A 69-year-old female presented with symptomatic atrial fibrillation. Cardiac amyloidosis was suspected due to an artificial intelligence clinical tool applied to the presenting electrocardiogram predicting a high probability for amyloidosis, and the subsequent unexpected finding of left atrial appendage thrombus reinforced this clinical suspicion. This facilitated an early diagnosis by the biopsy of AL cardiac amyloidosis and the prompt initiation of targeted therapy. This case highlights the utilization of an AI clinical tool and its impact on clinical care, particularly for the early detection of a rare and difficult to diagnose condition where early therapy is critical.
Keywords:artificial intelligence; electrocardiography; echocardiography; MRI; cardiac amyloidosis; case report
1. Introduction
AL amyloidosis is a rare condition characterized by the extracellular deposition of light chain protein fibers, also known as amyloid. Amyloid deposits can involve multiple organs, including, but not limited to, kidney, heart, gastrointestinal tract, nerves, and skin. The diagnosis of AL cardiac amyloidosis can be challenging and requires clinical expertise, and there is often a large testing burden [1,2]. A delay in its diagnosis is often a barrier to accessing appropriate care, and in the case of AL cardiac amyloidosis, which can be rapidly progressive and fatal, early diagnosis is critical [3,4]. Artificial intelligence applied to ECG for the screening and early detection of cardiac amyloidosis has recently been validated [5,6].
2. Clinical Case
A 69-year-old female with a medical background of ulcerative colitis status post remote ileostomy presented to the Emergency Department with shortness of breath and fatigue for 3 weeks. She was found to be in new onset atrial fibrillation (AF) with rapid ventricular response. A physical examination demonstrated volume overload with 1+ bilateral ankle pitting edema, jugular venous distension, and bilateral crepitations at the lung bases.
Electrocardiogram (ECG) demonstrated AF with rapid ventricular response (117 (bpm), low QRS voltage, and pseudo-infarct pattern. Artificial intelligence (AI) ECG analysis flagged an increased likelihood of amyloidosis (at 93.1% probability) (Figure 1). Laboratory cardiac biomarkers included mildly elevated troponin T (24 ng/L, reference value < 10 ng/L) without change on serial measurement and elevated NT-proBNP (2609 pg/mL, reference value < 540 pg/mL).