Detection of fungal pathogens by a histomolecular approach using targeted-massive parallel sequencing on formalin-fixed tissues: a retrospective study.

Abstract

Objectives: Since fungal infections (FI) are frequently encountered by pathologists, it is crucial to improve fungal diagnosis on formalin-fixed paraffin-embedded tissues (FT). We aimed to investigate if a histomolecular approach using targeted-massive parallel sequencing (MPS) could help detect and identify fungi on FT, when no mycological diagnosis is available on fresh tissue.

Methods: Forty-nine FT from 48 patients with histopathological FI diagnosis but without mycological identification were retrospectively included. Histopathology defined the fungal pattern and the tissue injuries. Panfungal PCRs were performed using ITS-3/ITS-4 and MITS-2A/MITS-2B primers. Amplicons were sequenced using Sanger sequencing and targeted-MPS. Probabilities of fungal identification for both sequencing techniques and both primers were compared.

Results: The median age was 57 years (Q1: 47; Q3: 64). Fungal cultures were performed in 22/49 (44.9%) samples but failed to identify the pathogenic fungi. Fungal identification by Sanger sequencing was successful in 17/49 (34.7%; [0.214-0.480]) FT; the probability of fungal identification was 32.7% (16/49; [0.195-0.458]) for ITS-3/ITS-4 and 22.4% (11/49; [0.108-0.341]) for MITS-2A/MITS-2B. Targeted-MPS was successful in 35/49 (71.4%; [0.588-0.841]) samples; the probability of fungal identification was 59.2% (29/49; [0.454-0.729]) for ITS-3/ITS-4 primers and 61.2% (30/49; [0.476-0.749]) for MITS-2A/MITS-2B. The probability of fungal identification by targeted-MPS (35/49 [71.4%]) was significantly higher than that of Sanger sequencing (17/49 [34.7%]; p<0.0001). We assessed that this approach could have optimised care for 22/48 (45.8%) patients.

Conclusions: Integrated histomolecular diagnosis using targeted-MPS could secure diagnosis and help clinicians prescribe the most appropriate antifungal therapy in the absence of mycological identification.