Myelodysplastic Syndromes and Acute Myelogenous Leukemia Resulting from Therapy for Autoimmune Disease, a Case-Control Cohort Study of 40,011 Patients

Background

Myeloid neoplasms (MN) occurring secondary to treatment of non-malignant disease, especially autoimmune diseases (AID), and the overlap between them, is an increasing clinically important observation. Confounding variables for myeloid transformation include the nature of autoimmune disease as an incessant, chronically activated inflammatory cascade that has been associated with an increased risk of myeloid malignancy, and disease modifying therapies, often with cytotoxic and immunosuppressive drugs. Thus, the addition of therapeutic agents for treatment of AID theoretically amplifies the opportunity for MN to develop in genetically susceptible individuals. To our knowledge, there is no large study evaluating a comprehensive primary AID population that developed MDS or AML with detailed assessment of therapeutic interventions for AID. The presented study queries the association of cytotoxic, anti-inflammatory, and immune-modulating agents to treat AID as risk factors for developing myelodysplastic syndromes (MDS) or acute myeloid leukemia (AML).

Methods

In a 10-year period from 2004-2014, a total of 40,011 patients with an ICD-9 coded diagnosis of AID were seen at Mayo Clinic Arizona and Florida campuses; 311 patients from this population had a concomitant coded diagnosis of MDS and/or AML. In an IRB approved, retrospective chart review, we confirmed 86 cases meeting strict inclusion criteria for AID and confirmed MDS or AML. We performed a case-control match from the same primary autoimmune population for age, gender, and autoimmune specific diagnosis at a 2:1 ratio querying for therapeutic interventions.

Results

Myeloid neoplasm specifics in our case cohort of 86 patients were 55 MDS, 21 de novo AML, and 10 AML with history of MDS. Average age was 72 years with a slight male predominance (57%). Rheumatoid arthritis (26.4%), psoriasis (20.7%), systemic lupus (13.8%), was the most common autoimmune profiles. Median onset of autoimmune disease to diagnosis of myeloid neoplasm was 6 years (range 1-54 years). Up to 26% controls documented no systemic agent for treatment of autoimmune disease and similarly 12.8% of the case cohort (p=0.142). Case and control populations treated with systemic therapies had similar exposures by category for immune-modulating agents (33.7% and 37.8% respectively) and cytotoxic therapies (47.7 and 44.8% respectively). A total of 57/86 cases (66.3%) received either a cytotoxic or immune-modulating agent. In comparison, 105/172 (61.1%) controls received either agent; p=0.495. Azathioprine was the only statistically significant agent exposure observed more frequently in the study cohort than controls with an OR of 7.05 (p= < 0.001). There was no increased incidence of MDS or AML in TNF-antagonist treated patients. No significant correlation for duration of agent exposure by drug category was observed.

Discussion

In a large primary AID population, azathioprine exposure was associated with a 7 fold risk of MDS or AML (p=< 0.001). Notable, but not statistically significant case cohort trends among cytotoxic agents was exposure to cyclophosphamide (OR 3.58) followed by mitoxantrone (OR 2.73). Methotrexate, mercaptopurine, and mycophenolate had favorable odd ratios, but these were not statistically significant. No association was found for TNF-inhibitory agents. Finally, there was no association of drug exposure time to the incidence in development of MDS or AML.