To the Editor

Interferon regulatory factor 4 (IRF4) is a transcription factor with an established role in the pathogenesis of various hematologic malignancies [1]. IRF4 expression has been related to the negative regulation of myeloid-derived suppressor cells (MDSCs), thereby altering immunosurveillance; moreover, its expression drives inflammation through the polarization of anti-inflammatory M2 macrophages [2, 3]. In chronic myeloid leukemia a downregulated expression of IRF4 has been found [4] and recent evidence supports its prognostic role in JAK2V617F mutated myeloproliferative neoplasms (MPNs) [5].

With these premises, we evaluated IRF4 expression in 119 Philadelphia negative MPNs (Ph- MPNs) patients (Additional file 2: Table S1) to verify its role on clinical outcome (median follow-up: 61.5 months, range: 1–238). The quantification was calculated as the ratio between IRF4 and GUSB number of copies (I/G) (Additional file 1).

The bone marrow (BM) IRF4 median value was 0.11 I/G (min. 0.11- max. 012) and 0.04 I/G (min. 0.001 – max. 0.27) in the healthy and in the MPNs groups, respectively (p < 0.0001). Considering the IRF4 median value for every MPN type, the difference compared to the healthy controls (HC) remained statistically significant (Fig. 1A). In particular, the primary myelofibrosis (PMF) patients showed a lower IRF4 median value than that of the other groups compared to the HC. Among the MPNs, the PMF IRF4 median value was lower than in the essential thrombocythemia (ET) group (p = 0.003). CALR mutated cases showed a higher IRF4 expression than those with JAK2 (0.06 vs 0.03, p = 0.007) or triple-negative (TN) (0.06 I/G vs 0.02 I/G, p = 0.0008) (Fig. 1B). IRF4 expression was not associated with variables as sex, age, risk group [6]. Fourteen (11.7%) patients showed leukemic transformation (LT): 8 PMF, 4 secondary myelofibrosis (SMF), and 2 polycythemia vera (PV); they had a lower IRF4 expression at diagnosis compared to the other MPN patients (0.01 I/G vs 0.04 I/G, p = 0.0005) (Fig. 1C). An optimal cutoff of the IRF4 expression value best identifying the possibility of MPN leukemic transformation was defined by ROC analysis. The area under the curve was 0.79 (95% CI 0.71–0.86; p < 0.0001). Representative cutoff values for sensitivity and specificity were calculated and plotted on the curve. An optimal value of 0.022 I/G was obtained, with a sensitivity of 76.9% (95% CI 46.2–95.0) and a specificity of 76.1% (95% CI 66.9–84.0). This value distinguished MPN patients with a higher probability of LT; in fact, the group with an IRF4 value < 0.022 I/G had shorter leukemia-free survival (LFS) (Fig. 2A). The LFS analysis was also considered only for the myelofibrosis (MF) group; patients with an IRF4 value < 0.022 I/G showed a shorter LFS (p = 0.001, Fig. 2B). In another LFS analysis, this difference was confirmed when considering only the PMF group: the median LFS for PMF patients with IRF4 < 0.022 I/G was 23 months, whereas LFS was not reached for the > 0.022 I/G group (p = 0.0009) (Fig. 2C). Overall survival (OS) analysis in the MF group showed that patients with IRF4 > 0.002 I/G had longer median survival than those in the < 0.022 I/G group (143.9 mo. versus 40.5 mo., p = 0.04) (Fig. 2D). Also, in the PMF group, the IRF4 value > 0.022 I/G was associated with a longer OS (75 mo versus 21 mo, p = 0.04) (Fig. 2E). Moreover, PV and ET patients with an IRF4 value < 0.022 I/G showed a shorter time to MF transformation (109 mo. versus 190 mo., p = 0.03) (Fig. 2F).

Fig. 1
figure 1

IRF4 expression in Ph-MPNs. A Comparison of IRF4 expression between healthy controls and MPN groups. The IRF4 expression was statistically significantly lower in every MPNs category than in healthy controls. B IRF4 expression according to the gene driver mutation in MPNs. A lower IRF4 expression was more frequently concomitant with JAK2 + and triple-negative status than with the CALR gene mutation MPN group. C The MPN patients group with leukemic transformation showed a lower IRF4 expression than the group without leukemic evolution. Boxplots representing the distribution. The box always extends from the 25th to 75th percentiles. The line in the box represents the median. The whiskers go down to the smallest value and up to the largest

Full size image
Fig. 2
figure 2

IRF4 prognostic impact in Ph-MPNs. A LFS analysis in all MPN patients included in this study according to the IRF4 expression value. B LFS analysis in MF patients according to the IRF4 expression value. C LFS analysis in PMF patients according to the IRF4 expression value. D OS analysis in MF patients according to the IRF4 expression value. E OS analysis in PMF patients according to the IRF4 expression value. F Probability analysis of evolution to post-PV and post-ET MF according to the IRF4 expression value. The cut off of the IRF4 expression (0.022) was calculated by ROC analysis

Full size image

Among the 14 patients with LT, 9 (64.2%), 7 PMF and 2 SMF, were analyzed in next-generation sequencing [7] to detect the high molecular risk (HMR) mutations (Additional file 3: Table S2). In 5 PMF patients, HMR mutations were found; moreover, all 9 patients showed at least one additional genetic lesion besides the gene driver mutation (Additional file 4: Figure S1). Despite the data paucity, just under half of the patients with MF who had undergone LT (44.4%) did not have HRM mutations. In three PMF cases, the IRF4 expression was evaluated during ruxolitinib treatment. All cases exhibited an increased IRF4 expression compared to the value at diagnosis (p = 0.003); in the two patients with a longer duration of ruxolitinib treatment (59 mo. and 72 mo., respectively), there was a one-log increment of IRF4 expression (Additional file 5: Figure S2).

Despite the main limits of our study (relatively small number of patients in each group, few NGS data, inability to determine the IRF4 production source) we demonstrate an IRF4 dysregulated expression in MPNs patients, particularly in PMF and in JAK2 + and TN + cases, distinguishing those with a higher probability of SMF. Furthermore, the IRF4 expression was associated with LT and a shorter LFS. Further studies are warranted to validate these data to confirm this biomarker as a new prognostic factor.