Association of socioeconomic indicators with survival in chronic myeloid leukemia: A retrospective analysis

Abstract

Abstract Background Tyrosine kinase inhibitors (TKIs) have improved outcomes of patients (pts) with chronic myeloid leukemia (CML). However, their high cost remains a major burden and leads to early switching to alternative TKIs or delayed treatment initiation. To date, no observational study has directly explored the impact of socioeconomic indicators (SEI) on outcomes of pts with CML treated with TKIs. Methods This is a retrospective study of adults with chronic or accelerated phase CML (1/2007–12/2022) at Cleveland Clinic. Collected data included demographics, comorbidities, hematologic parameters, first-line TKI, treatment response, and survival. SEI of interest obtained from electronic health record (EHR) with support from specialty pharmacy were: occupation, patient assistance, copay assistance, insurance, and median household income. Responses were assessed using BCR: ABL RT-qPCR and defined per ELN 2020 and NCCN 2024. Outcomes included overall survival (OS) and event-free survival (EFS), estimated via Kaplan-Meier. Associations between SEI and outcomes were analyzed using univariable and multivariable Cox regression adjusted for age, gender, race, and comorbidities. Insurance was classified by access level: 1) high-access (e.g., private, Medicare + supplements), 2) low-access (e.g., Medicaid, Medicare without supplements), and 3) minimal access (uninsured/self-pay). Occupation was grouped into: 1) professional/technical, 2) service/clerical, 3) manual labor/skilled trades, and 4) non-working. Copay assistance was defined as financial support (e.g., grant or copay card) to cover copay, co-insurance, or deductible. Patient assistance referred to manufacturer-provided medication at no cost, bypassing insurance. Median household income was estimated using the most recent U.S. Census ZIP code data. Results A total of 347 pts were included; median age 56 (range: 43-67), 86% (n=296) were White 58% were males (n=200). First-line TKIs were imatinib (n=183, 53%), dasatinib (n=105, 30%), nilotinib (n=51, 15%) & bosutinib (n=8, 2%). Median household income was $ 50,325 (41,674 - 65,113). Only 1% (n=5) discontinued first-line TKI due to financial toxicity. Median follow-up time was 7.2 years. Patient assistance: 35% (n=47 out of 136) of pts received patient assistance. The 5-year OS & EFS were 81% & 26% in those with patient assistance and 78% & 15% in those without (Log rank P OS: P=0.9, EFS =0.07). On MVR, there was no significant difference in OS (Hazard ratio (HR: 0.9, 95CI: 0.4-1.8) or EFS (HR: 0.8, 95CI 0.5-1.2) between those two groups (P>0.05). Copay assistance: In the following groups: 1. copay card (27%, n=40/147), 2. grant (29%, 29/147) and 3. received no copay assistance (54%, 78/147), the 5-year OS & EFS were: 100% & 19%; 70% & 33%; and 72% & 17%, respectively (Log rank P: OS<0.01, PFS =0.6). In the MVR for OS and compared to no copay assistance group as a reference, the copay card group had improved OS (HR: 0.2, 95CI: 0.03-0.8) but not the grant group (HR: 1, 95CI: 0.4-2.2) (P=0.02). For EFS, neither copay card group (HR: 1.1, 95CI: 0.7-1.7) or the grant group (HR: 0.8, 95CI: 0.5-1.3) were different than no copay assistance group (P=0.4). Occupation: In 272 pts with known occupation, 61% (n=166) were professional/technical jobs, 10% (n=27) were clerical jobs, 17% (n=46) were manual labor, and 12% (n=33) were unemployed. The 5-year OS & EFS were 82% & 24% in manual labor / skilled trades group, 68% & 27% in non-working group, 87% & 24% in professional / technical group and 89% & 39% in the service / clerical group (Log rank P: OS=0.2, PFS=0.6). The type of occupation did not predict OS or EFS on MVR. Insurance type: 68% (233/344) had comprehensive/high access insurance, 30% (104/344) had basic/low access insurance, and 2% (n=7/344) were uninsured. The 5-year OS & EFS were 75% & 18% for basic / low-access group, 86% & 27% for comprehensive / high-access and 86% & 48% uninsured / minimal access (Log rank P: OS<0.01, PFS =0.04). In the MVR model, no difference in OS (P=0.5) or EFS (P=0.2) was seen between the groups. Median household income: It did not predict OS or EFS on MVR (P>0.05) Conclusion In this real-world study, lack of copay assistance predicted worse OS in CML pts. The availability of patient assistance programs may have mitigated poor outcomes in pts who are uninsured or have financial difficulties. Our study findings highlight the complex and sometimes partial impact of SEI on long-term outcomes in CML.