An artificial intelligence-based model may facilitate selection of optimal frontline tyrosine kinase inhibitor (TKI) therapy for patients with newly diagnosed chronic phase chronic myeloid leukemia (CML-CP). These findings from a retrospective analysis were reported in the American Journal of Hematology.

Long-term use of BCR-ABL1 TKI, such as imatinib, dasatinib, nilotinib, or ponatinib, has resulted in deep and long-lasting responses in most patients with CML-CP. Nevertheless, data from clinical trials indicate that a substantial minority of patients with the disease will require a switch to an alternate BCR-ABL1 TKI at some point. Currently, the selection of first-line and subsequent TKI therapy is frequently based on subjective criteria.

In this study, a machine-learning assisted approach called the LEukemia Artificial intelligence Program (LEAP) was used to facilitate decision-making related to frontline BCR-ABL1 TKI selection in patients with CML-CP.


Continue Reading

This analysis included 630 patients with newly diagnosed CML-CP enrolled in prospective studies of frontline BCR-ABL1 TKI therapy between July 30, 2000, and November 25, 2014, conducted at the MD Anderson Cancer Center in Houston, Texas.

Of these patients, 504 were included in the training/validation cohorts for development of the LEAP model and 126 were included in the test cohort for individualized assessment of optimal TKI therapy using the LEAP model.  

The LEAP model was developed using 101 variables, including patient-related factors such as age at disease diagnosis, height, weight, bone mass index, gender, ethnicity, marital status, comorbidities, and residential area. Laboratory-based parameters included peripheral blood specimens such as white and red blood cell counts, platelet count, hemoglobin level, and percentages of different types of white blood cells. In addition, the blast level and levels of different types of white blood cells in bone marrow; the Philadelphia chromosome abnormality variant; BCR-ABL1 transcript type and level; risks scores determined according to the Sokal, Hosford, and European Treatment and Outcome Study (EUTOS) classification schemes; and daily doses of first-line TKI therapy, as well as many other factors, were also included.

The study investigators explained that “the performance of the LEAP CML-CP model was evaluated by calculating the hazard ratios for overall survival [OS] by treatment in the test cohort. The treatment option with the lowest hazard ratio was considered the best treatment option for individual patients.”

At a median follow-up of 139 months for the overall group, a key study finding was the 5-year OS rate for patients in the test group receiving the LEAP-recommended TKI (n=94) was 98% compared with 77% in the 32 patients treated with a non-LEAP-recommended TKI (P <.001).

Furthermore, on multivariate analysis, the LEAP model recommendation was independently associated with improved OS (hazard ratio [HR], 0.280; 95% CI, 0.087-0.895; P =.032). Age at diagnosis and comorbidity burden were also independent prognostic factors for OS.

“The LEAP CML-CP has the potential to support patients, caregivers, and physicians for personalized treatment recommendations, and contribute to further improvement of clinical outcome in patients with CML-CP,” stated the study investigators.

They further added that “given the impracticalities of conducting randomized clinical trials in specific subsets of patients with a relatively rare disease such as CML-CP, this approach may pave the way for a new era of personalized treatment recommendations for individual patients based on their unique clinical, social geodemographic, biological, chromosomal, and molecular features.”

Although this type of model may also be useful as an adjunct when selecting an alternative TKI therapy for patients whose response to treatment with a frontline TKI is not optimal, the study investigators clarified that a different set of patients corresponding to this clinical situation would be required to evaluate the model in this setting.

Disclosures: Multiple authors declared affiliations with or received funding from the pharmaceutical industry. Please refer to the original article for a full list of disclosures.

Reference

Sasaki K, Jabbour EJ, Ravandi F, et al. The LEukemia Artificial Intelligence Program (LEAP) in chronic myeloid leukemia in chronic phase: a model to improve patient outcomes. Am J Hematol. Published online November 12, 2020. doi:10.1002/ajh.26047