-Author name in bold denotes the presenting author
-Asterisk * with author name denotes a Non-ASH member
Clinically Relevant Abstract denotes an abstract that is clinically relevant.

PhD Trainee denotes that this is a recommended PHD Trainee Session.

Ticketed Session denotes that this is a ticketed session.

3239 Development of a Machine Learning–Based Risk Score for Predicting Atrial Fibrillation in Treatment-Naive CLL Patients Initiating BTK Inhibitor Therapy

Program: Oral and Poster Abstracts
Session: 642. Chronic Lymphocytic Leukemia: Clinical and Epidemiological: Poster II
Hematology Disease Topics & Pathways:
Adult, Therapy sequence, Treatment Considerations, Study Population, Human
Sunday, December 8, 2024, 6:00 PM-8:00 PM

Tamar Tadmor, MD1,2*, Guy Melamed3*, Hilel Alapi4*, Sivan Gazit5*, Tal Patalon6* and Lior Rokach, Prof7*

1Technion, Ruth and Bruce Rappaport Faculty of Medicine,, Haifa,, Israel
2Hematology Unit, Bnai Zion Medical Center, Haifa, Israel
3Kahn Sagol Maccabi Research & Innovation Center, Maccabi Healthcare Services, T, Tel-Aviv, Israel
4Kahn Sagol Maccabi Research & Innovation Center, Maccabi Healthcare Services,, Tel Aviv, Israel
5Kahn Sagol Maccabi Research & Innovation Center, Maccabi Healthcare Services, Tel Aviv, Israel
6Kahn Sagol Maccabi Research & Innovation Center, Maccabi Healthcare Services, Tel-Aviv, Israel, Tel Aviv, Israel
7Ben-Gurion University of the Negev,, Department of Software and Information Systems Engineering,, Beer-sheva, Israel

Background: One of the limiting toxicities of BTKi therapy is the development of atrial fibrillation (AF), with an incidence of 3% to 16%.

Aim: To identify patients with chronic lymphocytic leukemia (CLL) who are at high risk of developing AF, using a machine learning approach.

Methods: The CLL cohort was based on data obtained from electronic medical records from Maccabi, the second-largest healthcare organization in Israel. We evaluated more than 100 variables to develop the scoring schema. The optimal scoring model was determined using the code available at https://github.com/ustunb/risk-slim, implemented in Python 3.5 and CPLEX 12.6.

Results: A total of 3964 patients with a CLL diagnosis were available in the database. 208 patients started a BTKi during the study period, 16 of whom developed AF during follow-up. In addition to well-established factors (age, sex, and hypertension) that are used in many existing AF scores that were developed for the general population, the algorithm detected other factors that were associated with a high risk for AF, in particular: type of BTKi used, low eGFR (<30 mL/min/1.73m2), elevated absolute monocytes (>1100/µL), elevated CRP, elevated CK, and elevated B2MG (>2.5 mg/L). Based on the total score, we identified 3 main AF risk groups as following: low (0-6), intermediate (7-11) and high (≥12). The median AFS were 28 and 56 months for the high-risk and intermediate-risk groups, respectively, and was not reached for the low-risk group. The difference between the groups was statistically significant (P=0.0013).

The proposed scoring model reached a C-index of 0.744+/-0.082 and it outperformed the score ranking of Shanafelt et al that obtained a C-index of 0.626+/-0.138 on the same data. The improvement was found to be statistically significant (P=0.024).

Conclusion: Our novel score has a high concordance index to predict the development of AF in patients with CLL treated with BTKi.

Disclosures: Tadmor: Janssen, roche, abbvie, astra, takeda, novartis, beigene, medison: Consultancy, Research Funding.

*signifies non-member of ASH