AUTHORS
Samantha Atkinson, PhD, Katelyn E. Sadler Jr, PhD2, Jane Hankins, MD,MS3, Ashima Singh, PhD, MS4, Joshua Field, MD5, Mehri Bagherimohamadipour, PhD6, Dawn Retherford, BS7, Nicole Steiner, BS8, Janelle Highland, BS, MS9, Jessica J Rico, BS10, Michelle Brignac11, Hamda Khan12, Nita Salzman, MD, PhD13, Amanda M Brandow, DO, MS14
AFFILIATIONS
1 Medical College of Wisconsin, Milwaukee, Wisconsin, United States
2 University of Texas at Dallas, Dallas, United States
3 St. Jude Children's Research Hospital, Memphis, United States
4 Medical College of Wisconsin, Milwaukee, Wisconsin, United States
5 Medical College of Wisconsin, Milwaukee, Wisconsin, United States
6 Medical College of Wisconsin, Milwaukee, Wisconsin, United States
7 Medical College of Wisconsin, Milwaukee, Wisconsin, United States
8 Medical College of Wisconsin, Milwaukee, United States
9 Medical College of Wisconsin, Milwaukee, Wisconsin, United States
10 Medical College of Wisconsin, Milwaukee, Wisconsin, United States
11 St. Jude Children's Research Hospital, Memphis, United States
12 St. Jude Children's Research Hospital, Memphis, United States
13 Medical College of Wisconsin, Milwaukee, Wisconsin, United States
14 Medical College of Wisconsin, Milwaukee, Wisconsin, United States
ABSTRACT
Alterations in the intestinal microbiota (dysbiosis) contribute to inflammatory and immune-mediated disorders but remain understudied in sickle cell disease (SCD). Individuals with SCD have risk factors for dysbiosis (e.g., prophylactic antibiotics started in infancy, recurrent opioid exposure, intestinal ischemia from chronic vaso-occlusion), and exhibit reduced microbial diversity compared with healthy related and unrelated Black controls. However, data on differentially abundant taxa in SCD are limited. We leveraged 3 cohorts (SCD, related controls (R-controls), unrelated controls (U-controls)) to perform robust differential abundance analyses while controlling for covariates known to influence dysbiosis.
Individuals ≥2 months with SCD in baseline health (no acute visits for ≥2 weeks) and healthy Black individuals (related and unrelated to those in SCD cohort) were enrolled into a cross-sectional multi-site study. Any individual with acute illness requiring antibiotics within prior 4 weeks, those with SCD on chronic transfusions, and controls with a chronic inflammatory disorder were ineligible. 16S rDNA sequencing was completed on collected stool samples. Differential abundance analyses compared: 1) SCD vs R-controls and 2) SCD vs U-controls. Both analyses controlled for covariates that affect the intestinal microbiome (site of collection, age, sex, mode of birth, gestational age, infant feeding, relatedness, area deprivation index). Multivariate Association with Linear Models 2 (MaAsLin2), using a genus-level feature table filtered to retain taxa in at least 2 samples, was used to determine differentially abundant organisms for each group. A min_abundance of 10 and min_prevalence of 0.1 was used with a normalization method of “TSS.” The “relatedness” covariate was set as a random effect to account for confounding variables. All other variables from the ADONIS formula were set as fixed effects and the references for categorical variables were set as follows: gestational age-<33 weeks; mode of birth-vaginal; infant feeding-breast milk. The test was run twice; once with the reference for subject type as R-controls and once with U-controls. A q-value of <0.1 was significant and calculated using Benjamini-Hochberg tests. To identify taxa most relevant for future work, we determined the overlap of those that were significantly differentially abundant (q<0.1) in both independent analyses (SCD vs R-controls and SCD vs U-controls).
We enrolled 110 individuals with SCD, 53 related and 77 unrelated healthy Black individuals. Mean (SD) age (yrs) and sex (% female) were: SCD 16.5 (13.4), 46.4%; R-controls 28.7 (12.2), 75.5%; U-controls 27.3 (16.9), 71.4%. SCD genotype included HbSS 65.5%, HbSC 27.3%, HbSβ0thal 1.8%, HbSβ+thal 5.5%. Most differentially expressed taxa were significantly less abundant in SCD compared to both related and unrelated controls after controlling for covariates defined above and correcting for multiple comparisons (all q-value <0.1). Specifically, differential abundance analyses revealed: 1) SCD vs. R-controls: 20 taxa significantly depleted, 2 taxa significantly enriched and 2) SCD vs. U-controls: 16 taxa significantly depleted, 2 taxa significantly enriched. Of these differentially abundant taxa, 10 common taxa were significantly different between SCD and related and unrelated controls. Of these 10 taxa, 9 were significantly depleted (Oscillospira Negativibacillus, Faecalicoccus, Eubacterium nodatum group, Lachnospiraceae family (AC2044 group, UCG.010 group, FE2018 group, NK3A20 group, UCG.003 group) (all q-value <0.1, q-value range 0.07- 0.0004 for both comparisons) and 1 was significantly enriched (Cryptobacterium) in the SCD group (q-value 0.03 and 0.08 for both comparisons). Taxa common to both analyses support predominant depletion of 9 short-chain fatty acid (SCFA) producing bacteria in SCD, primarily from the Lachnospiraceae family.
We show SCD is characterized primarily by depletion of bacterial genera relative to healthy related and unrelated Black individuals. Two independent differential abundance analyses comparing SCD to related and unrelated controls show depletion of SCFA producing bacteria in SCD primarily from the Lachcnospiraceae family. SCD selective pressures likely reduce endogenous SCFA production which may influence disease biology. Future studies should assess clinical impact of these microbiome alterations and therapeutic potential of microbiome modulation in SCD.
DISCLOSURES
Amanda M Brandow, DO, MS: Pfizer - Consultancy (Includes expert testimony)