In this study, we analyzed clinical information and NGS data from a large cohort of patients with PV to reveal the risk factors for thrombosis. After comprehensive analysis, we identified age ≥ 57 years, CVF, previous thrombosis, and high-risk mutations (e.g., DNMT3A, ASXL1, and BCOR/BCORL1 mutations) as risk factors for thrombosis after diagnosis and then established a multiple factor-based prediction model classifying patients into low-risk, intermediate-risk, and high-risk groups. This new model incorporates genetic information into thrombosis prediction and put emphasis on development of precise thromboprophylaxis strategies applicable to patients in different risk groups.
Advanced age and prior thrombosis have been recognized as risk factors in the conventional two-tiered stratification since the publication of ECLAP trial which included 1638 patients.[5] Compared with conventional stratification, the new model has some differences and potential advantages.
Firstly, the MFPS-PV model was developed based on 2016 WHO-defined PV,[23] whereas the conventional model is based on PVSG-defined PV.[9, 10] The 2016 WHO-defined PV recognizes the importance of bone marrow morphology, thus patients previously considered to be “masked” or “prodromic” PV were included according to the new criteria.[28, 29] In fact, “masked” or “prodromic” PV displays a higher risk of thrombosis in younger patients probably due to the lower intensity of treatment.[30] Hence, MFPS-PV is more suitable for patients diagnosed using the current diagnostic standard than the traditional model.
Secondly, the MFPS-PV model highlights the significant predictive value of CVF. The current guideline recommendations from the National Comprehensive Cancer Network clearly emphasize that CVF in patients with PV need to be controlled.[7] Previous studies have also reported that CVF may increase the risk of thrombosis in patients with PV.[14, 15] Moreover, some experts commonly provide different treatment suggestions according to the presence or absence of CVF, as it plays an important role in the prevention and treatment of thrombosis.[1]
Thirdly and notably, we have incorporated genetic aberrations into the prognostic stratification system for the first time. The wide application of NGS in a large cohort of patients with PV allows us to evaluate the impact of specific mutations on thrombosis, which has not been mentioned in most previous studies. Several thrombogenic mutations have been proposed in our study, including mutations in DNMT3A, ASXL1, and BCOR/BCORL1. The epigenetic regulators DNMT3A and ASXL1 are commonly mutated in MPNs and have been proven to be associated with disease initiation and progression[31–33]. Previous reports have also confirmed that mutations in genes encoding epigenetic regulators are associated with increased cardiovascular risks and atherosclerosis, mainly due to upregulation of pro-inflammatory signaling triggered by macrophage deficiency.[34, 35] Moreover, a previous study has demonstrated that mutations in DNMT3A, TET2, and ASXL1 were strongly correlated with thrombosis occurrence, even if they evaluated a limited number of patients with PV.[11] A recent study has also reported that mutations in TET2, DNMT3A, and ASXL1 are significantly associated with prior stroke.[13] Those relevant findings partially support our conclusion. BCOR (BCL6 Corepressor) and BCORL1 (BCL6 Corepressor like 1) are two homologous genes located on chromosome X that have been described in approximately 4–6% of myelodysplastic syndrome cases and 16% of blastic phase chronic myeloid leukemia cases. Mutations in BCOR/BCORL1 are associated with shortened survival in myeloid malignancies.[36–39] In aplastic anemia, patients with BCOR and BCORL1 mutations have an improved response to immunosuppression.[40] Although reports of BCOR/BCORL1 mutations involved in Ph− MPNs are rare, there is still evidence supporting their role in treatment resistance and poor prognosis.[41–43] To the best of our knowledge, this is the first study to reveal that BCOR/BCORL1 mutations may increase the risk of thrombosis in patients with PV.
Fourthly, the MFPS-PV displayed better discrimination power than conventional stratification in predicting thrombosis. When regrouping according to the MFPS-PV, nearly half of the patients in the conventional low-risk group were regrouped as intermediate- or high-risk. Meanwhile, more than one-third of the patients in the conventional high-risk group were regarded as having low or intermediate risk. The actual incidence of thrombosis at follow-up confirmed that MFPS-PV predicted thrombosis more accurately than the traditional model, and that MFPS-PV had the potential to avoided under- or over-therapy in some patients. In addition, the higher C-index of MFPS-PV further supported the fact that MFPS-PV outperformed the conventional one. The predictive accuracy of MFPS-PV was also validated in an external cohort with similar baseline characteristics as the training set but with a much longer follow-up.
In addition to the above advantages, there are some differences between the MFPS-PV and the traditional model. Firstly, the cutoff value of age (57 years) for thrombosis prediction was slightly lower than that reported in previous investigations, mainly 60 years old.[14, 44] This may be due to differences in the prevalence of thrombosis between Chinese and Caucasian populations. According to a global epidemiologic report of thrombosis, the average age at first thrombosis was lower in developing countries where people < 50 years of age frequently experience ischemic heart disease, than in developed countries.[45] Additionally, the incidence of thrombosis is elevated among younger persons (18–45 years) due to the increasing prevalence of CVF in such population.[46] In a large cohort of Chinese patients with cardiovascular disease, the prevalence of coronary heart disease and stroke increased markedly from the age of 40 years.[47] Another study including 783 patients with deep vein thrombosis has reported three frequency peaks, including two smaller peaks at ages 20–24 and 70–74 years and the largest peak at age 45–59 years.[48] In other studies conducted in China, the median age was only 52 years in patients suffering from the first venous thrombosis and even lower in patients suffering from cerebral venous sinus thrombosis (median, 37 years).[49–50] These results indicate that improved awareness of thrombosis prevention and management is urgently required in China. In this study, we present a thrombosis prediction model for PV that is suitable for the current epidemiology of thrombosis in China, and even in the East Asia. A lower age cutoff may help raise public concerns regarding early thrombosis prevention in young patients with PV.
In addition, we found that JAK2V617F mutational status or allele burden had no influence on thrombosis, which has been proven to be significant in some, but not all, studies.[16–18] The possible reasons are that JAK2V617F alterations affect thrombosis in ET rather than in PV,[51–53] and more likely, JAK2V617F mutant burden may be in salient associations with blast transformation instead of thrombosis.[54–56] Moreover, we only evaluated a single time point JAK2V617F at first diagnosis, so the significance of the changes in JAK2V617F allele burden occurring during disease evolution may be ignored. Thus, long-term sequential monitoring of JAK2V617F allele burden during follow-up should be encouraged in future studies to better predict thrombotic risks.
According to the widely accepted treatment recommendations, phlebotomy and low-dose acetyl salicylic acid are first-line therapies for all patients with PV.[1, 7, 8, 26, 27] Additionally, high-risk patients in conventional stratification are advised to receive either hydroxyurea or interferon as cytoreductive therapy. In this study, we analyzed the incidence of thrombosis after diagnosis under different treatment strategies in low-, intermediate-, and high-risk patients according to the MFPS-PV, respectively, and proposed an evidence-based risk-adapted therapeutic regime. We supported conservative management in patients in low-risk group. For these patients, antiplatelet therapy and regular phlebotomy to reach a target HCT level of < 45% can effectively prevent thrombosis. Furthermore, MFPS-PV delineated thrombosis history and CVF as the two most detrimental thrombotic risk factors, and patients with previous thrombosis or CVF treated with combination therapy had a lower risk of thrombosis than those treated with antiplatelet monotherapy. Thus, cytoreduction is strongly recommended in intermediate-risk patients with one of these two risk factors. Antiplatelet and cytoreductive combination therapy is recommended for high-risk patients in order to avoid recurrent thrombosis and improve TFS.
In conclusion, by integrating basic clinical data together with thrombotic high-risk mutations, we presented a multiple factor-based, three-tiered, and four-factor thrombosis prediction model that was applicable to both training and external cohorts of the 2016 WHO-defined PV. This model improves the predictive power of thrombosis and has the potential to realize more precise risk-adapted management for patients with PV.