Background:Polycythemia vera (PV) is a chronic myeloproliferative neoplasm (MPN) associated with risk of leukemic or fibrotic progression (Tefferi and Barbui. Am J Hematol. 2017;92;94-108). In addition to the known risk of disease transformation, patients with MPNs have a 60% higher risk of developing second non-hematologic cancers compared to matched controls. Skin cancers are among the most prevalent second cancers, with a reported 2.8-fold increase in risk of non-melanoma skin cancer (NMSC) (Landtblom et al., Leukemia. 2018;32:2203-10). The inherent risk associated with MPNs is exacerbated by cytoreductive treatments, as patients treated with hydroxyurea (HU) were found to have a two-fold higher risk of NMSC compared to patients who were not treated with HU (Barbui et al., Leukemia. 2019;33:1996-2005). Previous studies examined fewer than 2000 patients; the study populations may have included patients with other MPNs and PV. Our objective was to evaluate the frequency of second cancers in patients with a primary PV diagnosis who were treated with HU compared to phlebotomy only using a large US real-world claims database. Methods:This real-world, retrospective cohort study used Optum's deidentified Market Clarity Data consisting of electronic health records (EHR) on 105 million US patients and linked medical and prescription claims between 1/1/2007 to 12/31/2019. More than 82,000 patients with ≥1 PV diagnosis at any time were identified within the extracted dataset. For this analysis, eligible patients were ≥18 years old and required ≥2 PV diagnosis codes with ≥60 days between the first and last diagnosis code in the data period. Patients were indexed at the first diagnosis code of PV preceded by ≥1 year of EHR data (“index”) and required ≥1 year of post-index data or death prior to data period end. Patients with a pre-index diagnosis of secondary polycythemia, leukemia, myelodysplastic syndrome, myelofibrosis, or stem cell transplant were excluded. Second cancers were identified at first occurrence of a diagnosis code for a defined cancer in EHR during specified time periods. Analysis of post-treatment cancer rates used a subset of patients with pre-index medication data and excluded cancer types identified prior to treatment. Results:The total study cohort included 20,089 patients with PV (mean age, 63.0 years; 57.8% male; 88.6% Caucasian). Some patients (15.5%; n=3112) had evidence of a pre-index cancer. Median duration of follow-up was 4.3 years (IQR, 2.4-5.9). Among the total cohort, 35.7% (7181 patients) reported at least one second cancer in the post-index period (“post-index period prevalence”). The rate of second cancers in the post-index period was 109.7 events per 1000 patient years (PY). The most common non-hematologic malignancies were skin cancers (22.1/1000 PY) followed by prostate cancer (8.1/1000 PY) and breast cancer (6.7/1000 PY) (Figure 1). Within the total cohort, 9.1% (1830 patients) reported at least one skin cancer in the post-index period, with 8.3% (1659 patients) experiencing NMSC and 1.4% (286 patients) experiencing melanoma (Figure 1). The post-index period prevalence rate of NMSC was 6 times higher vs. the rate of melanoma (19.8 vs. 3.3/1000 PY). Among NMSCs, the rate of basal cell carcinoma was higher than the rate of squamous cell carcinoma (8.7 vs. 6.9/1000 PY). However, the highest rate was observed for non-specific or unspecified malignant neoplasms (8.9/1000 PY). Within a subset of 17,402 patients with pre-index medication data, 20.2% (n=3512) were exposed to at least 4 weeks of HU (without ruxolitinib) at any time; 19.0% (n=3314) were treated with phlebotomy (PHL) only. The rate of any second malignancy subsequent to treatment was >2.5-times higher in the HU vs. PHL group (140.4 vs. 55.3/1000 PY) (Table 1). Rates of skin cancers were nearly two times higher in the HU vs. PHL group for any skin cancer (31.6 vs. 17.7/1000 PY), NMSC (28.9 vs. 15.7/1000 PY), and melanoma (4.5 vs. 2.9/1000 PY). Conclusion:To our knowledge, this is the largest retrospective study evaluating the frequency of second cancers in a cohort of US patients with PV to-date. We found that patients with PV had high rates of malignancies associated with their primary PV diagnosis. The highest rates were observed for skin cancers, driven by the frequency of NMSC. Patients treated with HU had nearly twice the rate of skin cancers (including NMSC and melanoma) compared to PHL-only treated patients.