Abstract Introduction: Early molecular response (EMR, BCR-ABL (IS) ≤ 10% at 3 months) is a strong predictor of outcome in imatinib-treated chronic phase chronic myeloid leukemia (CP-CML) patients, but for patients who transform early 3 months may be too late for effective therapeutic intervention. Thus, alternative approaches are required to identify poor responders at the time of diagnosis. The aim of this study was to identify plasma biomarkers at diagnosis that will predict for subsequent EMR failure, early transformation or the development of BCR-ABL1 kinase domain mutations. Cytokine profiling has proven valuable in identifying prognostic factors in myelofibrosis and myelodysplastic syndromes; however, similar comprehensive studies are lacking to date in CML. Methods: Plasma samples from CP-CML patients enrolled to the TIDEL II trial were collected prior to starting imatinib treatment (n=186) and after 6 months on TKI (n=17); and compared to those of healthy donors (n=19). The levels of 39 cytokines, chemokines and growth factors (CC&GF) were measured using a Luminex multiplex assay. To identify potential biomarkers to predict EMR failure, random forest analysis and recursive partitioning techniques in R were applied as statistical methods. Results: Plasma concentrations of 13/39 CC&GF were significantly elevated at CP-CML diagnosis compared to healthy donor samples. Most (EGF, bFGF, VEGF, TGF-α, CXCL1, CCL4, sCD40L and IL-4) werenormalized after 6 months of TKI treatment while others (TNF-α, sIL-2Ra, IL-8, IL-10, IL-1a) remained at higher levels, possibly reflecting persistent disease-induced alterations within the microenvironment. A third subset of CC&GF, such as CCL2, CCL3 and CCL22, showed higher circulation levels only in TKI-treated patients but not at diagnosis, suggesting that changes in these CC&GF could be treatment-related. 183/186 patients had BCR-ABL1 assessments available at 3 months, and 23/183 (13%) did not achieve EMR. Random forest analysis identified TGF-α, IL-6 and IFN-α as the most important CC&GF associated with EMR failure. Recursive partitioning incorporating these three variables produced a classification tree based only on TGF-α and IL-6, and demonstrated that 12/20 (60%) of patients who were TGF-αhi/IL-6hi failed to achieve EMR (Table 1). Importantly, this group contained 3/3 (100%) patients who transformed within the first 12 months of TKI treatment. Both TGF-α (7.99 vs. 60.57 pg/ml, p<0.001) and IL-6 (0.26 vs. 8.35 pg/ml, p=0.004) were increased in plasma samples of EMR failure patients compared to those who achieved EMR, and this was independent of white cell count (both for TGF-α and IL-6) and Sokal Score (TGF-α only). While less compelling, high TGF-α alone was associated with EMR failure in 15/39 (38%) patients compared to 8/144 (6%) in the TGF-αlo group (p<0.001) and reduced transformation-free survival (92% vs. 99%, p=0.029). Interestingly, the TGF-αlo group also had higher rates of complete molecular response (38% vs. 8%, p=0.005). While the relevance of IL-6 in CML pathogenesis has been previously proposed the association of the EGFR ligand TGF-α with CML treatment outcomes is novel. Conclusion: These data highlight for the first time the prognostic value of cytokine profiling in CML patients. The combination of TGF-α and IL-6 plasma levels at CML diagnosis is strongly predictive for EMR, transformation and CMR. While TGF-α alone delineates a poor response group of patients, the addition of IL-6 provides significant additional power and strongly selects for high-risk patients who may benefit significantly from very early proactive intervention. Thus, incorporation of these simple measurements to the diagnostic work-up of CP-CML patients may enable therapy intensity to be individualized according to the risk profile of the patient. Abstract 1788. Table 1: Treatment outcomes in patients grouped according to predictive plasma biomarkers N(of 183 total patients) EMR failure Trans-formation Mutation MMR by 24 months CMR by 24 months + TGF-αhi/IL-6hi 20 12 (60%) 3# 5 9 (45%) 1 (5%) TGF-αhi/IL-6lo 19 3 (16%) 0 0 12 (63%) 2 (11%) TGF-αlo 144 8 (6%) 2## 6 117 (81%) 55 (38%) Overall p value (log-rank) < 0.001* <0.001 <0.001 0.004 0.004 MMR – major molecular response CMR – complete molecular response (undetectable BCR-ABL1) * p value is only indicative as the groups were formed based on EMR # all early transformations (< 1 year of TKI treatment) ## all late transformations Disclosures Reynolds: Novartis: Shareholder Other. Biondo:CSL Limited: Employment. Busfield:CSL Limited: Employment. Vairo:CSL Limited: Employment. Yeung:Novartis: Honoraria, Membership on an entity's Board of Directors or advisory committees, Research Funding; BMS: Honoraria, Membership on an entity's Board of Directors or advisory committees, Research Funding. White:Novartis: Honoraria, Research Funding; BMS: Honoraria, Research Funding; Ariad: Research Funding. Hughes:Novartis: Honoraria, Membership on an entity's Board of Directors or advisory committees, Research Funding; BMS: Honoraria, Membership on an entity's Board of Directors or advisory committees, Research Funding; Ariad: Honoraria, Membership on an entity's Board of Directors or advisory committees; Pfizer: Membership on an entity's Board of Directors or advisory committees.
Read full abstract