders but also with numerous extraintestinal diseases. Growth hormone-secreting pituitary adenoma (GHPA) is an insidious disease with persistent hypersecretion of GH and IGF-1, causing increased morbidity and mortality. Researches have reported that the GH/IGF-1 axis exerts its own influence on the intestinal microflora. Here, the results showed that compared with healthy controls, GHPA patients not only decreased the alpha diversity of the intestinal flora but also significantly changed their beta diversity. Further, metagenomics shotgun sequencing in the present study exhibited that Enterobacter sp. DC1 and Enterobacter sp. 940 PEND were enriched in patients. Also, we were pleasantly surprised to find that the Enterobacter genus was strongly positively correlated with baseline IGF-1 levels. Collectively, our work provides the first glimpse of the dysbiosis of the gut microbiota at species level, providing a better understanding of the pathophysiological process of GHPA.

Keywords: clinical therapeutics; metabolism; metagenomics.

FIG 1

Microbial diversity was notably decreased in patients with GH-secreting pituitary adenoma. (A) A Venn diagram displaying the overlaps between groups showed that 516 of the total richness of 742 OTUs were shared among the two groups, while 42 and 184 OTUs were found exclusively in GH group and HC group, respectively. Compared with that in the controls, fecal microbial diversity, as estimated by the Shannon index (B) and Simpson index (C), was significantly decreased in patients with GH (P = 0.036 and 0.013). Beta diversity was calculated using weighted UniFrac and further analyzed by MRPP (D) and Adonis (E), both obtaining P values equal to 0.002. Further, the former and latter tests were presented via NMDS (D) and PCoA (E), respectively. Cooperatively, our results demonstrated a significant clustering tendency among all samples. GH, growth hormone-secreting pituitary adenoma; HC, healthy control; OTUs, operational taxonomy units; MRPP, multi response permutation procedure; NMDS, nonmetric multidimensional scaling; PCoA, principal coordinates analysis.

FIG 2

Taxonomic changes of gut microbiota at genus level in patients with GH-secreting pituitary adenoma. (A) A histogram displayed the relative abundance of the 20 main genera in all samples, simultaneously showing the clustering relationships among all samples. (B) A barplot presenting the phylogenetic profiles at genus level of the two groups, with only Oscillibacter genus and Enterobacter genus increased in GH group while Blautia genus and Romboutsia genus ranked top 2 among the decreased genera. (C) Two violin plots demonstrate significantly increased abundance of Oscillibacter genus and Enterobacter genus in GH group. (D) Two violin plots show significantly decreased abundance of Blautia genus and Romboutsia genus in GH group. (E) A histogram exhibits genera with significant effects on the division between groups, assessed by LEfSe analysis and corresponding influence represented as LDA score, further confirming the prominent roles of Oscillibacter, Enterobacter, Blautia, and Romboutsia genera. GH, growth hormone-secreting pituitary adenoma; HC, healthy control; g_ in figure c and d is short for genus, dotted line and dashed line represent quartiles and median, respectively; LDA, linear discriminant analysis.

FIG 3

Taxonomic differences at species level between patients with GH-secreting pituitary adenoma and healthy controls. (A) A histogram shows the relative abundance of the 20 main species in all samples, with a clustering tree of the total samples listed on the left side and the specific species exhibited on the right side. (B) The histogram demonstrates species with absolute LDA score exceeding two, identifying species with significant influence on sample division. (C) Boxplots illustrate differential phyla at species level in patients and healthy individuals detected by rank-sum test, with two different colors representing different sets of samples, while the short bar and the plus sign indicate the median and the mean, respectively. Only the 20 most abundant species in each group are shown for clarity. GH, growth hormone-secreting pituitary adenoma; HC, healthy control; LDA, linear discriminant analysis.

FIG 4

Functional characterization of microbiome in GH patients. (A) Boxplots illustrate differential KOs between GH group and HC group detected by rank-sum test, with two different colors representing different sets of samples while the short bar and the plus sign indicating the median and the mean, respectively. Only the 20 most abundant KOs in each group are shown for clarity. (B) The histogram shows KOs with absolute LDA score exceeding two, identifying KOs with significant influence on sample division. (C) The barplot displays the percentage of the overall KOs occupied by each pathway in both groups, respectively, with amino acid metabolism pathway notably enriched in GHPA while carbohydrate metabolism pathway remarkably enriched in controls. GH, growth hormone-secreting pituitary adenoma; HC, healthy control; LDA, linear discriminant analysis; KO, KEGG ortholog.

FIG 5

Correlation analysis between the differential taxa and clinical indices of patients with GH-secreting pituitary adenoma. (A) The heatmap delineates the correlation between significantly varied genera detected by LEfSe analysis and clinical indices of participants enrolled in 16S rRNA sequencing after spearman correlation analysis, among which prominent positive correlation was observed between Enterobacter genus and pre-op IGF-1. (B) Similarly, spearman correlation analysis was conducted between remarkably differed species observed in LEfSe analysis and clinical parameters of patients registered in metagenomics sequencing. The depth of color directly shows the degree of correlation between taxon and environmental factors. At the same time, correlation significance test was carried out, with + and * symbolizing P < 0.05 and P < 0.01, respectively. (C) The boxplots demonstrate that significant differences existed in pre-op IGF-1, pre-op IGF-1 index, change in IGF-1 index, ratio of change in IGF-1 index, and ratio of change in nadir GH between low-abundance group and high-abundance group, divided according to relative abundance of Enterobacter genus in GH group. Detailed definitions of clinical indices involved in correlation analysis are illustrated in Materials and Methods.

FIG 6

Gut microbes discriminate patients with GH-secreting pituitary adenoma from healthy controls. The scatter diagrams on the left side display the respective top 6 genera with largest contributions to the division of the two sample sets, corresponding to mean decrease accuracy (A) and mean decrease Gini (B) separately. (C) The ROC curve of the random forest model constructed on the basis of ordered differential genera, in which the abscissa axis is 1 − specificity while the ordinate is sensitivity. The point marked in the figure is the closest one to the top left corner, with a specificity and sensitivity equal to 1.000 and 0.875, respectively. AUC, the area under the curve, has reached 0.981 in our model, and the larger the AUC, the better the prediction effect of the model. g_ in panels A and B is an abbreviation of genus.