Myocarditis and endomyocardial biopsy: achieving consensus diagnosis on 100 cases.

• One hundred digital cases were evaluated by cardiovascular pathologists • Consensus was achieved on 90 cases • Subtle forms of myocarditis are challenging to diagnose The two histopathology benchmarks used to diagnose myocarditis are the Dallas Criteria, developed in 1984 and the European Society of Cardiology criteria, developed in 2013, which added immunohistochemistry for the detection of CD3+ T cells (lymphocytes) and CD68+ macrophages. Despite their near universal acceptance, the extent to which pathologists use these criteria or their own criteria to consistently render the diagnosis of myocarditis on endomyocardial biopsy (EMB) is unknown. We digitally scanned slides from 100 heart biopsies, including a trichrome stain and immunostaining, that were chosen as representative of myocarditis, non-myocarditis, and borderline myocarditis, as diagnosed per one institution's use of the Dallas Criteria. Eight blinded international cardiovascular experts were asked to render diagnoses and offer a confidence score on each case. No clinical histories were shared. There was full initial agreement across all experts on 37 cases (16 myocarditis and 21 non-myocarditis) and moderate consensus on 35 cases. After individual inquiries and group discussion, consensus was reached on 90 cases. Diagnostic confidence was highest among the myocarditis diagnoses, lowest for borderline cases, and significantly different between the three diagnostic categories (myocarditis, borderline myocarditis, non-myocarditis; P -value=8.49 × 10−57; ANOVA). Diagnosing myocarditis, particularly in cases with limited inflammation and injury, remains a challenge even for experts in the field. Intermediate cases, termed "borderline" in the Dallas Criteria, represent those for which consensus is particularly hard to achieve. To increase consistency for the histopathologic diagnosis of myocarditis, we will need more specifically defined criteria, more granular descriptions of positive and negative features, clarity on how to incorporate immunohistochemistry findings, and improved nomenclature. [ABSTRACT FROM AUTHOR]