Basophilia Predicts Poorer Outcomes in Essential Thrombocythemia, Polycythemia Vera, Primary Myelofibrosis, and Myeloproliferative Neoplasm, Unclassifiable

Basophils are hematopoietic cells derived from myeloid progenitor cells and are found in increased numbers in some myeloid neoplasms, particularly chronic myeloid leukemia (CML). Marked basophilia is a poor prognostic indicator in CML and defines accelerated phase according to the revised 4th edition World Health Organization (WHO4R) classification and the International Consensus Classification (ICC). Basophilia is less well-documented in the classic BCR::ABL1-negative myeloproliferative neoplasms (MPNs) essential thrombocythemia (ET), polycythemia vera (PV), and primary myelofibrosis (PMF). The clinical behavior of these BCR::ABL1-negative MPNs is heterogeneous, with varying propensity to progress to fibrotic or blast phase. Prior studies have reported that patients with primary or secondary myelofibrosis (MF) with high basophil counts have increased risk of progression to blast phase, shortened survival, and more frequent CALR mutations [1]. The significance of basophilia across ET, PV, and PMF, as well as its association with driver mutations other than the classic JAK2 V617F, CALR, and MPL, have not been previously characterized. In the current study, we examined a broad cohort of MPNs to determine associations of basophilia with clinical and molecular features and patient outcome. We performed a search of bone marrow (BM) biopsies reviewed at the Massachusetts General Hospital between 2008 and 2019 for patients diagnosed with an MPN other than CML, chronic eosinophilic leukemia, or chronic neutrophilic leukemia. All cases were classified according to the WHO4R. We identified 195 patients: 79 ET, 14 PV, 16 pre-fibrotic PMF (pre-PMF), 45 overt fibrotic phase PMF (F-PMF), 21 post-ET-MF, 4 post-PV-MF, and 16 MPN, unclassifiable (MPN-U). Patients with prior reported diagnoses of MPN were classified according to their first BM biopsy at our institution. We defined basophilia as both relative (> 3% of all white blood cells) and absolute (> 0.3 K/μL) increase in blood basophils, which are the upper limits of the reference ranges in our hospital, and/or ≥ 3% basophils in the BM aspirate, occurring within 6 months of the patient's first biopsy diagnostic of MPN at our institution. We assigned each pre-PMF, F-PMF, post-ET-MF, post-PV-MF, and MPN-U patient a prognostic score according to the Mutation-Enhanced International Prognostic Score System (MIPSS70+ ver2.0) [2, 3]. For statistical analysis, Chi-square test was used to compare categorical variables. Student's t-test was used to compare continuous variables summarized as mean values. Leukemia-free survival (LFS, determined from the date of first diagnostic assessment at our institution to progression to blast phase or death) and overall survival (OS, determined from the date of first diagnostic assessment at our institution to death from any cause) were assessed using the Kaplan–Meier method and were compared between groups using the log-rank test. Statistical significance was defined as 2-tailed p < 0.05. Multivariable analysis was performed including all variables significant in univariate analyses, using stepwise elimination to arrive at the final models. Of the 195 patients with MPN, 43 patients (22%) were noted to meet our strict definition of basophilia. Patients with ET and PV had a lower incidence of basophilia (8%) compared with patients with pre-PMF, F-PMF, post-ET-MF, post-PV-MF, or MPN-U (together 35%, p < 0.0001). Compared with ET patients, pre-PMF patients trended toward having a higher incidence of basophilia (19% vs. 5%, p = 0.056). 4/43 patients (9%) had increased basophils only in the BM without basophilia in the peripheral blood. In the basophilia (baso+) group, the median basophil percentage was 4.1% (range 0%–13.0%) and the median absolute basophil count was 0.7 K/μL (range 0–7.53 K/μL) in blood. In the no basophilia (baso−) group, 18/152 patients (12%) subsequently developed basophilia during follow-up, at a median of 19.6 months (range 6.3–128.3 months) after the initial MPN diagnosis. Characteristics for the 152 baso− and the 43 baso+ cases are shown in Tables S1 and S2, respectively. Patients in the baso+ group were significantly older (p < 0.001) and had higher white blood cell count (p < 0.001) and reticulin grade (p = 0.0007), and lower hemoglobin (p = 0.01) and platelet counts (p < 0.001) than the baso− group. Baso+ cases had more frequent abnormal cytogenetics (p = 0.0002), a greater number of total mutations detected on next-generation sequencing (mean 2.9 vs. 1.7, p < 0.0001), and more frequent JAK2 (p = 0.0004), SF3B1 (p = 0.0083), and SRSF2 (p = 0.0159) mutations, but less frequent CALR mutations (p = 0.0018) (Figure S1). For pre-PMF, F-PMF, post-ET-MF, post-PV-MF, and MPN-U patients, fewer baso+ patients had therapy that consisted of observation or aspirin alone (p = 0.048), more patients required therapy with a JAK2 inhibitor (p = 0.014), and similar percentages underwent hematopoietic stem cell transplant (HCT) (22.7% vs. 19.4%, respectively, p = 0.68) (Tables S1 and S2). In the ET and PV patients, there was no significant difference in treatments between the baso− and baso+ groups. Median follow-up time of all patients was 63 months (range 0–199 months). The baso+ group had significantly shorter OS (54 months) and LFS (46 months) compared to the baso− group (both not reached, p < 0.0001) (Figure 1). OS and LFS were also significantly shorter in basophilia cases when ET and PV patients were excluded from analysis (p = 0.003 and p = 0.002, respectively) (Figure 1). 18.6% of baso+ cases progressed to blast phase compared to only 2.6% of baso− cases (p < 0.0001). In a multivariable analysis (MVA) of all PMF patients (n = 61) that included basophilia, age, F-PMF versus pre-PMF, and the MIPSS70+ ver2.0 score, basophilia was an independent prognostic variable for both OS and LFS: basophilia (HR 2.714 [1.209–6.095], p = 0.016) and MIPSS70+ ver2.0 score (very high: HR 5.466 [1.180–25.311], p = 0.03; high: HR 5.963 [1.666–21.349], p = 0.006; and intermediate: HR 3.695 [1.106–12.343], p = 0.034 versus low/very low risk) were independently significantly associated with OS. Basophilia (HR 4.109 [1.703–9.913], p = 0.002) and MIPSS70+ ver2.0 score (very high: HR 5.661 [1.214–26.372], p = 0.027; high: HR 11.410 [3.004–43.338], p < 0.001; and intermediate: HR 3.200 [0.933–10.979], p = 0.064 vs. low/very low risk) were independently significantly associated with LFS; neither age nor F-PMF versus pre-PMF were independently prognostic in the final models. In this retrospective study, we investigated 195 patients diagnosed with MPNs and classified the cases according to the WHO4R as ET, PV, pre-PMF, F-PMF, post-ET-MF, post-PV-MF, or MPN-U. The presence of increased basophils has been anecdotally noted as a poor prognostic factor in MPNs, but this has not been rigorously studied across the spectrum of MPN. Compared with prior studies that have evaluated basophilia in MPN, our cohort of MPN cases was larger (195 patients compared to 32 patients [1]) and included cases of non-fibrotic ET and PV. We also analyzed for the presence of basophilia in all MPN patients at the time of diagnosis rather than retrospectively evaluating for basophilia in MPN cases that had already progressed to blast phase. We found that the presence of basophilia in MPN is associated with more aggressive disease, as characterized by abnormal cytogenetics, mutation burden, BM reticulin grade, and poorer outcomes. Patients with PMF, post-ET-MF, post-PV-MF, and MPN-U with basophilia had higher MIPSS70+ ver2.0 scores compared to those without basophilia. Overall, baso+ patients more frequently received therapies beyond aspirin alone, had shorter OS and LFS, and higher incidence of progression to blast phase compared to baso− patients. The increased mutation burden in basophilic MPN cases manifested as both a greater number of total mutations as well as a different mutation profile (more frequent JAK2, SF3B1, and SRSF2 mutations). SF3B1 mutation correlates with shorter time to myelofibrosis progression [4] and has been associated with higher-risk MPN disease subtypes (primary and secondary MF) [5]. SRSF2 is known to confer a poor prognosis in patients with MPNs [6]. In contrast to the prior study on basophilia in myelofibrosis cases [1], we noted that baso+ cases were less likely to have CALR mutations, which may reflect our inclusion of early/pre-fibrotic MPN cases rather than only those that had progressed to secondary MF or AML. It has been previously postulated that TGF-β secretion by megakaryocytes leads to increased basophil differentiation and proliferation via IL-3, as well as increased fibrosis, suggesting that basophilia may merely be a secondary indicator of more advanced BM fibrosis [1]. However, even after excluding cases of nonfibrotic ET and PV, cases with basophilia had a shortened OS and LFS, suggesting that basophilia and fibrosis may be independent prognostic factors. Moreover, we found that the influence of basophilia on patient outcome in PMF was independent of the MIPSS70+ ver2.0 score that accounts for clinical features, cytogenetic data, mutation profile, and reticulin grade. In summary, we found that basophilia is rare in non-fibrotic ET and PV patients at the time of diagnosis but was present in a significant subset of patients with pre-PMF, F-PMF, post-ET-MF, post-PV-MF, and MPN-U. Basophilia at the time of initial diagnosis predicted worse outcome in PMF patients that was independent of the MIPSS70+ ver2.0 prognostic scoring system. These results suggest that basophilia could represent an additional prognostic marker in MPN patients, by highlighting patients who may have shorter survival and higher risk of progression to blast phase than would be predicted by current risk modeling. All authors were up to date on research, ethics, compliance, and safety training (CITI Program) and protocols were approved by the local Institutional Review Board (IRB) before starting research studies. No written consent has been obtained from the patients as there is no patient identifiable data included. The authors declare no conflicts of interest. The data that support the findings of this study are available from the corresponding author upon reasonable request. 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.