S1607 MOLECULAR ANALYSIS IN MAJIC PV CORRELATION WITH CLINICAL ENDPOINTS.

Abstract

Background: MAJIC PV is a randomised open label phase II study, in polycythaemia vera (PV) patients who are resistant or intolerant to hydroxycarbamide(HC) and is conducted in 38 UK centres. The primary objective of the study is to evaluate the activity (in terms of complete haematological response within one year) of ruxolitinib(Rux) compared to best available therapy(BAT). The molecular response by JAK2 V617F allele burden quantification was a secondary objective. Aims: We present here the results of the molecular assessment at the first year in MAJIC PV and correlation with clinical endpoints. Methods: 190 patients with PV were randomized 1:1 to Rux and BAT. Paired DNA samples from either whole blood or granulocytes were tested for JAK2 V617F allele burden by qPCR and results were defined according to European LeukemiaNet criteria (ELN) (Barosi et al. 2009). Statistical analysis of correlation between molecular response at one year and clinical endpoints (including best clinical haematological response, disease transformation, occurrence of haemorrhagic event and thrombotic event) was performed using Spearman's rank correlation test. Results: Paired samples were available for 124 patients: 70 were male, average age 64.7 years, 68 patients were on Rux and 56 on BAT arm (comprising 25 hydroxycarbamide, 8 interferon, 8 pegylated-interferon,7 anagrelide, 8 other treatment). The dose of Rux was 15 mg bd for 1 patient, 10 mg bd for 55 patients and 5 mg bd for 12 patients. The best clinical haematological response per ELN criteria at 1 year was for the RUX arm: CR in 28 patients, PR for 38 and NR in 2; for BAT arm responses were 14 CR, 39 PR and 3 NR. Of these patients 8 (12%) patients on Rux and 10 (18%) patients on BAT experienced at least thrombotic event; furthermore 11 transformations occurred: 4 (6%) on the Rux and 7 (13%) on the BAT arm. According to ELN molecular response criteria: Rux patients had 20 PR and 48 NR; and BAT patients: 1 CR, 9 PR and 46 NR. Demonstrating more molecular responses in the Rux arm with 29.4% CR/PR vs 17.9% for the BAT arm. We also further subcategorised the ELN NR category into progression (PROG+ 25–50% increase and PROG++ if >50% increase in JAK2 V617F allele burden). Here of the 48 NR patients treated with RUX: 2 had PROG+ and 6 PROG++, for BAT of 46 NR there were 5 PROG+ and 3 PROG++. Concerning correlations: we found no significant correlation between ELN molecular response (CR/PR, NR) and best clinical haematological response within 1 year (p = 0.22), disease transformation (p = 0.63) and occurrence of haemorrhagic event (p = 0.83). There was however a significant correlation between molecular response at one year and occurrence of thrombotic event at any time in the study. None of the 30 patients with CR/PR molecular response experienced a thrombotic event whereas 18 (19.1%) patients with NR experienced at least 1 thrombotic event (p = 0.01). Summary/Conclusion: This analysis shows that as well as having clinical and symptom responses to Rux, PV patients may also have molecular responses. Molecular responses also occurred in patients on the BAT arm including the only complete molecular response. Previous studies have suggested that molecular responses can be slow for patients with PV and further time points from our study will be assessed. Importantly we show for the first time that molecular responses may correlate with reduced risk of thrombosis, this finding has the potential to change the therapeutic paradigm in PV.