EDIT-B consortium study: new validation for a blood test for the diagnosis of bipolar disorder

Introduction

Differentiating bipolar disorder (BD) from major depressive disorder (MDD) during depressive episodes remains challenging due to underreporting of prior hypomanic or manic episodes. As BD often begins with depression, diagnostic approaches relying on symptoms and interviews lack precision, leading to the misdiagnosis of BD patients as MDD. The average diagnostic delay (about 10 years) has serious clinical consequences, including inappropriate treatment, risk of manic switching and increased suicide rates. A-to-I RNA editing is a highly dynamic process implicated in regulatory mechanisms relevant to psychiatry. Dysregulated editing patterns have been reported in BD and other neuropsychiatric disorders, implicating this process in mood regulation and neuroinflammation. Recent evidence has highlighted the potential of a blood-based RNA editing biomarker assay, comprising eight RNA sequences and an algorithm, to discriminate between BD and MDD with high sensitivity and specificity [1, 2].

Aims

The objective of this study is to validate and confirm the diagnostic performance of this test in a novel independent multicenter European cohort.

Methods

This trial (NCT05603819) is a double-blind multicenter cross-sectional study enrolling adults diagnosed with BD or MDD with a current major depressive episode across four clinical sites: two in Barcelona (Spain), one in Paris (France), and one in Copenhagen (Denmark). Participants were enrolled between June 2022 and September 2024, and written informed consent was obtained prior to enrollment. This study was conducted in accordance with the ethical principles outlined in the World Medical Association’s Declaration of Helsinki (2013). Blood samples were collected, RNA was extracted, and sequencing data were processed via a secure HDS-certified platform. Algorithm outputs were compared with physician diagnoses (MINI, MADRS, YMRS and EQ-5D-5L).

Results

The test was evaluated on an analytical dataset (N = 393) and its performance was assessed using the area under the ROC curve (AUC-ROC), in comparison with previously published results [1]. No statistically significant difference was observed between the two ROC curves (DeLong’s test, p = 0.33). The test demonstrated a sensitivity of 82.6% (95% CI: 75.7% - 88.2%), a specificity of 80.3% (95% CI: 74.6% - 85.1%), a negative predictive value (NPV) of 87.6% (95% CI: 82.5% - 91.7%), a positive predictive value (PPV) of 73.1% (95% CI: 65.9% - 79.6%), and an overall accuracy of 81.2% (95% CI: 77.0% - 84.9%). These results confirm that the test provides a robust discrimination between patients with MDD and those with BD.

Conclusions

The study confirms the diagnostic potential of RNA editing signatures combined with Artificial Intelligence for differentiating BD from MDD. The test’s performance was validated in a large, independent European multicenter cohort using the same biomarker panel and algorithm, demonstrating high discriminative performance (accuracy, sensitivity, and specificity >80%) and successfully replicating previous findings, thereby confirming its generalizability across diverse clinical settings.

References

Salvetat N, Checa-Robles FJ, Delacrétaz A, Cayzac C, Dubuc B, Vetter D, Dainat J, Lang J-P, Gamma F, Weissmann D (2024) AI algorithm combined with RNA editing-based blood biomarkers to discriminate bipolar from major depressive disorders in an external validation multicentric cohort. J Affect Disord 356:385–393. https://doi.org/10.1016/j.jad.2024.04.022
Salvetat N, Checa-Robles FJ, Patel V, Cayzac C, Dubuc B, Chimienti F, Abraham J-D, Dupré P, Vetter D, Méreuze S, Lang J-P, Kupfer DJ, Courtet P, Weissmann D (2022) A game changer for bipolar disorder diagnosis using RNA editing-based biomarkers. Transl Psychiatry 12:182. https://doi.org/10.1038/s41398-022-01938-6