Background: Non-CML myeloproliferative neoplasms (MPNs), which include essential thrombocythemia (ET), polycythemia vera (PV), myelofibrosis (MF) and MPN not otherwise specified (MPN-NOS), are characterized by activation of JAK2 signaling and abnormal blood cell production. MPN patients, especially those aged ≥65 years (in whom MPN incidence is highest), are at increased risk for cardiovascular and other comorbidities (Vannucchi et al. Blood 2007;110:840-6, Marchioli et al. J Clin Oncol 2005;23:2224-32). However, comorbidity rates in elderly MPN patients as compared with non-MPN controls have not been described in previous literature. MPNs are more prevalent in the elderly and therefore Medicare enrollees are a highly relevant source for US-based epidemiological data for these diseases.
Objective: To compare comorbidity rates from four subtypes of elderly MPN patients (ET, PV, MF and MPN-NOS) with matched non-MPN/non-cancer controls.
Methods: Retrospective data were taken from the Survey, Epidemiology, and End Results (SEER)-Medicare linked database in the US, which combines clinical information from the SEER cancer registry (MPN reporting has been required since 2001) with medical and pharmacy claims for Medicare enrollees. Patients with a new MPN diagnosis between Jan 1, 2001 and Dec 31, 2007 were selected and evaluated for comorbidities from Jan 1, 2008 (index date) through Dec 31, 2008 (follow-up end date). Patients were classified by MPN subtype based on the most recent diagnosis information (ICD-O-3 from the SEER registry or ICD-9-CM from Medicare claims) before the index date. Patients who died before follow-up end, had HMO or discontinuous Medicare enrollment during the follow-up year, had enrollment based on end stage renal disease, or were diagnosed with a non-MPN malignancy before follow-up end were excluded. Comorbidities were defined by ICD-9-CM diagnosis codes comprising the Charlson Comorbidity Index (CCI) (Charlson et al. J Chron Dis 1987;40:373–83; Deyo et al. J Clin Epidemiol 1992;45:613-9) as well as other adverse conditions. Separate non-MPN/non-cancer control groups were selected for each MPN subtype and matched (5:1) on birth year, gender, ethnicity, geography, and reason for Medicare eligibility. The proportion of patients with individual comorbidities and mean CCI during the follow-up year were compared between MPN cases and controls using univariate chi-square tests and t-tests.
Results: A total of 1,355 MPN patients (n = 445 ET, 684 PV, 81 MF, 145 MPN-NOS) were identified for inclusion and assigned matching controls. For ET, PV, MF and MPN-NOS cases, respectively, mean [SD] age at index was 75.5 [9.7], 70.8 [11.3], 70.8 [10.4] and 74.1 [8.9] years and % female was 69.0, 43.9, 54.3, and 55.2. Mean [SD] years between first MPN diagnosis and study index date was 3.1 [2.0], 3.4[1.9], 2.7 [2.0], and 3.1 [2.1] for ET, PV, MF and MPN-NOS cases, respectively. Comorbidity rates during the follow-up period for MPN cases and matched controls are shown in the figure. Compared with controls, ET, PV and MPN-NOS cases had significantly (p<0.05) higher rates of serious cardiovascular events and comorbidities during the follow-up year, including myocardial infarction (ET vs. control: 8.1% vs. 4.0%, PV vs. control: 8.6% vs. 4.3%, MPN-NOS vs. control: 9.7% vs. 5.0%), congestive heart failure (CHF) (ET vs. control: 16.4% vs. 12.7%, PV vs. control: 18.4% vs. 10.1%, MPN-NOS vs. control: 22.1% vs. 12.4%), peripheral vascular disease (PVD) (ET vs. control: 20.0% vs. 15.4%, PV vs. control: 19.4% vs. 13.6%, MPN-NOS vs. control: 27.6% vs. 15.7%), and stroke (ET vs. control: 17.8% vs. 13.4%, PV vs. control: 17.8% vs. 13.1%, MPN-NOS vs. control: 22.1% vs. 13.9%). MF cases also had higher rates of CHF, PVD and stroke, but due to small sample size, only congestive heart failure was significant. Other comorbidities were significantly higher in all MPN subtypes, notably thromboembolism, renal disease, moderate-to-severe liver disease, and infections.
Conclusions: Medicare enrollees with MPNs generally experienced significantly higher comorbidity rates and overall comorbidity burden (based on mean CCI scores) than matched controls. These findings have implications for both the clinical management of MPN patients as well as for health economic assessments, since a substantial portion of the cost of care for MPNs may reside in treatment of comorbidities not directly coded to MPNs.
