(SLB-017) A real-world database study of myelodysplastic syndrome and acute myeloid leukemia in prostate cancer patients

Background: Myelodysplastic syndrome (MDS) and the subsequent evolution of acute myeloid leukemia (AML) are second primary malignancies that have been associated with Poly(ADP-ribose) polymerase (PARP) inhibitors treatment in multiple cancers. With an increasing use of PARP inhibitors in patients with prostate cancer (PC), there is need to understand the risk of developing MDS/AML in PC patients.

Objectives: To estimate the incidence of MDS/AML in patients with PCs in real-world setting.

Methods: We conducted a retrospective cohort study across 7 US real-world databases (MarketScan Commercial Claims and Encounters, Multi-State Medicaid, Medicare Supplemental Beneficiaries, Optum’s de-identified Clinformatics® Data Mart, PharMetrics, HealthVerity Commercial Claims, and Optum® de-identified Electronic Health Record). Events of MDS and AML were identified using both a specific (requiring ≥ 2 diagnosis codes) and a sensitive (requiring a single diagnosis code) definition. The incidence rate (IR, events per 1000 person-years [p1000py] at risk) and incidence proportion (IP, persons with events per 1000 persons [p1000p] at risk) of MDS/AML in adult male patients were calculated in three cohorts: overall PC, metastatic PC (mPC), and metastatic castration-resistant PC (mCRPC). Cohorts of mPC and mCRPC were generated by using customized algorithms to determine the metastasis and castration-resistance status. The observation period for the outcome event started from the earliest eligible event of cohort entry till the end of observation in each database. All analyses were restricted to individuals with the first recorded diagnosis of PC in or after 2019 and having ≥ 365 days of continuous observation prior to the cohort entry.

Results: Using the more sensitive definition for a composite outcome of MDS or AML, in the overall PC cohorts, the estimated IRs of MDS/AML ranged from 0.58 to 2.53 events p1000py, and the estimated IPs ranged from 1.26 to 6.23 persons p1000p across 7 databases. The estimated incidence of MDS/AML were higher in the mPC cohorts (IRs: 1.91 to 5.92 events p1000py; IPs: 3.05 to 8.25 persons p1000p) and the mCRPC cohorts (IRs: 2.63 to 6.64 events p1000py; IPs: 4.36 to 8.96 persons p1000p). Across all cohorts in all databases, the incidence of MDS/AML were higher in patients aged ≥65 years than in those aged 18-64 years, and the higher incidence in mPC and mCRPC cohorts were consistently observed within both age categories. When using the more specific definition for MDS/AML, the estimated incidence were lower in all cohorts and all databases.

Conclusions: The findings from real-world data demonstrated that the occurrences of MDS/AML are generally rare in PC patients, but slightly more commonly observed in patients in the later stage of the disease, such as mPC and mCRPC.