Associations between Myeloid Malignancies and Osteoporosis Using Real-World Data from the Trinetx Database

Myelodysplastic syndromes (MDS), myeloproliferative neoplasms (MPNs), and clonal hematopoiesis have been linked with increased risk of osteoporosis. However, current clinical practice guidelines do not recommend routine bone density screening for all patients with these diseases and rates of screening are unknown. Further, outcomes of patients with myeloid malignancies and concurrent osteoporosis are not well understood.

The objectives of this retrospective observational study were to examine the frequency of osteoporosis in patients with MDS or MPNs, assess the rate of bone density screening in these populations, and evaluate outcomes including fracture risk and mortality in patients with concurrent osteoporosis and myeloid diseases. We collected real-world data from the TriNetX Research Network, a global federated network of de-identified electronic medical record data including approximately 131 million patients from 93 healthcare organizations. Data were queried for demographic characteristics and ICD-10 codes between 2007 and 2024. Specifically, we surveyed ICD-10 codes for osteoporosis, MDS, acute myeloid leukemia (AML), myelofibrosis (MF), chronic myeloid leukemia (CML), chronic MPNs, polycythemia vera (PV), and essential thrombocythemia (ET).

The frequency of an ICD-10 diagnosis of osteoporosis in the general TriNetX population was approximately 2% and the rate of documented bone density screening in patients with osteoporosis was 40%. The frequency of osteoporosis in patients with myeloid malignancies in this population was 14% in MDS, 10% AML, 19% MF, 8% CML, 13% chronic MPN, 11% PV, and 12% ET. To better estimate the actual frequency of osteoporosis in tested patients, we narrowed our query to a population that received screening with a dual-energy x-ray absorptiometry (DEXA) scan which included around 2.4 million patients.

The rate of bone density screening in patients with myeloid diseases ranged between 8-19%. Surveyed demographic characteristics in patients with myeloid diseases who underwent DEXA screening included sex (74% female, 21% male, 5% unknown), race (74% White, 11% unknown, 9% Black, 3% Asian, 3% other race), and mean age of 70. Out of all patients who had a DEXA scan, 41% had a diagnosis of osteoporosis. Of patients with myeloid malignancy who underwent DEXA screening, osteoporosis was diagnosed in 55% with MDS, 43% AML, 56% MF, 43% CML, 52% chronic MPN, 52% PV, and 52% ET.

We also evaluated outcomes in patients with concurrent myeloid malignancy and osteoporosis including frequency of osteoporotic fractures and mortality. The rate of osteoporotic fractures was 11% in all patients with a diagnosis of osteoporosis versus 23% in patients with MDS, 20% AML, 21% MF, 19% CML, 21% chronic MPN, 16% PV, and 18% ET. The TriNetX platform was also used to compare outcomes between cohorts and perform statistical analysis. This analysis included DEXA-screened patients who had a diagnosis of any myeloid disease with either an additional diagnosis of osteoporosis (Cohort A) or no osteoporosis (Cohort B). Propensity score matching was performed based on demographic factors (age, sex, race) and type of myeloid malignancy to reduce bias resulting in 26,815 patients in each cohort. The risk of death was 21% in Cohort A versus 18% in Cohort B with an odds ratio of 1.23 (95% CI 1.182-1.287). Median survival in Cohort A was 12.4 years versus 15.2 years in Cohort B with a hazard ratio of 1.16 (95% CI 1.117-1.207). Survival probability at approximately 17 years after index event (diagnosis of myeloid disease and osteoporosis) was 36% in Cohort A versus 47% in Cohort B.

Population health informatics provide an efficient and powerful tool to study observational real-world data. This analysis found the frequency of osteoporosis was higher in DEXA-screened patients with myeloid malignancies compared to a general population of DEXA-screened patients. We also showed that patients with myeloid malignancies and osteoporosis may have worse outcomes compared to those without osteoporosis. Our analysis is limited by accuracy of provider-entered ICD-10 codes and bias towards female sex within a DEXA-screened patient population. Nevertheless, our findings underscore a need for improvement in osteoporosis screening among patients with myeloid diseases and may provide groundwork for use of therapeutic interventions in this high-risk population.