PF424 DISTINCT CYTOKINE PROFILES CHARACTERISE CML PATIENTS WHO HAVE HIGHER RISK OF EMR FAILURE AND BLAST CRISIS PROGRESSION

Background: Tyrosine kinase inhibitors (TKIs), imatinib (IM), nilotinib (NIL) and dasatinib (DAS), are the frontline treatment in chronic myeloid leukaemia (CML). Failure to achieve early molecular response (EMR) at 3 months ( BCR‐ABL1 level >10%) is predictive of inferior clinical outcomes such as blast crisis (BC) progression. Identifying patients at high‐risk of EMR failure provides an opportunity to improve outcomes by interventions at diagnosis, as later intervention may be too late to prevent adverse outcomes. Aims: To identify CML patients who are at high‐risk of EMR failure and BC progression based on plasma cytokine levels at diagnosis. Methods: 347 CML patients (9% EMR failure), who received IM, NIL, or DAS treatment had plasma collected at diagnosis. Luminex‐multiplex‐assay was used to quantify the level of 38 cytokines, chemokines and growth factors. The expression of each cytokines was log2 transformed and subjected to batch effect removal using Combat algorithm implemented in sva Bioconductor package. Heatmap was generated using pheatmap Bioconductor package with unsupervised clustering performed using Euclidean distance with ward.D linkage. Results: The cytokine data was filtered with low variance across samples (<2 in log2 scale) leaving 25 significant cytokines. Unsupervised clustering identified 8 clusters with distinct cytokine expression patterns (Figure 1A). Among these 8 clusters, cluster 6 and cluster 7 were identified with higher risk of EMR failure (24% and 50% respectively), compared to the remaining clusters with EMR failure rate all lower than 10% (Figure 1B). Interestingly, cluster 6 demonstrated high expression of a few cytokines such as TGFabut not most of the other cytokines, while cluster 7 demonstrated high expression of all of the cytokines except for GRO, sCD40L and EGF. This suggested the heterogeneity of cytokine expression for these patients who failed EMR. These two clusters also showed higher risk of BC progression (12% and 13% for cluster 6 and 7 respectively) compared to the remaining clusters (all lower then 5%) (Figure 1B). We therefore refered combined clusters (6 and 7) as high risk group. It is known that patients who receive second generation TKI treatment have lower risk of EMR failure compared to those receiving imatinib. We therefore investigated whether NIL/DAS treatment overcomes the risk of EMR failure in the high risk cluster. In the high risk group, EMR failure rate was 30% (10/33). In these patients, 20 out of 33 received IM treatment, of which 7 failed EMR (35% failure); while 13 patients received NIL/DAS treatment, and 3 failed EMR (23% failure). These suggested that this high risk group of patients may not benefit from more potent second generation TKI (compare to IM, p = 0.7). Summary/Conclusion: We identified a subgroup of patients who have higher risk of EMR failure and BC progression based on unsupervised clustering by cytokine profiling that were partially resistant to both first and second generation TKIs. Importantly, receiving second generation TKI NIL/DAS could not overcome the high risk of poor response in most of these patients. We postulate that cytokine profiling at diagnosis may contribute to a biomarker‐driven risk‐adapted approach to selection of frontline therapy.