Introduction

Essential thrombocythemia (ET) and polycythemia vera (PV), classified under Philadelphia chromosome-negative myeloproliferative neoplasms (MPNs), represent unique clinical entities with excessive production of platelets and/or erythroid cells [1]. Despite their generally benign nature, both conditions carry a predisposition for thrombotic events, which influence patient prognosis. Within the framework of MPN management, stratified approaches to thrombosis prevention are instrumental and underscore the need for robust risk assessment tools.

While international criteria such as the European LeukemiaNet (ELN) criteria [2], International Prognostic Score of Thrombosis for Essential Thrombocythemia (IPSET-thrombosis) [3], and revised-IPSET [4] have been instrumental in risk stratification for ET, the ELN criteria [5] and the criteria proposed by Tefferi et al. [6] have been useful for PV, and their application has been predominantly validated in Western populations. The JSH-MPN-R18 trial, a nationwide retrospective analysis conducted in Japan, extended these findings to Asian populations and revealed potential cross-racial applicability [7, 8].

In the evolving landscape of thrombotic risk assessment for MPNs, leukocytosis [9,10,11,12,13,14,15] and neutrophil-to-lymphocyte ratio (NLR) [16, 17] have emerged as promising markers of thrombotic events and mortality. However, extrapolation of these findings to a Japanese cohort remains unexplored. This study aimed to bridge this knowledge gap by leveraging data from the JSH-MPN-R18 trial to ascertain the prognostic value of the neutrophil ratio in thrombosis prediction within a Japanese MPN population.

Methods

Study design

This multicenter retrospective analysis was performed under the auspices of the Japanese Society of Hematology (JSH), to examine the clinical profiles of Japanese patients diagnosed with PV or ET. Participants were part of the JSH-MPN-R18 cohort, aged 20 years or older, and received a diagnosis of PV or ET according to the WHO 2008 or 2017 criteria [1, 18] during the period from April 2005 to March 2018. Details of the inclusion and exclusion criteria have been described previously [7]. This retrospective analysis involving human participants was in accordance with the ethical standards of the institutional and national research committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. Ethical approval was granted by the Mie University Hospital Ethical Committee (approval number H2022-165).

Data collection

Diagnostic laboratory data of the enrolled patients were collected retrospectively. Medical records included date of diagnosis, age at diagnosis, sex, presence of driver gene mutations, laboratory parameters (white blood cell [WBC] counts, neutrophil ratio [rNeu] to WBC, absolute number of neutrophils [aNeu], red blood cell [RBC] counts, haemoglobin [Hb], haematocrit [Hct], and platelet counts), incidence of post-diagnostic thrombosis, and mortality.

Outcome measures

The primary outcome measure was overall incidence of thrombotic events. Thrombosis-free survival was calculated from the time of diagnosis to the first event of each type. Patients without adverse events were censored at the last follow-up visit.

Statistical analysis

Univariable and multivariable analyses using Cox proportional hazards regression models were conducted to identify risk factors for thrombotic events. Based on previous studies [2,3,4,5,6,7,8], we included WBC, rNeu, RBC, Hb, Hct, and platelet counts in the models. Because of the high collinearity resulting from the similarity between rNeu and aNeu, a multivariate analysis was conducted using rNeu, excluding aNeu. Subjects with missing data for the variables included in the models were excluded from the analysis.

Fine–Gray competing risk model [19] was used to calculate cumulative incidences of thrombosis. All deaths were considered for the cumulative incidence in Fine–Gray analysis. Identified predictors for thrombosis were tested using receiver operating characteristic (ROC) curve analysis, and area under ROC curve values were determined along with their 95% confidence intervals. Decision tree analysis was performed using R package (rpart).

All analyses were two-sided, with a p-value of < 0.05 considered statistically significant. Statistical analyses were conducted using R software (version 4.2).

Results

In this nationwide retrospective analysis of JSH-MPN-R18 results, we included 1152 patients with ET (Table 1) and 596 patients with PV (Online Resource 1). Regarding PV diagnosis, the current cohort included data from before JAK2 mutation tests became available in Japan, and some cases lacked information on the JAK2 mutation, as observed in raw data of PV cases (Online Resource 1). Given the high number of patients who did not harbour JAK2 mutations, we excluded cases where the JAK2 mutation was negative or unknown from further analysis. Finally, a total of 456 cases with PV were evaluated in the following study, including 446 and 10 patients with JAK2 V617F and JAK2 exon12 mutation, respectively.

Table 1 Patients’ characteristics
Full size table

Table 1 outlines the baseline characteristics, laboratory data, and mutation status. Among the patients, 74 and 22 cases of ET and PV, respectively, developed thrombosis, as shown in Table 2. To evaluate the utility of clinical items against future thrombosis, we performed a multivariate analysis. For patients with ET, rNeu was a significant factor associated with thrombosis (Table 3). Besides, decision tree analysis identified rNeu as the most important factor for future thrombosis (see figure, Online Resource 1), and hematocrit was found to be a significant predictive factor for PV (data not shown) in decision tree analysis.

Table 2 Characteristics of thrombotic and non-thrombotic cases in ET and PV
Full size table
Table 3 Results of multivariate analysis for ET and PV
Full size table

For patients with ET, we defined the rNeu cut-off value as 80% (predicted by decision tree analysis; see figure, Online Resource 2). Fine–Gray cumulative incidence analysis for patients stratified by rNeu revealed that those with high rNeu were more susceptible to future thrombosis (Fig. 1A). We also analyzed the role of aNeu in thrombosis, and the ROC analysis identified an aNeu cutoff value of 5.0 × 109 /L and revealed the potential role of aNeu in thrombosis (Fig. 1B). Interestingly, the importance of aNeu was emphasized in the lower thrombotic risk group (including low/very low in revised-IPSET) (Fig. 2).

Fig. 1
figure 1

Thrombosis-free survival using Fine–Gray analysis The cumulative incidence of thrombosis from the time of essential thrombocythemia (ET) diagnosis was calculated for patients with neutrophilia and those without neutrophilia. The definition of neutrophilia was based on the neutrophil ratio (cut-off value, 80%) and absolute neutrophil count (cut-off value, 5000/µL). The cumulative incidence was significantly lower (Fine–Gray test, p = 0.0152 and p = 0.00073, respectively) for ET patients with neutrophilia than for those without neutrophilia

Full size image
Fig. 2
figure 2

Thrombosis-free survival according to the revised International Prognostic Score of Thrombosis for Essential Thrombocythemia (rIPSET). The cumulative incidence of thrombosis was calculated for patients with essential thrombocythemia (ET) with high rIPSET and those with low rIPSET. Fine–Gray analysis revealed the significance of aNEU for ET patients with low rIPSET, but not for those with high rIPSET

Full size image

As there was a significantly higher incidence of patients with JAK2 mutations (Chi-square test, p-value = 5.22 × 10− 14) in the group with rNeu values of ≥ 80%, we further investigated whether the predictive value of rNeu could be applied to all driver mutations. The multivariate analysis identified rNeu as a predictive tool for thrombosis only in JAK2 mutated ETs but not in CALR/MPL mutated ETs (Table 3).

Next, we performed a similar analysis in patients with PV. The multivariate analysis revealed that Hct was a predictive factor for thrombosis, as predicted. Previous reports [12, 20] have also highlighted that a higher WBC count can be a predictive tool for thrombosis; however, there has been little evidence of the predictive value of aNeu and rNeu for thrombosis in our dataset. ROC analysis of WBC against thrombosis revealed a cut-off value of 7.05 × 109/L (Fig. 3A). The Fine–Gray method revealed that thrombosis rates for JAK2-positive PV with aNeu ≥ 7.05 × 109/L were higher, although the difference was not statistically significant (Fig. 3B).

Fig. 3
figure 3

Prognostic utility of the absolute neutrophil count for thrombosis

Receiver operating characteristic (ROC) curve analysis was performed for the neutrophil ratio (rNeu) against future thrombosis for patients with polycythemia vera (PV) (A). The cumulative incidence of thrombosis from the time of PV diagnosis in patients with neutrophilia (aNeu ≥ 7000) and those with lower neutrophil counts was calculated (B). The incidence of thrombosis was significantly lower (Fine–Gray test, p = 0.0295) for patients with PV and neutrophilia than for those without neutrophilia

Full size image

Discussion

This analysis of the JSH-MPN-R18 dataset underscores the prognostic significance of elevated neutrophil levels in predicting thrombotic events in patients with MPNs. Similar to previous studies that emphasized the NLR as a risk marker, our study revealed that increased rNeu and aNeu were associated with a greater risk of thrombosis.

The biological mechanisms linking heightened neutrophil activity to thrombotic complications remain unclear. However, the observed correlation was particularly strong in ET patients with JAK2 mutations; this implied that JAK2 mutations are associated with thrombosis. Previous studies have identified the ERK pathway as JAK2-specific [21] and characteristic of neutrophil activation [22, 23]. Furthermore, as JAK2 V617F allelic burden in MPNs correlates with leukocyte counts, higher rNeu might reflect higher JAK2 signaling activity [23, 24].

Our study has certain limitations, notably the absence of lymphocyte counts, which prevents a complete evaluation of the predictive value of NLR or rNeu for thrombosis. In addition, we did not have molecular background or allelic frequency information of the driver mutations. Registration of patients with MPNs began in 2005, and continuous follow-up for therapeutic interventions was not included. Because of incomplete data for some cases, we could not uniformly apply the latest WHO classification for disease definition. These limitations indicate that our study is not comprehensive, and future validation is warranted. To overcome these limitations, we are planning a novel cohort study in Japan. Nonetheless, our data showed that simple rNeu can identify ET patients with a high risk of thrombosis. Further data collection on these risk factors is required in future studies.

In conclusion, our study underscores the significance of rNeu, aNeu, and JAK2 mutations in predicting thrombosis in Japanese patients with ET. Further accumulation of evidence and the development of new therapeutic strategies tailored to the unique characteristics of Japanese patients are imperative. Preventive interventions that consider the specific genetic and clinical traits of Japanese patients may play a key role in mitigating the thrombotic risks associated with MPNs.