Microbiome data enhances predictive models of lung function in people with cystic fibrosis

Abstract

The polymicrobial context of chronic infection has received increasing attention due to widespread use of microbiome sequencing technology. However, clinical microbiology analysis of infection samples in hospitals continues to focus only on established human pathogens. This disconnect between diverse ‘infection microbiomes’ and limited clinical microbiology profiling leaves open the possibility that important risk markers are being unexploited during infection management. To address this disconnect, we focus on lung infections in people with Cystic Fibrosis (CF). A cohort of CF patients (N=77) were recruited for this study. We collected health information (age, BMI, lung function) and clinical microbiology records for each patient. We also collected sputum samples during a period of clinical stability, and determined lung microbiome compositions through 16S rDNA sequencing. We use a regularized linear regression algorithm (ElasticNet) to select informative features to predict lung function. We find that models including whole microbiome quantitation outperform models trained on pathogen quantitation alone, with or without the inclusion of patient metadata. Our most predictive models retain key pathogens as negative predictors (Pseudomonas, Achromobacter) along with established correlates of CF disease state (age, BMI, CF related diabetes). In addition, our models select specific non-pathogen taxa (Fusobacterium, Rothia) as positive predictors of lung health. Our analysis does not address causality, leaving open whether these non-pathogen taxa are playing an active role in promoting lung health (e.g. by suppressing pathogens), or are simply informative biomarkers of patient health (orthogonal to age, BMI, etc). Our results support a reconsideration of clinical microbiology pipelines to ensure the provision of the most informative data to guide clinical practice.

Rothia mucilaginosais an anti-inflammatory bacterium in the respiratory tract of patients with chronic lung disease

“…and sputum inflammatory markers, such as neutrophil elastase, IL-8 and IL-1β in CF patients [66,67]. A recent report found the relative abundance of the genus Rothia in expectorated sputum to be a positive predictor of lung function in patients with CF [68]. Not only were we able to confirm a negative correlation between Rothia spp.…”

Single nucleotide variants in Pseudomonas aeruginosa populations from sputum correlate with baseline lung function and predict disease progression in individuals with cystic fibrosis

“…2 ), indicating that Pa abundance can be considered as a proxy for lung microbiome diversity in our dataset. However, Shannon and Simpson diversity indices were not selected as predictive features in our model, consistent with a previous work (Acosta et al 2018; Zhao, Hao et al 2020). This suggests that, even if low microbiome diversity indices are associated with CF disease progression, the low diversity is likely driven by the dominance of key pathogens such as Pa .…”

“…2 ). Multiple studies have shown an association between poor lung function and low BMI (Snell, Bennetts et al 1998, Cystic Fibrosis Foundation 2006, Kumru, Emiralioğlu et al 2018), high abundance of Pa (Cox, Allgaier et al 2010) and age (Zhao, Hao et al 2020, Cox, Allgaier et al 2010). As expected, Pa relative abundance also showed a strong negative correlation with Shannon and Simpson microbiome diversity indices ( Supplementary fig.…”

“…However, microbiome diversity may have limited predictive value as there is high interpersonal variability in lung microbiomes (Cuthbertson, Walker et al 2020), and a large number of adult CF individuals have microbiomes dominated by pathogens such as Pa . Zhao, Hao et al (2020) also recently showed that microbiome composition data does not improve machine learning (ML) prediction performance compared with using only clinical factors. Our results suggest that genetic diversity within key pathogens like Pa could complement or even supersede microbiome community diversity for predicting clinical outcomes in specific patient subsets.…”

“…Among the set of clinical factors studied, BMI, Pa abundance, and age were identified as important predictors of both baseline lung function and lung function decline. These are all known risk factors for CF disease severity and progression (Acosta et al 2018, Snell, Bennetts et al 1998, Kumru, Emiralioğlu et al 2018, Cox, Allgaier et al 2010, Zhao, Hao et al 2020). By including these features in our prediction models, we noted a moderate increase across all the measured metrics relative to using only AmpliSeq data.…”