Truncation Products of Stromal Cell Derived Factor-1 (CXCL12) Quantified By Mass Spectrometry in Patients with Myelodysplastic Syndrome (MDS) or Acute Myeloid Leukemia (AML) Treated with Rigosertib in a Phase I-II Study

Introduction: CXCL12 (stromal cell derived factor-1, SDF-1), an 8 kDa peptide chemokine, ligates chemokine receptor 4 (CXCR4) and activates migration of normal and leukemic stem cells from the bone marrow into the blood. Rigosertib is a synthetic benzyl styrene sulfone with evidence of activity in certain subsets of patients with MDS and AML (Garcia-Manero et al., Lancet Oncology 2016; 17; 496). Objective: To characterize and quantify intact CXCL12 and its protease-induced truncation products in plasma of MDS and AML patients before and after treatment with Rigosertib in Phase I/II dose escalation trials. Methods: Rigosertib was infused continuously for three days (dose: 650-1,700 mg/m2/d), and samples were obtained from patients with MDS (n=8) or AML (n=12). Plasma samples were taken at 0 and 72 h and centrifuged at 300 g for 10 min. After dilution with equal volume of water, plasma was ultrafiltered (30 kDa exclusion). Aliquots of the ultrafiltrates were analyzed by liquid chromatograph/mass spectrometry (LC/MS). LC: Tosoh C18 column (TSK gel, ODS-100V), gradient elution. MS: positive electrospray ionization (ESI) with selected ion monitoring, m/z 980 for intact CXCL12, m/z 952, 940, 929, and 914 for -2 amino acids (aa, KP removed), -3 aa (KPV removed), -4 aa (KPVS removed), and -5 aa (KPVSL removed) truncation products, respectively. The presence of truncates was confirmed using standards incubated with various proteases by determining molecular masses using ESI. Calibration curves for quantification were established using synthetic standards. Results & Discussion: In normal subjects (n=10), the concentration of intact CXCL12 was 16.6±9.4 ng/m, compared to 3.1± 3.2 ng/mL in MDS patients (n=8) and 2.4±1.7 ng/mL in AML patients (n=12) prior to therapy with Rigosertib. No truncation products (≤1.0 ng/mL) were detected in normal subjects, despite the relatively high concentration of intact CXCL12. In contrast, truncation products, corresponding to the removal of 2, 3, 4 and 5 amino acids, were detected in all patients prior to treatment: 2.5±1.6 ng/mL, 2.7±2.1 ng/mL, 2.4±1.7 ng/mL, 4.2± 3.6 ng/mL in MDS patients; 3.3±2.9 ng/mL, 2.4±1.7 ng/mL, 3.1±2.5 ng/mL, 2.2±1.7 ng/mL in AML patients. The total concentration of all truncation products was ~12 ng/mL in MDS patients and ~11ng/mL in AML patients. We have also reported CXCL12 truncation products in patients with primary myelofibrosis (~29 ng/mL) and polycythemia vera (~31 ng/mL) (SC, JR et al., Cancer Res. 70, 3402, 2010). Of the 20 AML and MDS patients treated with rigosertib, six attained partial or complete bone marrow remission, based on the International Working Group criteria. Among the responding patients, intact CXCL12 concentration increased from 3.4±3.6 ng/mL to 5.6±3.7 ng/mL. Among non-responders, intact CXCL12 underwent little change, from 2.7±2.2 to 2.9±2.3 ng/mL. Conclusion: Proteolytic degradation of CXCL12 may be characteristic of the pathobiology of homing and release from the marrow niche in patients with myeloid malignancies and this process changes in response to treatment. Our findings suggest that CXCL12 may be a biomarker for patients with MDS or AML who respond to rigosertib on clinical trials. Further investigation of the potential role of intact CXCL12 and its truncation products in plasma as a biomarker in these diseases is warranted.