Validation of a Salivary miRNA Signature of Endometriosis - Interim Data
Sofiane Bendifallah,Yohann Dabi,Stéphane Suisse,Léa Delbos,Andrew Spiers,Mathieu Poilblanc,Francois Golfier,Ludmila Jornea,Delphine Bouteiller,Hervé Fernandez,Alexandra Madar,Erick Petit,Frédérique Perotte,Raffaèle Fauvet,Michael Benjoar,Cherif Akladios,Vincent Lavoué,Thomas Darnaud,Benjamin Merlot,Horace Roman,Cyril Touboul,Philippe Descamps
DOI: https://doi.org/10.1056/EVIDoa2200282
Abstract:BACKGROUND: The discovery of a saliva-based micro–ribonucleic acid (miRNA) signature for endometriosis in 2022 opened up new perspectives for early and noninvasive diagnosis of the disease. The 109-miRNA saliva signature is the product of miRNA biomarkers and artificial intelligence (AI) modeling. We designed a multicenter study to provide external validation of its diagnostic accuracy. We present here an interim analysis. METHODS: The first 200 patients included in the multicenter prospective ENDOmiRNA Saliva Test study (NCT05244668) were included for interim analysis. The study population comprised women from 18 to 43 years of age with a formal diagnosis of endometriosis or with suspected endometriosis. Epidemiologic, clinical, and saliva sequencing data were collected between November 2021 and March 2022. Genomewide miRNA expression profiling by small RNA sequencing using next-generation sequencing (NGS) was performed, and a random forest algorithm was used to assess the diagnostic accuracy. RESULTS: In this interim analysis of the external validation cohort, with a population prevalence of 79.5%, the 109-miRNA saliva diagnostic signature for endometriosis had a sensitivity of 96.2% (95% confidence interval [CI], 93.7 to 97.3%), specificity of 95.1% (95% CI, 85.2 to 99.1%), positive predictive value of 95.1% (95% CI, 85.2 to 99.1%), negative predictive value of 86.7% (95% CI, 77.6 to 90.3%), positive likelihood ratio of 19.7 (95% CI, 6.3 to 108.8), negative likelihood ratio of 0.04 (95% CI, 0.03 to 0.07), and area under the receiver operating characteristic curve of 0.96 (95% CI, 0.92 to 0.98). CONCLUSIONS: The use of NGS and AI in the sequencing and analysis of miRNA provided a saliva-based miRNA signature for endometriosis. Our interim analysis of a prospective multicenter external validation study provides support for its ongoing investigation as a diagnostic tool. (Funded by Ziwig and the Conseil Régional d’Ile de France [Grant EX024087]; ClinicalTrials.gov number, NCT05244668.)