Posted: 12/13/2023
A novel artificial intelligence (AI) model may help physicians distinguish and identify prefibrotic primary myelofibrosis from essential thrombocythemia, according to new findings presented by Srisuwananukorn et al at the 2023 American Society of Hematology (ASH) Annual Meeting and Exposition (Abstract 901).
Background
Myeloproliferative neoplasms are a type of cancer in which the bone marrow overproduces certain types of blood cells. Prefibrotic primary myelofibrosis is rarer and has a much worse prognosis than essential thrombocythemia—with a median survival of 12 years vs 22 years, respectively. As a result, prefibrotic primary myelofibrosis may require more aggressive treatment; however, experts may not always agree on a definitive diagnosis when interpreting laboratory and biopsy results.
Despite being integral to informing treatment approaches and enrolling patients in clinical trials, distinguishing the two hematologic malignancies is often challenging with current diagnostic methods.
Study Methods and Results
In the new study, researchers used a novel AI model—which had previously been trained with 32,000 pan-cancer biopsy images and was familiar with general pathologic features—to analyze images from U.S. and Italian patients in order to differentiate between prefibrotic primary myelofibrosis and essential thrombocythemia.
To aid diagnosis, the researchers trained an AI model to distinguish features indicating the two conditions in bone marrow biopsy images from 200 patients. They then tested the model’s ability to differentiate the two types of myeloproliferative neoplasms in biopsies from 26 additional patients.
The researchers found that the AI model was able to return results in an average of just over 6 seconds for a new patient and performed well, demonstrating a 92.3% rate of agreement with human experts. The sensitivity and specificity for prefibrotic primary myelofibrosis diagnosis was 66.6% and 100%, respectively.
Conclusions
“With the combined accuracy, sensitivity, and specificity we saw, it would allow the physician to be confident in one diagnosis vs another and help rule in or rule out the rarer [prefibrotic primary myelofibrosis] diagnosis, particularly for clinical trials,” emphasized lead study author Andrew Srisuwananukorn, MD, Assistant Professor at The Ohio State University Comprehensive Cancer Center. “[Our] hope is that it would maintain this accuracy when tested in larger cohorts,” he added.
The researchers hope that with further testing, the novel AI model could potentially be used as a companion tool for clinical diagnoses and may help physicians match patients with the most appropriate clinical trials—which could result in more effective treatments. However, the researchers cautioned that the model was intended to complement, not replace, human experts.
“What we’re trying to develop is a clinical decision support tool, with an emphasis on support. Physicians with no computer science backgrounds are increasingly recognizing the value of AI [models] and closer to being able to use them for their clinical practice. [M]ore investigations would be needed for this [model] to be used in clinical practice, including testing in cohorts with different racial backgrounds,” underscored Dr. Srisuwananukorn.
The researchers plan to continue refining the AI model and hope to test it with larger data sets. The researchers concluded that AI models could potentially be utilized in the advancement of basic research on myeloproliferative neoplasms to link biologic processes with particular morphological features visible on biopsy slides and develop strategies to predict prognoses or response to treatment.
Disclosure: For full disclosures of the study authors, visit ash.confex.com.
The content in this post has not been reviewed by the American Society of Clinical Oncology, Inc. (ASCO®) and does not necessarily reflect the ideas and opinions of ASCO®.