Award: Presidential Poster Award
Jessica R. Allegretti, MD, MPH1, Benjamin Mullish, MBBS2, Lotem Nativ, BA1, Jenna Marcus, BA1, Julian Marchesi, PhD3, Julie A.K. McDonald, PhD3, Alexandros Pechlivanis, PhD3, Kevin Kennedy, MS4, Georg Gerber, MD, PhD, MPH1, Lynn Bry, MD, PhD1
1Brigham and Women's Hospital, Boston, MA; 2Imperial College of London, London, Northern Ireland, United Kingdom; 3Imperial College of London, London, England, United Kingdom; 4Beth Israel Deaconess Hospital, Boston, MA
Introduction: Recurrent Clostridioides difficile infection (CDI) is a major public health problem. The ability of commensal gut microbiota to metabolize primary into secondary bile acids plays a role in protection against this infection. Current clinical prediction tools for CDI recurrence do not incorporate biomarkers predictive of protective microbiota functionalities. We investigated metabolomic predictors of C. difficile recurrence.
Methods: We conducted a prospective longitudinal study of patients experiencing a first CDI episode. Patients testing positive with either enzyme immunoassay (EIA) toxin or polymerase chain reaction (PCR), and being treated for CDI, were eligible for inclusion. Serial stool samples were collected at diagnosis through week-8 post-completion of anti-CDI therapy if no recurrence, or until the point of recurrence (defined as diarrhea with positive C. difficile EIA toxin stool test). Liquid chromatography-mass spectrometry was performed to profile fecal bile acids. The week 1 post-antibiotic time point was chosen to assess for potential predictors. We derived a univariate logistic regression model predicting recurrence and computed the AUC (c-statistic) on discriminatory ability. The Youden index was calculated as the value that maximizes sensitivity and specificity.
Results: 29 first episode CDI patients were enrolled. 10 patients recurred (34.5%) during 8-weeks of follow-up. Average time to recurrence was 1.9 weeks. Two clinical factors were assessed a-priori for recurrence: use of metronidazole vs vancomycin (OR: 3.56, 95% CI 0.61-20.81, p=0.15) and diagnosis with EIA vs PCR (OR 8.75, 95% CI 1.53-50.1, p=0.01). At week 1, significantly more of the secondary bile acids lithocholic acid (LCA) and isolithocholic acid were found in non-recurrers vs recurrers (relative abundance per mg of stool: 20395.0 vs 4542.5, p=0.037 and 4318.0 vs 1122.7, p=0.031 respectively) (Table 1). As these are co-linear, we proceeded with LCA alone. With Youden index 17025, sensitivity was 80% and specificity was 57.9%, and the overall AUC resulted in a c-statistic of .753 (Figure 1A). However, when clinical risk factors were combined with LCA, sensitivity became 90% and specificity was 79%, with c-statistic of .91 (Figure 1B).
Discussion: The secondary bile acid, LCA, a known inhibitor of C. difficile’s vegetative growth, predicted eventual recurrence of CDI with moderate accuracy; however, the model improved significantly with the addition of simple clinical variables.
Citation: Jessica R. Allegretti, MD, MPH; Benjamin Mullish, MBBS; Lotem Nativ, BA; Jenna Marcus, BA; Julian Marchesi, PhD; Julie A.K. McDonald, PhD; Alexandros Pechlivanis, PhD; Kevin Kennedy, MS; Georg Gerber, MD, PhD, MPH; Lynn Bry, MD, PhD. P1035 - EVALUATING DYNAMICS OF BILE ACID METABOLISM TO PREDICT RECURRENCE OF CLOSTRIDIOIDES DIFFICILE INFECTION. Program No. P1035. ACG 2019 Annual Scientific Meeting Abstracts. San Antonio, Texas: American College of Gastroenterology.