We reviewed 71 patients with Philadelphia chromosome-negative MPNs diagnosed at the Hakodate Municipal Hospital (Hakodate, Japan) between April, 2001 and April, 2014. The diagnosis was based on the World Health Organization criteria (10). Patients diagnosed at our department but followed up in other hospitals were excluded from the analysis. This study was approved by the Hakodate Municipal Hospital Institutional Review Board. All the patients provided written informed consent according to the Declaration of Helsinki.

Abnormal values of white blood cell (WBC) count, hemoglobin (Hb) concentration, hematocrit (Hct) and platelet (PLT) count were defined as higher than the upper limits of normal values (ULN), or lower than the lower limits of normal values (LLN). Abnormal erythrocytic lineages were reflected in abnormal Hb and/or Hct values. In addition to the number of abnormal blood cell lineages, one point each was assigned for age >65 years and splenomegaly. Each MPN patient was subsequently assigned a score of 1–6.

For the detection of the JAK2V617F mutation, allele-specific polymerase chain reaction analysis was performed as previously reported (11). We were unable to investigate other JAK2 mutations, as this is not covered by insurance in Japan. In our institute, detection of JAK2V617F mutation became accessible in 2013.

The baseline characteristics of the patients were described using median and range, and compared using the median test, χ² analysis and the Fisher's exact test, as appropriate. Overall survival (OS) was defined as the time between the first patient contact and death or the time of the last follow-up. The endpoint of event-free survival (EFS) was defined as transformation to AML, new appearance or aggravation of splenomegaly, addition of an abnormal karyotype, new complication of MF, increasing blood cell counts that required chemotherapy, thrombosis, or patient death from any cause. OS and EFS were evaluated by the Kaplan-Meier method and Mantel-Cox log-rank tests. For univariate and multivariate analysis, logistic regression was used to investigate the association of baseline characteristics with OS and EFS. Statistical analysis was performed with StatMate V software (ATMS Co., Ltd., Tokyo, Japan). P<0.05 was considered to indicate statistically significant differences.

Table I provides a comparative presentation of the clinical and laboratory characteristics of each type of MPN [PV, n=17; ET, n=34; PMF, n=16; and MPN, unclassifiable (u), n=4]. Significant differences were observed in Hb, Hct, PLT count and splenomegaly, but not in age, gender distribution, WBC count, abnormal karyotype, or frequency of the JAKV617F mutation.

The causes of death and endpoints of EFS are presented in Table II. A total of 3 patients succumbed due to transformation to AML, 2 patients due to renal failure and 1 patient each due to pneumonia, hemophagocytic syndrome, uncontrollable bleeding due to thrombocytopenia, malignant pleural effusion, stroke, hepatic failure due to hepatitis B virus infection and unexpected cardiopulmonary arrest. All these patients had multilineage abnormalities in blood counts. The numbers of events were as follows: Increasing blood cell counts and chemotherapy, n=10; appearance or aggravation of splenomegaly, n=6; complications of MF, n=3; transformation to AML, n=3; and death, n=4 (2 from renal failure and 1 each from stroke and hepatic failure). Addition of an abnormal karyotype was observed in 1 patient concurrently with complications of MF. Thrombosis was not observed in this analysis.

The median follow-up duration for all MPN patients was 37 months (range, 2–151 months). No significant differences were observed among different types of MPN (P=0.10).

The survival curves for each type of MPN (excluding MPN, u) are presented in Fig. 1. The median OS of PV, ET and PMF was 43 (range, 12–100), 38 (range, 2–151) and 41 (range, 4–105) months, respectively. The respective median EFSs were 31 (12–97), 30 (2–112) and 22 (4–100) months. As reported, the OS of PMF was worse compared with that of PV and ET (P=0.0117). Similarly, EFS was significantly difference among the different types of MPN (P=0.0210).

The prognostic factors for OS and EFS on the univariate and multivariate analysis are summarized in Tables III and IV. On univariate analysis, age (P=0.043), gender (P=0.036), number of abnormal cell count lineages (P=0.020) and splenomegaly (P=0.0055) were found to be significant prognostic factors for OS. However, for EFS, only the number of abnormal cell lineages (P=0.0065) was identified as a significant prognostic factor. Our results suggest that abnormal karyotype and JAK2V167F mutation did not significantly affect survival.

On multivariate analysis, a significantly worse overall survival was observed for patients who were elderly (P=0.042) and had splenomegaly (P=0.016). A significantly worse EFS was observed with higher numbers of abnormal cell lineages (P=0.014).

According to the previously described risk scoring system based on age, abnormal blood cell lineages and splenomegaly, each patient was assigned a sum score of 1–5 adverse points. The five patient groups were then consolidated into three risk groups based on the margin of intergroup survival differences (Fig. 2) as follows: i) Score 1–2 (n=23; median OS, 45 months; median EFS, 38 months); ii) score 3 (n=24; median OS, 39 months; median EFS, 29.5 months); and iii) score 4–5 (n=24; median OS, 32 months; median EFS, 26.5 months). Significant differences were observed among these three groups in terms of OS (P=0.011) and EFS (P=0.0059).

As illustrated in Fig. 1, PMF was associated with a worse prognosis compared with PV and ET (17). Thus, we created two categories of PV-ET analysis (Fig. 3) and PMF analysis (Fig. 4). All the PV/ET patients belonging to the score 1–2 and 3 groups remained alive during the entire duration of the follow-up, with a median survival of 46 months for the score 1–2 and 29.5 months for the score 3 group. However, 3 patients succumbed to the disease in the score 4–5 group (median survival, 34 months); this group had a significantly poorer prognosis (P=0.0053). The respective median EFSs of the 3 groups were 38.5, 29 and 30 months. Similarly, the EFS of the score 4–5 group was significantly worse compared with that of the other groups (P=0.022).

The OS and EFS curves of PMF patients are presented in Fig. 4. Only 2 patients had a score of 1–2, 6 patients had a score of 3, and 8 patients scored 4–5. No significant differences were observed within each group (OS, P=0.177; EFS, P=0.520). However, our results suggested that PMF patients with a score of 4–5 may have a worse outcome compared with the lower scoring groups (OS, P=0.023 and EFS, P=0.142 in a statistical analysis of the score 3 vs. the score 4–5 group).

A total of 47 patients were tested for the presence of the JAK2V617F mutation. The survival curves are presented in Fig. 5. We observed no statistical association of OS and EFS with the JAK2V617F mutation in MPN patients (OS, P=0.281; EFS, P=0.174).