A population‐based study of outcomes in polycythemia vera, essential thrombocythemia, and primary myelofibrosis in the United States from 2001 to 2015: Comparison with data from a Mayo Clinic single institutional series

Polycythemia vera (PV), essential thrombocythemia (ET), and primary myelofibrosis (PMF) are BCR/ABL1-(Philadelphia chromosome [Ph−]) negative myeloproliferative neoplasms (MPNs).1 Mutations of Janus Kinase 2 (JAK2), calreticulin (CALR), and myeloproliferative leukemia virus oncogene (MPL) are implicated in these MPNs, leading to clonal hematopoietic stem cell proliferation.1 The most dreaded complication of PV, ET, and PMF is transformation to acute myeloid leukemia (AML).2 While ET and PV have significantly longer survival compared to PMF, their life expectancy has been shown to be inferior to the age- and sex-matched US population.2, 3 There is a plurality of factors leading to impaired longevity in MPN patients; however, the relative contributions of these complications to mortality in the MPN phenotypes are not known. Similarly, the propensity for leukemic transformation has not been studied at the population level. This study aims to describe population-based outcomes of these MPNs, with an emphasis on cause-specific mortality and leukemic transformation. We studied outcomes of PV, ET, and PMF using the Surveillance, Epidemiology, and End Results (SEER) Registry. MPN patients were identified using the International Classification of Diseases for Oncology, third revision (ICD-O-3) codes for PV (9950/3), ET (9962/3), and PMF (9961/3). The database was searched for cases diagnosed between 2001 and 2015 and SEER*Stat was utilized to calculate standardized mortality ratios (SMRs) in order to evaluate the comparative risk of cause-specific mortality. We compared the survival and leukemic transformation trends to a contemporaneous cohort of validated Ph− MPN patients at Mayo Clinic, Rochester. Please see the Supplement for additional information regarding our study methods. 10 725 patients with PV, 8768 patients with ET, and 3689 with PMF are the subject of this report. Additional information regarding patient demographics, incidence, and cohort definition is available in the corresponding Supplement (Tables S1 and S2, Figure S1). The median follow-up was 5.8 years for PV, 5.7 years for ET, and 3.1 years for PMF. Five-year relative survival (RS) for PV, ET, and PMF was 88.3%, 88.7%, and 44.9% respectively. The median overall survival (OS) in PV, ET, and PMF was 11.9 years, 12.1 years, and 4.0 years respectively. Incidence and OS noted in our study are predictably similar to a recent SEER analysis.4 Graphical representations of RS and OS are available in the Supplement (Figures S2 and S3). During the study period, 105 (1.0%) PV, 134 (1.5%) ET, and 122 (3.3%) PMF patients progressed to AML. The median time to AML transformation was 4.4 years, 5.8 years, and 2.3 years for PV, ET, and PMF respectively. At Mayo Clinic, 1952 patients were diagnosed with the MPNs of interest between 2000 and 2017 (444 with PV, 551 with ET, and 957 with PMF). The median follow-up in this cohort was 5.3 years. Rates of leukemic transformation were as follows: 2.7% in PV, 1.6% in ET, and 9.2% in PMF. The median OS in this case series was 12.7 years for PV, 14.9 years for ET, and 4.4 years for PMF (Figure S4). OS between SEER and our case series was in general agreement for PV and PMF; however, shorter survival for ET was noted in the population database. In a previously published Mayo Clinic series, the risk of leukemic transformation at 20 years was 3.9%, 2.6%, and 9.3% in PV, ET, and PMF respectively.3 The proportion of patients in SEER with leukemic transformation was lower at 1%, 1.5%, and 3.3%. Two potential explanations for the perceived discrepancy are (i) inaccurate record of leukemic transformation in SEER and (ii) short duration follow-up. Given that our median follow-up was 3.1–5.8 years, we suspected that the lower proportion of leukemic transformation was primarily due to the latter. The contemporaneous Mayo Clinic cohort demonstrated almost identical leukemia-free survival compared to SEER. However, an exception was noted in PMF, where our institutional series demonstrated a rate of transformation of 9.2%. Though this discrepancy may be partially rooted in referral center bias, when considering that a similar rate of transformation was also published in an Italian cohort of 755 MPN patients, a potential limitation in the population database is apparent.2 We utilized SEER to evaluate cause-specific mortality. 3840 (35.8%) PV patients, 3095 (35.3%) ET patients, and 2430 (65.9%) PMF patients died during the study period. Causes of death were divided into eight categories: myeloid cancer, nonmyeloid hematologic malignancy, solid tumor, other malignancies (referring to "in situ, benign or unknown behavior neoplasm" and "miscellaneous malignant cancer"), cerebrovascular disease, cardiovascular disease, infection, and "other." A summary of these categories is available in the Supplement. Common causes of death in PV included cardiovascular disease (26.1%) and solid tumors (12.7%). A total of 193 (5%) PV patient deaths were related to myeloid cancer. In ET, cardiovascular disease and solid tumors contributed 25.0% and 12.6% of deaths, respectively. Around 282 (9.1%) ET patient deaths were related to myeloid cancer. In PMF, "other malignancies" and myeloid cancer contributed to 41.8% and 15.9% of patient deaths. SMRs are presented in Table 1. Patients with PV, ET, and PMF had more observed deaths than expected when compared with the general population. Patients with PV were at the greatest risk of dying from myeloid cancer (SMR = 14.30, p < .05) and "other malignancies" (SMR 10.03). Similar findings were noted in ET (myeloid cancer SMR = 26.31 p < .05, other malignancies SMR = 8.87 p < .05). For PMF, patients had increased observed mortality from myeloid cancer (SMR = 129.97, p < .05), "other malignancies" (SMR = 97.14, p < .05), nonmyeloid hematologic malignancies (SMR = 8.44, p < .05), and infections (SMR = 4.84, p < .05). When stratified by age, we noted an increased relative burden of observed mortality in younger patients. For PV, this was prominently noted in deaths related to myeloid cancer, "other malignancies," cerebrovascular disease, and cardiovascular disease. In ET, we noted this phenomenon in myeloid cancer, other malignancies, and cerebrovascular disease. In PMF, the greatest observed-to-expected mortality ratios in younger patients were in myeloid cancer, other malignancies, nonmyeloid hematologic malignancies, infection, and cardiovascular disease. The cause-specific mortality data with corresponding SMR provide important insights into these MPNs. First, while considered to be a lower risk for leukemic transformation, both PV and ET patients were at 14.3- and 26.3-fold higher observed risk of death from myeloid malignancies compared to the general population. Additionally, cardiovascular disease was a prominent cause of death in both PV and ET, accounting for 26.1% and 25.0% of all deaths, respectively. The SMR data clearly demonstrate an increased relative risk of mortality from cardiovascular disease and cerebrovascular disease, especially in younger patients with PV and ET. This finding, not previously described at a population level, supports the dictum that in Ph− MPNs, the management of cardiovascular and cerebrovascular complications is of utmost interest, especially in younger patients. Most patients with PMF died from myeloid cancers and "other malignancies" (15.9% and 41.8% of all deaths, respectively), while a high rate of cardiovascular-related and infection-related death was also noted. Similar findings were demonstrated in a previous report of cause-specific mortality in PMF.5 Though the SMR, as expected, demonstrated a markedly elevated observed/expected ratio for mortality from myeloid cancers in PMF, an increased risk of mortality related to infections and cardiovascular disease was also noted. This finding is perhaps related to the overall increased morbidity associated with PMF3 and should prompt clinicians to address and optimize comorbidities accordingly. Interestingly, patients with PMF had an increased observance of nonmyeloid hematologic cancers (SMR = 8.44), corroborating findings of prior reports.6 Given the low incidence of myeloid malignancies, the SMR for myeloid cancer is extremely high for all three Ph− MPN phenotypes across all age demographics. In the absence of disease-modifying therapies, it is not unsurprising to see that Ph− MPN is associated with significantly shorter survival, attesting to the increased morbidity and mortality they cause. Study limitations include the use of the population database which contains limitations including uncertainty about the diagnosis, lack of molecular data which are known to be prognostic, and lack of treatment data. As no cause of death output code exists in SEER for PV, ET, and PMF, respectively, our analysis was limited in capacity to describe specific MPN-related mortality. Additionally, the high proportion of patient deaths attributed to "other malignancies," has likely resulted in underreported disease-related mortality, namely progression and transformation to secondary AML. Finally, the study observation time is relatively short to assess survival in ET and PV. We have attempted to overcome some of these limitations by using a Mayo Clinic cohort for comparison. In summary, the incidence and OS of BCR-ABL-1 negative MPNs were similar to prior reports. MPN patients succumbed to a plurality of conditions, with the greatest increased risk of mortality stemming from subsequent malignancies, namely myeloid cancers. PMF portended the worst prognosis among MPN phenotypes, with increased observed mortality not only to malignancies but also to infections and cardiovascular disease, demonstrating the morbidity associated with this MPN. An increased relative burden of cardiovascular and cerebrovascular disease was noted in younger patients with PV and ET, which should prompt clinicians to address and optimize cardiovascular risk factors in the longitudinal care of MPN patients. The authors report no conflict of interest related to the presented work. The data is publicly available at the National Cancer Institute's Surveillance, Epidemiology, and End Result Program website. The study was determined to be exempt from the requirement for IRB approval. The data is publicly available at the National Cancer Institute's Surveillance, Epidemiology, and End Result Program website. Data S1. Supporting information. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.