In this paper we show that CML cells have a low level of SPARC and there is also a reduction of the secreted form of SPARC in the serum of CML patients. Once treatment with IM was initiated, we observed an increase of SPARC mRNA and protein in the PBMCs, reaching levels that are much higher than normal controls. This increase was evident already at 3 months and was maintained at least for the 18 months of observation. Accordingly, the concentration of SPARC in the serum showed a progressive increase during IM treatment and reached normal values at 18 months. In fact, IM is a very effective treatment for CML. At 3 months, most patients are in complete hematological response and at 12 months most of them are in complete cytogenetic or even molecular response. In these conditions, CML cells are virtually absent in the PBMCs of CML patients and, however, the contribution of these cells to the PBMC composition is negligible. Therefore, the measured SPARC mRNA after 3 months of treatment and throughout the subsequent evaluations was produced by normal cells, i.e. lymphoid cells, monocytes and non-clonal granulocytes. This production is probably favored by IM since we observed that, in one patient, discontinuation of IM therapy resulted in a reduction of the SPARC level while it rose again after restarting treatment. The only two patients treated with an alternation of IM and NI, showed the same pattern as patients treated with IM alone, thus favoring the hypothesis that the increase of SPARC is a common phenomenon after any TKI treatment.
We show that SPARC is mainly expressed by monocytes with respect to the total number of analyzed cells but, during IM treatment, other cell populations such as normal lymphocytes and granulocytes increase their SPARC production.
We also demonstrated that SPARC may synergize in vitro with IM by blocking the cell cycle of CML cells in the G0/G1 phase. We therefore hypothesize that the increased production of SPARC by normal cells may contribute to the efficacy of IM in reducing CML clone. In this perspective, studies are ongoing in our lab in order to evaluate if the levels of SPARC that can be achieved after treatment may be correlated with the magnitude of the response to IM.
The study of SPARC in hematopoietic malignancies led to conflicting reports about its role as a tumor suppressor or promoter. A study by Li et al. supports an anti-proliferative effect of SPARC on leukemic cells. The authors reported that transfection of K562 cells with SATB1 plasmid induced SPARC over-expression, resulting in a reduction of cell proliferation . In contrast, a recent study  reports intracellular SPARC as a potential cause of resistance to IM in CML cells. However, the same authors pointed out that the cellular localization of SPARC could be determinant in predicting its activity. The discrepancy that exists in the literature on the activity of SPARC on different diseases is probably linked to the fact that there are different sources and different isoforms and even in the same disease SPARC may have different activities. In addition, the activity of SPARC may depend on whether it is secreted and associated with the extracellular matrix or retained inside the cell. In this perspective, if the intracellular form of SPARC may protect CML cells from IM induced cell death , its secreted form could even be toxic for the CML cells as it has been demonstrated not only by our experiments but also for some subtypes of acute myeloid leukaemia .
Initial studies showed that SPARC is important in bone mineralization . In this perspective, we  and others [22–25] have demonstrated that in CML patients treated with IM there is a transient increase in bone-formation markers that could be linked to inhibition of PDGFR signaling. We have also demonstrated that also NI induced osteoblastogenesis in vitro, probably triggering the same IM targets . Therefore the increase of SPARC induced by TKIs could be related to the well documented modification of the bone metabolism observed in IM treated patients.
In CML cells, the BCR/ABL oncoprotein has proliferative effects activating the Ras/Raf/MEK/ERK, JAK/STAT and PI3K/Akt pathways [27–30]. The autophosphorylation site Y177 on BCR/ABL binds the scaffolding adaptor protein Gab2  that activate both PI3K/Akt and Raf/MEK/ERK pathways . Inhibition of BCR/ABL by IM results in a G1 cell cycle arrest mediated by the PI3K pathway . One method that is showing success against IM-resistant CML cells is the treatment with inhibitors that target Ras or PI3K . Because SPARC affects multiple downstream signaling pathways, such as PI3K/AKT , in our study we focused on its potential role as a tumor suppressor protein in CML cells. Many studies have focused on the role of the protein in tumorigenesis, but few on its potential role in modulating therapy sensitivity . Here, we demonstrate that exogenous exposure to SPARC in K562 cells induced a G0/G1 cell cycle arrest reducing the growth rate of these cells and appeared to confer an increased sensitivity of K562 cells to IM. In addition, residual leukemia stem and progenitor cells persist in IM-responsive patients and may be a potential source of relapse. While integrins such as very late antigen-4 and the adhesion molecule CD44 have been demonstrated to be crucial for the persistence of leukemic progenitors in the bone marrow , SPARC is an anti-adhesive molecule and modulates the cell matrix ; its down-regulation may increase the adhesion of the hematopoietic stem cells to the supporting stromal cells and provide a clonal advantage . Therefore the SPARC/TKI association could serve to enhance the TKI effect on leukemic stem cells that are not eliminated by BCR/ABL inhibitors [37–39].
On the basis of 1) our demonstration that in vitro SPARC has an anti-proliferative effect on BCR/ABL positive cells and influences the sensitivity of leukemic cells to therapy and 2) the putative modulation of the tumor microenviroment by this protein, it is conceivable that the effects of exogenous SPARC could be exploited therapeutically by using recombinant SPARC or one of its derivate peptides. This possible approach is supported by experiments on xenograft tumor models where intraperitoneal or subcutaneous injections of SPARC inhibited tumor growth [33, 40].