MSACL 2023 Abstract

Introduction

Connections between gut microbial metabolism and disease have been of heightened interest in recent years, with multiple gut microbial metabolic by-products showing negative associations with patient outcomes. However, the downstream activity of the gut microbiota is not solely linked to deleterious consequences, with the production of short-chain fatty acids (SCFAs) shown to correlate with improved immunity, decreased inflammation, and elevated protection from disease characteristics. Multiple SCFAs (namely acetate, propionate, and butyrate) have demonstrated protective effects for in vitro and in vivo investigations of kidney disease. However, it is not currently known how basal circulating levels of SCFAs associate with disease outcomes in longitudinal human cohorts of chronic kidney disease (CKD).

Objectives

This project aimed to analyze serum SCFAs in a large cohort of CKD patients in order to investigate associations of SCFAs with medium-term clinical outcomes.

Methods

Serum samples were analyzed from CKD patients enrolled into the Renal Risk in Derby (RRID, UK) longitudinal cohort study and a total of 1,472 patient samples were included in the analyses (age 74 ± 9 years, 60% female). SCFAs including four straight chain [acetate (C2), propionate (C3), butyrate (C4), valerate (C5)] and three branched chain molecules [isobutyrate (isoC4), 2-methyl butyrate (metC4), isovalerate (isoC5)] were quantitated using a fully validated gas chromatography-mass spectrometry assay (previously published in JMSACL). Each SCFA was investigated by Cox’s proportional hazards regression (COX) for univariable association with clinical outcomes of mortality (death) and a composite of mortality or cardiovascular event (death/CVE). Patients were followed up for a median of 1,994 days (range 4-2408 days), and a total of 218 (14.8%, death) and 477 (32.4%, death/CVE) events were recorded for each outcome. Significant univariable associations of SCFAs with clinical outcomes were further assessed by Kaplan-Meier survival analyses (KM) for group differences of event occurrence at standardized stratification levels (e.g. median), and via a series of multivariable COX models adjusted for increasing quantities of relevant CKD risk factor variables. SCFAs that remained independently associated with outcomes in a multivariable model were assessed for their additive prognostic nature to a base risk score model (age, sex, estimated glomerular filtration rate (eGFR), and plasma albumin:creatinine) by net reclassification improvement (NRI) investigations. Furthermore, prognostic cut off values were identified for associated SCFAs by calculating the optimal Youden index value derived from the receiver operator characteristic (ROC) curve for each relevant outcome parameter. KM was performed using the calculated cut off values and patient groupings assessed for differences. Cut off values were also investigated as a discrete variable for improvement in risk predication by NRI to compare the base model and the base model plus optimized SCFA cut off. All data were analyzed using R statistics or IBM SPSS Statistics.

Results

SCFAs were detected in all patient samples with levels at or above the limit of quantitation for C2 (99.6%), C3 (100%), C4 (100%), C5 (60.5%), isoC4 (99.8%), metC4 (82.7%), and isoC5 (96.8%). Butyrate levels were decreased in patients who died during the follow up period and when the composite event was recorded (death/CVE). Univariable investigations with outcomes showed that decreased butyrate levels were the only SCFA to show association with death and death/CVE. Butyrate retained independent predictive characteristics for both outcome variables when the predictions were adjusted in a basic (5 variables), intermediate (9 variables), and expanded (15 variables) model. No differences in KM analyses were observed at any standardized cut off point. When butyrate was added as a continuous variable to the base model, no improvements in predictive capability were seen for NRI investigations. To assess the prognostic quality of butyrate as a dichotomized variable, the optimal Youden index was calculated to identify prognostic cut off values. When patients were categorized by these cut off values, KM analyses demonstrated a difference in survival statistics between patient groupings. Furthermore, butyrate dichotomized by optimal Youden index cut off groupings was able to improve prognostic capacity on top of the base model for death and death/CVE for NRI investigations.

Conclusion

Elevated levels of serum butyrate were associated with improved outcomes in CKD patients from a single-center longitudinal study. Butyrate was independently prognostic for outcomes relating to death and death/CVE after adjustment for a series of common CKD risk factor variables. The identification of prognostic cut off values for serum butyrate via Youden index values provided additive value to improve outcome risk prediction when added to a base prediction model including commonly used CKD risk factor variables. Overall, serum butyrate was demonstrated as a valid prognostic indicator in CKD and associations indicated that increasing serum butyrate concentrations showed a protective association with adverse clinical outcomes. Future research in multi-center, multi-national cohorts should be performed, with the potential to investigate if an intervention to increase circulating butyrate levels improves outcomes in CKD patients.