Early diagnosis of cardiac amyloidosis (CA) can prevent irreversible heart damage, guide appropriate disease management and improve quality of life and prognosis. The complexity of diagnostic techniques, however, can made CA detection extremely challenging.
Fortunately, automated, artificial intelligence (AI)-driven echocardiography can now provide clinicians with a powerful diagnostic tool. Advanced solutions can analyze echo images to quickly detect signs of disease, and make recommendations on the likelihood of CA presence, guiding further diagnostic workup and putting patients on the path to effective care.
CA's Diagnostic Challenges
CA is a rare condition in which misfolded amyloid proteins accumulate in the heart tissue, leading to progressive heart dysfunction. Over time, these deposits can reduce the heart’s ability to pump blood effectively, often leading to heart failure.
The prognosis is variable and depends on the CA subtype. Untreated patients with amyloid light chain (AL) CA have an average survival time of just 9 to 24 months, but in later-stage disease, this may drop to as little as 5 months. Those with transthyretin amyloidosis (ATTR) can live between seven and ten years, though in early-stage disease, survival can extend beyond 69 months.
Accurate and early detection is crucial because treatments, particularly transthyretin stabilizers, are most effective when initiated in the early stages of the disease. Traditional methods, such as Doppler echocardiograms, however, can lead to missed diagnoses due to the high variability in interpretation and overlapping features with other cardiac diseases like hypertrophic cardiomyopathy.
CA is characterized by biventricular wall thickening on echo, with thickening of the left ventricle (LV) being a common red flag. However, the broad overlap in clinical features means it can mimic hypertrophic cardiomyopathy. Accurate interpretation requires a high degree of specialist knowledge, which is not always available due to the rarity of CA.
What’s more, there is a high degree of inter- and intra-reader variability, and interpretations often need to be made in real-time, adding to the complexity and increasing the margin for error. It addition, it is a time-consuming procedure, taking around 30 minutes.
AI in Echocardiography
In recent years, the use of AI in diagnostic imaging has gained significant momentum, particularly in the field of echocardiography. These models can analyze echo video clips, down to the pixel level, and recognize patterns and features that may go unnoticed by the human eye.
It is an advanced capability that enables models to identify key images and quantify critical areas, and link them to specific disease patterns.
Combining AI-driven insights with clinician expertise makes echocardiography more accurate, more consistent, and reduces variability between operators. Crucially, because it has the potential to uncover predictive insights beyond human perception, it allows for earlier identification of diseases.
AI in CA Detection
In CA, AI-driven echocardiography allows for the quantification of results for disease detection. This enables clinicians to quickly initiate further diagnostic work up, such as cardiac MRI, electrocardiogram and bone scintigraphy, as well as genetic testing where indicated.
Advanced AI solutions utilise a number of features to detect the signs of CA and hypertrophic cardiomyopathy, quickly and accurately, within a single apical four chamber view.
EchoGo® Amyloidosis, for example, analyzes an echocardiogram video clip and identifies suspected disease, providing a diagnostic output of either 'suggestive' or 'not suggestive' of CA, based on the automated analysis of several echocardiographic markers, including ventricular volume, cardiac function, and myocardial movement. These insights are automatically sent to clinicians to support them in clinical decision making.
The technology is FDA-approved and incredibly accurate. In the pivotal clinical trial, conducted at 15 sites with 3,593 patients, CA detection was achieved with a sensitivity of 85% and a specificity of 94.9%. [i] The tool also performed consistently across different CA subtypes, with sensitivity rates of 84.4% for AL Amyloidosis, 85.8% for wild-type ATTR, and 86.3% for variant ATTR.
EchoGo® Amyloidosis, which was developed in collaboration with Janssen Biotech, Pfizer, and the Mayo Clinic, could help to close the diagnostic gap in CA.
It has been rigorously tested in a control group of people with clinically relevant cardiac diseases, to simulate the real-world clinical environment. It excelled in identifying underlying amyloidosis in patients with heart failure with preserved ejection fraction (HFpEF), a population in which the condition is significantly underdiagnosed. Throughout the trial, it consistently outperformed current clinical practices in detecting amyloidosis, while maintaining high sensitivity and specificity across all major subtypes.2
During external validation, carried out with 2,719 Mayo Clinic patients, the AI tool showed an area under the curve (AUC) of 0.93, with 85% sensitivity, 93% specificity, 78% positive predictive value, and 96% negative predictive value. It also performed consistently across different CA subtypes, with 84% sensitivity for AL amyloidosis, 85% for wild-type ATTR, and 86% for variant ATTR.2
Importantly, the platform integrates seamlessly with the clinical workflow to provide clinicians with definitive results on the presence of CA within 30 minutes.
Early Detection, Early Intervention
As the burden of cardiovascular disease continues to grow, healthcare institutions are increasingly looking for innovative solutions.
AI-powered echocardiography is one such solution. By enabling earlier detection of CA, it allows for earlier intervention, helping to prevent complications, improve prognoses, and enhancing the lives of people living with the condition.
References:
-
Shams, P., & Ahmed, I. (2023). Cardiac amyloidosis. In StatPearls [Internet]. StatPearls Publishing.
-
EchoGo Amyloidosis 510K FDA submission.
