Risk of Myelodysplasia and Acute Myeloid Leukemia in Patients with Chronic Lymphocytic Leukemia Treated with Fludarabine and Bendamustine

Background: Patients with Chronic Lymphocytic Leukemia (CLL) receive a variety of treatments that have the potential to induce myeloid malignancies. One such treatment, fludarabine, has been associated with a five-year risk of therapy-related myeloid neoplasia (t-MN) of 4.8% (Smith et. al 2011). This finding was supported by others showing a rate of t-MN of approximately 2.8% to 11% associated with fludarabine (Short et. al 2015). It is not known where new therapies, such as bendamustine, have a similar rate of t-MN compared to fludarabine. Using the Veterans Administration Central Cancer Registry (VACCR), we sought to identify the association between CLL-directed therapy and t-MN.

Methods: We used the VACCR to identify patients diagnosed with CLL between September 1999 and October 2015. Pharmacy Benefits Management records were used identify patients who received bendamustine or fludarabine as initial treatment of CLL. ICD-9 codes for myeloid leukemia, monocytic leukemia, and other unspecified leukemia (205-208.8) as well as myelodysplastic syndrome (238.72-238.76) were used to identify cases of t-MN. Medical records of all cases were reviewed for diagnosis of t-MN through review of clinical notes, laboratory data, and subsequent treatment. Comparisons between groups were performed with independent samples t-test. Comorbidity was assessed using the Charlson (Romano) index score. Cox proportional hazards regression modeling was used to assess associations between treatment and development of t-MN while controlling for age and comorbidity index. The study was approved the Saint Louis VA Medical Center institutional review board.

Results: A cohort of 7155 patients with CLL were available for analysis of which 1084 received fludarabine and 460 received bendamustine as initial therapy. Mean age at initial therapy was higher with bendamustine (68.9 years vs. 64.9, p<0.001) and there was a longer mean duration of time between CLL diagnosis and treatment in patients who received bendamustine (44.9 months vs. 26.9, p<0.001). Comorbidities were higher in patients receiving bendamustine compared to fludarabine, (2.1 vs. 1.6, p<0.001). Patients who were followed a median of 42.8 months after fludarabine and 23.6 months after bendamustine (p<0.001). Median year of diagnosis of CLL was 2005 compared to 2008 for bendamustine with median age of diagnosis of 62 for fludarabine versus 64 for bendamustine. Of 148 patients identified by ICD9 codes, 31 had confirmed t-MN. There were 26 (2.4%) cases of t-MN in the fludarabine group compared to 5 (1.1%) in the bendamustine group (Pearson chi-square p=0.093). In adjusted analysis, there was no significant difference between development of t-MN associated with bendamustine versus fludarabine (adjusted Hazard Ratio aHR 0.75; 95% Confidence Interval: 0.27-2.05).

Conclusion: In this retrospective analysis of treatment for CLL, bendamustine was not associated with reduced rates of t-MN compared to fludarabine. This study is limited by a shorter follow-up period for patients treated upfront with bendamustine, longer follow-up is needed.