A Novel Score to Predict Interferon-Alpha Therapy Responsiveness in Patients with Essential Thrombocythemia

Interferon alpha (IFN-α) is an attractive agent for the treatment of Essential Thrombocythemia (ET) due to its ability to induce clonal complete remission, sometimes lasting beyond treatment discontinuation, and to its recognized non-leukemogenicity. However, despite decades of clinical experience with IFN-α in patients with MPNs, optimal dose schedules, treatment duration and the ultimate molecular basis of the heterogeneous response still remain undefined. Hence, the early identification of IFN-sensitive patients may help limit IFN-α exposure to those who really benefit from treatment. Aim. Here we report the results of a trial involving 61 ET patients treated with IFN-α, aimed to identify the baseline molecular and clinical parameters able to predict response to treatment. Methods. IFN treatment schedule implied an initial induction phase with 3MU/five times a week; in patients who reached a platelet count <400x109/L, IFN dose density was progressively reduced and patients considered Good-responders (Good-R). Patients who failed to achieve platelet normalization after induction, underwent an additional three-month period of IFN treatment at 3MU/five times a week. In case platelet count was still >600x109/L or platelet reduction was <50% the baseline level, the patient was considered resistant to the IFN therapy (Bad-R). Careful medical history, main laboratory data and spleen volume, assessed by ultrasonography scan were recorded in all patients at presentation and during follow-up. Complete hematological response (CHR) is defined as the normalization of both platelet and WBC counts (<400x109/L and <10x109/L, respectively) and the absence of disease-related symptoms. mRNA levels of JAK1, JAK2, STAT1, STAT3, SOCS1, SOCS3 and TYK2 were assayed in pre-treatment bone marrow specimens by Real-Time PCR using the SYBR Green method. Results. After a median follow-up of 41.2 months, 72% of patients achieved CHR and were considered Good-Rs for subsequent analysis, whereas the remaining 17 were considered Bad-Rs. Among the Good-Rs, 24 (54%) are still on therapy with standard IFN-α doses (i.e. 3 MU 3 or 2 times a week), whereas 10 (23%) are maintained in CHR by the administration of very low doses of IFN-α-2b (3 MU every 7 or 15 days), and 3 (7%) have maintained CHR after therapy discontinuation (up to a median time of 31 months). The initial univariate analysis indicated that the mRNA levels of JAK1, STAT3, SOCS3 were significantly lower in Good-R than in Bad-R patients. Interestingly, among the different genes involved in the IFN-α receptor pathway, the expression levels of JAK1, together with spleen volume and platelet count, were selected by the stepwise multivariate analysis as the variables that independently correlate with IFN-α response. We used the relative HRs and the optimal cut-offs for response calculated for each variable by the ROC analysis to develop a prognostic score able to predict IFN-α response. This score has an overall 87% diagnostic efficiency in discriminating IFN-α response and unambiguously identifies the response to IFN-α in most patients, avoiding treatment in those with no probability of gaining benefit from this therapy. In addition, this score is able to identify unambiguously the response to IFN in a sizeable proportion of patients: an IFN-R score of 3 or 4 (31 patients, corresponding to 70.4% of the Good-R) indicates a 100% odd to obtain CHR, while a score of 0 indicates no chance of achieving a response. Conclusion. In conclusion, this study shows for the first time that the use of three simples parameters predicts the response to IFN in ET patients.