Dysglycemia and the airway microbiome in cystic fibrosis

Abstract

Background Cystic fibrosis-related diabetes (CFRD) is one of the most common non-pulmonary complications in people living with cystic fibrosis (pwCF), seen in up to 50% of adults. Even when correcting for severity of CFTR mutations, those with CFRD have more pulmonary exacerbations, lower lung function, and increased mortality than those with normal glucose tolerance (NGT). Methods Expectorated sputum samples were collected from 63 pwCF during routine outpatient visits (29 with CFRD, 12 with IGT and 22 with NGT). Oral glucose tolerance test results, A1c levels, and pulmonary function tests closest to the time of sputum collection were obtained from the medical record. Samples underwent metagenomics sequencing and raw reads were processed through the bioBakery workflow for taxonomic profiling at the species level as well as predicted functional profiling and antibiotic resistance profiling. Viral profiling was performed with Marker-MAGu. Differences in alpha diversity, beta diversity, and differential abundance were assessed. Microbiome and phage signatures of CFRD were generated using sparse partial least squares models which were subsequently used as a primary predictor of lung function using multivariate linear regression. Results In linear models, CFRD status compared to NGT was associated with a lower alpha diversity (reciprocal Simpson −1.98 [−3.80,-0.16], p = 0.033) and differences in microbial community composition (Bray Curtis dissimilarity PERMANOVA R2 0.17, p = 0.011). Pseudomonas aeruginosa and Streptococcus gordonii had higher relative abundance in CRFD vs NGT participants (2.43 [0.027, 4.82], unadjusted p = 0.056 and 1.11 [0.58, 1.64] unadjusted p= < .001 respectively). There were global differences between CFRD vs NGT in both functional pathways and antibiotic resistance genes. In multivariate models adjusting for age, sex, antibiotic use, and modulator therapies, virome but not microbiome signatures of CFRD were associated with lower FEV1 percent predicted (−6.4 [95% CI −10.2, −2.6]%, p = 0.001 for each 10% increase in virome score). Conclusion Differences in the airway microbiome in those with dysglycemia in CF are associated with poorer lung function.