The CXCR4-CXCL12 axis is involved in many different types of cancer (Teicher & Fricker, 2010). Regarding haematological malignancies, CXCR4 is best studied in patients with acute myeloid leukaemia (AML). In this group of patients, overexpression of CXCR4 is an independent predictor of a poor prognosis (Rombouts et al, 2004; Dommange et al, 2006; Spoo et al, 2007; Tavernier-Tardy et al, 2009). Little is known regarding CXCR4 expression in BCP-ALL patients. Crazzolara et al (2001) demonstrated that paediatric B-ALL cases with high CXCR4 protein expression had significantly more prominent leukaemic cell infiltration of the liver or spleen compared with cases that had low CXCR4 expression. However, Schneider et al (2002) analysed a small cohort of paediatric BCP-ALL patients and could not confirm these observations, although they did report a shortened disease-free survival in patients with high CXCR4 expression. Our data supports the conclusion that high CXCR4 protein expression is correlated with an increased risk of relapse and poor outcome in paediatric BCP-ALL patients. In addition, we showed that this is independent of WBC count, age at diagnosis, cytogenetic subtype and sample source. Interestingly, CXCR4 expression levels were found to be significantly higher in samples isolated from the PB compartment compared to samples isolated from the BM compartment (Table SI and Figure S1). This might be explained by the fact that BCP-ALL cells localized in the BM compartment no longer need CXCR4 and/or that CXCR4 is internalized upon CXCL12 binding.
Thus far, reports regarding CXCL12 serum levels in leukaemic patients are limited. Khandany et al (2012) reported that CXCL12 serum levels were significantly increased in adult ALL patients compared with controls. On the other hand, Colmone et al (2008) found decreased CXCL12 levels in the serum of NALM6-engrafted (SCID) mice (Colmone et al, 2008). In the present study, we found that, in ALL patients, CXCL12 serum levels were significantly lower at diagnosis compared to remission after chemotherapy (day 79) and non-leukaemic control patients. This lower serum level could not be explained by consumption of CXCL12 by the leukaemic cells or by an altered production capacity of MSC at the time of diagnosis. In contrast, we observed that leukaemic cells negatively affected CXCL12 production by the MSC. We checked whether the variance in secreted levels correlated to MRD levels at the end of induction (ALL in remission). However, MRD status did not correlate with CXCL12 expression levels in these patients (Figure S4). Of interest, we observed that a drop in CXCL12 production coincided with an elevated level of G-CSF when leukaemic cell numbers increased. G-CSF has been shown to mobilize HSC and myeloid cells from the BM into the blood compartment by down-regulating CXCR4 expression and attenuating their responsiveness to CXCL12 (Kim et al, 2006). Hypothetically, the increase in G-CSF and concomitant drop in CXCL12 production observed in the present study may trigger the release of lymphatic leukaemic cells from a crowded BM into the blood compartment and may foster the spreading of the disease. These circulating leukaemic cells can be targeted with traditional chemotherapeutic drugs. Adult patients with relapsed or refractory AML treated with Plerixafor in combination with low-dose cytarabine, aclarubicin and G-CSF had higher response rates than those without Plerixafor (Saito et al, 2000; Lowenberg et al, 2003; Qian et al, 2007; Wei et al, 2011). Interestingly, combination of G-CSF with Plerixafor increases the percentage of persons that mobilize sufficient stem cells for transplantation compared to G-CSF alone (Anonymous, 2007). Plerixafor is now approved by the US Food and Drug Administration in combination with G-CSF to mobilize HSC to the PB for collection and subsequent autologous transplantation in patients with non-Hodgkin lymphoma and multiple myeloma (DiPersio et al, 2009a,b). We observed that CXCL12 levels were low at the time of diagnosis, when leukaemic cells were numerous in the bone marrow, and increased upon chemotherapy treatment to levels seen in non-leukaemic controls. This, together with the observation that these lower levels seem to be caused by leukaemic cells, suggests that interference with the CXCR4/CXCL12 axis may be an effective way to mobilize ALL cells; the more ALL cells become mobilized, the less ALL cells may escape from combination chemotherapy. Promising results have already been obtained in a paediatric BCP-ALL mouse model, in which NOD/SCID mice were transplanted with newly diagnosed primary patients cells. Prolonged exposure to small molecule CXCR4 antagonists (including Plerixafor) reduced the number of leukaemic cells in the spleen and PB, and significantly reduced their dissemination to extramedullary sites (Juarez et al, 2007; Welschinger et al, 2013). As a next step, clinical trials using AMD3100 (in combination with other mobilizing agents such as G-CSF) to sensitize BCP-ALL cells to chemotherapy need to be conducted to determine whether such a strategy could improve responses to chemotherapy and patient outcomes.