CXCR4 expression on CD34+ blood cells is a predictive marker of outcomes in patients with primary myelofibrosis

Previous studies by the authors have demonstrated that decreased CXCR4 surface expression on CD34⁺ cells (CD34/CXCR4-se) is associated with mobilization of these cells in myeloproliferative neoplasms (MPN)-associated myelofibrosis. Furthermore, mobilization of CD34⁺ blood cells is correlated with increased disease activity and is considered a hallmark of primary myelofibrosis (PMF) along with splenomegaly and bone marrow fibrosis.      

To analyze the prognostic value and quantitative relevance of CD34/CXCR4-se, MPN Hub Steering Committee member, Giovanni Barosi, et al. published in Leukemia a cytofluorometric assessment from a large perspective study of patients with MPN.¹

Study design

Overall, 2,468 blood samples were analyzed for CD34/CXCR4-se from 1,284 patients with MPN who were referred to the Center for the Study of Myelofibrosis at Fondazione IRCCS Policlinico San Matteo, Pavia, IT. All participants were confirmed to have polycythemia vera (PV), essential thrombocytosis (ET), or MPN-associated myelofibrosis, based on the review of the initial bone marrow biopsy using the most recent World Health Organization criteria.

Additionally, blood samples were collected from 71 healthy individuals, age 20–80 years, as a reference point for normal for CD34/CXCR4-se values.

Results

Reduced CD34/CXCR4-se levels were observed in patients with MPN-associated myelofibrosis, who had a CD34/CXCR4-se level 1.7 times lower than the controls and 1.5 times lower than patients with PV or ET.

CD34/CXCR4-se < 39% was defined as an accurate discriminatory cut-off to distinguish MPN-associated myelofibrosis from ET or PV. It yielded a sensitivity and specificity for diagnosis of MPN-associated myelofibrosis of 54.3% (95% CI, 52.3–56.4) and 96.3% (95% CI, 92.1–98.6) with a positive predictive value of 99.5%.

There was a correlation between CD34/CXCR4-se and age. A total of 63.8% of patients aged > 60 years (n = 350) had a CD34/CXCR4-se proportion < 39% vs 49.5% of patients aged ≤ 60 years (n = 850; r = –0.159; p < 0.001).

Low CD34/CXCR4-se was shown to be correlated with lower hemoglobin and platelet concentrations, increased spleen size, percentages of blood blasts, and CD34⁺ blood cells in patients with PMF. Table 1 shows the analysis of the correlation coefficients of CD34/CXCR4-se with the listed covariates.

Table 1. Correlation between the frequency of CD34/CXCR4-se and hematological parameters in patients with primary myelofibrosis (n = 1,200)¹

Hematological parameter	r	p
Hemoglobin concentration, g/dL	0.29	< 0.001
White blood cell count, × 109/L (all measurements)	–0.04	0.166
White blood cell count, × 109/L (measurements with values < 10 × 109/L)	0.09	0.004
Spleen index, cm2	–0.25	< 0.001
Liver size, cm from the costal margin	–0.22	< 0.001
Blood monocyte count, × 109/L	–0.056	0.016
CD34+ blood cells, × 106/L	–0.08	< 0.001
Serum cholesterol level, mg/dL	0.189	< 0.001
Lactic dehydrogenase (times × upper limit of normal)	–0.056	0.107
Blood blasts, %	–0.32	< 0.001
Blood immature myeloid cells, %	–0.364	< 0.001
Blood basophils, %	–0.158	< 0.001

There was a strong correlation between CD34/CXCR4-se and degree of bone marrow fibrosis seen in 389 patients. In total, 23.7% of patients with a bone marrow fibrosis Grade 0 (n = 80) had a CD34/ CXCR4-se < 39% in comparison with 46.5% of patients with Grade 1 fibrosis (n = 71; p = 0.004), 62.2% with Grade 2 fibrosis (n = 119; p < 0.001), and 78.9% with Grade 3 fibrosis (n = 119; p < 0.001)

Out of the 84 patients that were analyzed for additional myeloid mutations, 14.3% had an ASXL1 mutation and 7.1% had an EZH2 mutation. Patients with less than one mutation had low CD34/CXCR4-se in the blood (31% ± 22%). The frequency of CD34/CXCR4-se < 39% was 71% in patients that had additional myeloid mutations compared with 43% in controls (p = 0.016).

A 67-month follow-up analysis was completed to determine if there was a correlation between CD34/CXCR4-se and disease evolution in patients with PMF whose CD34/CXCR4-se value was determined at diagnosis (n = 297). The authors confirmed that low CD34/CXCR4-se was strongly correlated with development of severe anemia, thrombocytopenia, leukopenia, increased levels of blood CD34⁺ cells, larger spleen size, and severe bone marrow fibrosis in patients with PMF. The outcomes of these patients can be seen in Table 2.

Table 2. Hazard ratio of the outcomes of patients with PMF analyzed at diagnosis and stratified according to decreased (< 39%) value and CD34/CXCR4-se frequency¹

HR, hazard ratio Bold indicates statistical significance.
Outcome of patients (n = 297)	HR (95% CI)	p
Severe anemia (hemoglobin > 100 g/L)	2.48 (1.61–3.81)	< 0.001
Large splenomegaly (spleen > 10 cm from the left costal margin)	2.33 (1.45–3.74)	< 0.001
Leukocytosis (white blood count > 12 × 109/L)	1.52 (0.97–2.36)	0.060
Leukopenia (white blood count < 4 × 109/L)	2.28 (1.20–4.32)	0.011
Thrombocytopenia (platelet count < 150 × 109/L)	2.99 (1.80–4.95)	< 0.001
Elevated CD34+ blood cells (> 100 × 106/L)	7.89 (3.73–16.73)	< 0.001
Death for any cause	3.80 (1.93–7.50)	< 0.001
Blast transformation	5.53 (2.28–13.40)	< 0.001

The interval from diagnosis to disease progression was significantly shorter in subjects with CD34/CXCR4-se < 39% measured at diagnosis in patients with PMF (n = 1200; r = –0.075; p = 0.009).

The authors built a prognostic model of survival in patients with PMF using CD34/CXCR4-se < 39%, age > 65 years, hemoglobin < 100 g/L, and CD34⁺ blood cell concentration > 50 × 10⁶/L.  This new risk model has only three risk classifications: low risk (median survival not reached), intermediate risk (median survival 15.8 years), and high risk (median survival 3.5 years). The new scoring system had a higher Akaike Information Criterion of 515 compared with 472 for the International Prognostic Scoring System (IPSS), indicating that the new score model is more accurate and informative. The new scoring system displayed good agreement with IPSS in defining low-risk patients; 90.6% of patients in the new scoring system that were classified as low-risk were also low-risk according to the IPSSscore. However, only 52% of patients in the new scoring system that were classified as high-risk were also IPSS high-risk. For patients classified as the new intermediate-risk cohort, only 52% were also assigned to the IPSS intermediate-1 or intermediate-2 IPSS categories, with the remainder of the group being IPSS low-risk (36%) and IPSS high-risk (12%).

Conclusion

In conclusion, the authors showed that CD34/CXCR4-se < 39% can be used as a new potential diagnostic and prognostic biomarker in patients with PMF. Furthermore, based on the CD34/CXCR4-se < 39% cut-off, a new risk scoring model was developed which predicts survival in patients with PMF more accurately than the IPSS scoring system. The limitations of this study include the lack of a validation cohort, and its single-center nature, as the chosen CD34/CXCR4-se cut-off may not apply to other test centers.