A Predictive Diagnostic Model for IgA Vasculitis Based on a Metabolomic Approach

Alexandre Boissais¹, Hélène Blasco², Patrick Emond³, Antoine Lefevre², Adrien Bigot⁴, denis Mulleman⁵, François Maillot⁴ and Alexandra Audemard-Verger⁴, ¹Medical University of Tours, Tours, France, ²Biochemistry and Molecular Biology Department, Tours, France, ³In vitro Nuclear Medicine Department, Tours, France, ⁴Department of Internal Medicine, Tours, France, ⁵Rheumatology Department, Tours, France

Meeting: ACR Convergence 2021

Date of first publication: October 22, 2021

Keywords: Biomarkers, Late-Breaking 2021, metabolomics, spondyloarthritis, Vasculitis

Session Information

Title: Late-Breaking Posters (L01 - L15)

Session Type: Poster Session D

Background/Purpose: IgA vasculitis is a rare systemic disease that is life-threatening mainly due to digestive or renal involvement. To date, there is no reliable diagnostic marker.

The objective was to explore the serum metabolome of patients with IgA vasculitis to identify potential diagnostic biomarkers.

Methods: We performed a cross-sectional study comparing the serum metabolome of patients with IgA vasculitis and controls with spondyloarthritis. Serum analysis was performed by high-performance liquid chromatography-mass spectrometry. Univariate and multivariate analyses were performed.

Results: We compared the metabolome of 55 patients with IgA vasculitis versus 77 controls with spondyloarthritis; matched for age and gender.

The median age of IgA vasculitis patients was 53 years old. Two-thirds of the patients were female (n=32). At diagnosis of vasculitis, 100% of patients had skin involvement and there was renal involvement in 69% (n= 38), joint involvement in 56% (n= 31), and digestive involvement in 42% (n= 23).

The multivariate metabolomic model resulting from patient serum analyses discriminated IgA vasculitis and spondyloarthritis with >90% accuracy; significantly on permutation tests (p< 0.01). Validation on the test sets revealed excellent predictive values on independent cohorts: sensitivity 98%; specificity 98%, positive predictive value 97% and negative predictive value 98%. These models identified 4 discriminative metabolites: 1-methyladenosine, L-glutamic acid, serotonin, and thymidine.

Conclusion: This study demonstrated an excellent diagnostic predictive model for IgA vasculitis based on the serum metabolome. These results need to be confirmed in a larger IgA vasculitis cohort, as well as with other control populations.

– Score scatter plot based on the PLS-DA models to explain the diagnosis (green for IgA vasculitis and red for Rheumatism inflammatory.
– Rank of the different metabolites (the top 15) identified by the PLS-DA according to the VIP score on the x-axis. Coloured boxes on the right indicate the relative concentrations of the corresponding metabolite in each group.

Prediction of IgA vasculitis in an independent cohort from plasma metabolome profile of the patients from the training set. ROC curves, obtained after PLS-DA models, allowed to compare the diagnosis prediction to the observed diagnosis.

A. Boissais, None; H. Blasco, None; P. Emond, None; A. Lefevre, None; A. Bigot, None; d. Mulleman, None; F. Maillot, None; A. Audemard-Verger, None.

To cite this abstract in AMA style:

Boissais A, Blasco H, Emond P, Lefevre A, Bigot A, Mulleman d, Maillot F, Audemard-Verger A. A Predictive Diagnostic Model for IgA Vasculitis Based on a Metabolomic Approach [abstract]. Arthritis Rheumatol. 2021; 73 (suppl 9). https://acrabstracts.org/abstract/a-predictive-diagnostic-model-for-iga-vasculitis-based-on-a-metabolomic-approach/. Accessed .

ACR Meeting Abstracts - https://acrabstracts.org/abstract/a-predictive-diagnostic-model-for-iga-vasculitis-based-on-a-metabolomic-approach/