Integrating cytokine profiles with clinical and molecular biomarkers to predict molecular relapse after TKI discontinuation in CML patients

Abstract Background: The introduction of tyrosine kinase inhibitors (TKIs) has improved the management of chronic myeloid leukemia (CML), making treatment-free remission (TFR) an attainable objective. Despite this, half of the patients lose their major molecular response (MMR) within the first six months after TKIs discontinuation. TFR continues to pose a significant challenge in low middle income countries. There is currently no biomarker that reliably predicts TFR in CML. The aim of our study is to identify plasma cytokines signature as predictive biomarkers that distinguish CML patients maintaining molecular remission following TKI discontinuation from those who relapse. Methods This study is a subanalysis within the multicentric AST- Argentina Stop Trial which attempted TKI in CML patients meeting the main inclusion criteria of a sustained Deep Molecular Response (DMR) (BCR-ABL1IS ≤ 0.01%, or undetectable BCR-ABL1 i.e., MR4.0 or better) for a minimum of two years.A total of 81 CML patients from 15 centers in Argentina were included between February 2019 and April 2023. Peripheral blood samples were collected before stopping TKI treatment. Cytokines and chemokines in plasma were measured using multiplex bead assays on a Magpix® system. IL6, MCP-1, GCSF, IL-10, IL-12 (p70), MIP-1a, TNF-a, VEGFA were assessed. Patients were classified into two groups based on their response to TKI discontinuation the Relapsed (R) group, defined by the loss of MMR, and the Non-Relapsed (NR) group, comprising those who maintained molecular remission. Molecular relapse-free survival rates were estimated using Kaplan-Meier analysis, and group comparisons were performed with the log-rank test and Man whitney. A p-value of less than 0.05 was considered statistically significant.A classification tree algorithm was applied to the variables selected after preprocessing. All analyses were conducted using R statistical software (version 4.3.1). Results At the time of the analysis, the median molecular follow-up after TKI discontinuation was 60 months (range, 20 to 69 months). Thirty-two patients (39,5%) lost MMR, most frequently within the first 6 months, leading to a molecular relapse-free survival of 62% at 24 months and 60,5% for the entire follow-up period. Significant differences were found between NR and R patients in both the duration of DMR until discontinuation and the overall treatment duration (Mann–Whitney test; p =0.030 and p =0.013, respectively). stratification by a BCR-ABL/ABLIS(%) threshold of 0.0036%, close to the MR4.5 international standard (0.0032%), showed that patients below this threshold had significantly better molecular relapse-free survival time than those above it 61% vs 14% p=0.0010. The median duration of DMR was 90 m, (34-203) for NR patients and 74m ( 30-179) for R patients p=0.030. Regarding cytokine panel, significant differences between R and NR patients were observed; in particular, MCP-1 and IL-6 being significantly higher in the NR group with median concentrations of 289.1 vs 253.5 pg/mL (Mann-Whitney test; p = 0.012) and 3.9 vs 2.9 pg/mL (Mann-Whitney test; p = 0.007), respectively Decision tree analysis identified MCP-1 as the primary stratifying factor: patients with MCP-1 ≥ 264 pg/mL had a lower relapse risk (23%), while those with lower levels relapsed more frequently (58%). Among MCP-1^low patients, IL-6 further refined prognosis: relapse occurred in 72% of MCP-1^low/IL-6^low patients versus 12% in MCP-1^low/IL-6^high. This cytokine-based stratification significantly discriminated molecular relapse-free survival (log-rank p < 0.001). Predictive performance of the model was confirmed in a validation cohort, showing good sensitivity and no false positives Conclusion A novel contribution of this work is the integration of cytokine profiles into the prediction of molecular relapse, extending beyond traditional clinical and functional molecular biomarkers. Our study suggests a potential role for IL-6 and MCP-1, in predicting molecular relapse following TKI discontinuation. The cytokine model achieved high specificity and a Positive Predictive Value of 100%, indicating that when predicts relapse, it is highly reliable.These findings support the notion that cytokine profiles may reflect aspects of the immune environment, suggesting an active role of these biomarkers in CML immunobiology, that help to early identify patients at higher risk of relapse toward a personalized TKI discontinuation approach.