Comprehensive haematological control with ruxolitinib in patients with polycythaemia vera resistant to or intolerant of hydroxycarbamide

Polycythaemia vera (PV) is a myeloproliferative neoplasm characterized by excessive erythrocytosis and elevated haematocrit (HCT) (Barbui et al, 2011; Tefferi & Barbui, 2015). Patients have an increased risk of thrombotic and cardiovascular events (Tefferi & Barbui, 2015); therefore, therapeutic approaches focus on lowering this risk and include low-dose aspirin and phlebotomy for low-risk patients or cytoreductive treatment (e.g, hydroxycarbamide [HC]) for high-risk patients (Barbui et al, 2011). Given findings from the Cytoreductive Therapy in PV (CYTO-PV) study, achieving HCT <45% is a key therapeutic goal (Marchioli et al, 2013); however, leucocytosis and thrombocytosis are often present and are associated with an increased risk of complications and adverse outcomes (Landolfi et al, 2007; Alvarez-Larran et al, 2012; Marchioli et al, 2013; Barbui et al, 2015). Ruxolitinib is a Janus kinase (JAK) 1/JAK2 inhibitor approved for the treatment of patients with PV who are resistant to or intolerant of HC based on the phase 3 RESPONSE study, in which ruxolitinib (n = 110) proved superior to standard therapy (best available therapy [BAT]; n = 112) (Vannucchi et al, 2015). A higher proportion of ruxolitinib-treated patients achieved HCT control with fewer phlebotomies compared with BAT-treated patients (60% vs. 19%) (Vannucchi et al, 2015; Verstovsek et al, 2016); frequent phlebotomies have been associated with uncontrolled HCT (Barbui et al, 2017). Despite having their HCT temporarily controlled with phlebotomies 2 weeks prior to randomization, many patients had features of uncontrolled myeloproliferation at baseline, including HCT >45% and elevated white blood cell (WBC) and platelet (PLT) counts (Table SI). Because comprehensive haematological control is a suggested management goal by the European LeukemiaNet (ELN) (Barbui et al, 2011) and is considered an important endpoint in clinical studies (Barosi et al, 2015), we compared the efficacy of ruxolitinib with BAT in achieving target HCT, WBC, and PLT levels in RESPONSE. For ruxolitinib-treated patients, HCT decreased from weeks 2 (44·3%) to 12 (40·1%) and remained relatively stable thereafter (mean change of −3·12% to −4·36% between weeks 12 and 80). HCT in the BAT arm was higher than baseline throughout treatment (mean change of 0·06% to 1·03% between weeks 12 and 32). For patients with HCT >45% at baseline, mean HCT in the ruxolitinib arm declined to <45% by week 8 and was lower compared with BAT (Fig 1A), which remained mostly >45% throughout treatment. Although HCT in individual patients can fluctuate over time, HCT remained <45% in most patients randomized to ruxolitinib (Fig 2A); in contrast, a large proportion of patients randomized to BAT had HCT >45% (Fig 2B). Similarly, for patients with higher WBC counts at baseline (Q3, >16·4 to ≤22·6 × 109/l; Q4, >22·6 to 82·8 × 109/l), mean values improved with ruxolitinib and were lower than with BAT. Patients with the highest WBC counts (Q4) had the largest reductions (Fig 1B), whereas those with near-normal WBC counts (Q1, 2·9 to ≤10·7 × 109/l; Q2, >10·7 to ≤16·4 × 109/l) had levels that remained at or below baseline values throughout treatment (Figure S1). Moreover, patients had greater mean reductions in PLT counts with ruxolitinib versus BAT, with the largest reductions seen in the highest quartile (Q4, PLTs >650 to 1852 × 109/l; Fig 1C; Figure S2). In patients with baseline WBC ≥11 × 109/l, those treated with ruxolitinib had greater mean reductions compared with BAT (week 12, −7·7 vs. −3·2 × 109/l; week 32, −7·2 vs. −4·2 × 109/l) (Figure S3A); reductions were smaller among patients who received HC (n = 46: week 12, −1·2 × 109/l; week 32, −2·2 × 109/l). Furthermore, a greater proportion of patients who received ruxolitinib achieved WBC ≤10 × 109/l or a ≥50% reduction in WBCs (week 12: ruxolitinib, 41% [34/83] vs. BAT, 19% [15/79] vs. HC, 13% [6/46]; week 32: 45% [37/83] vs. 22% [17/79] vs. 9% [4/46]). A smaller proportion of patients treated with ruxolitinib had increases in WBCs compared with BAT or HC at week 12 (10·8% vs. 35·4% vs. 47·8%). In patients with baseline WBC <11 × 109/l, mean values remained stable over time in all 3 groups (Figure S3B), and there was no significant difference in the proportion of patients with worsening WBC counts at week 12 (14·8% vs. 24·2% vs. 20·0%). Among patients with WBC >10 or >15 × 109/l at baseline, higher proportions of ruxolitinib-treated patients had WBC ≤10 × 109/l (i.e., an ELN response) at week 32 compared with BAT (WBC >10 × 109/l: ruxolitinib, 31·0% [27/87] vs. BAT, 21·6% [19/88]; WBC >15 × 109/l: 26·6% [17/64] vs. 14·5% [9/62]), and responses improved at week 80 with ruxolitinib (WBC >10 × 109/l: 47·1% [41/87]; WBC >15 × 109/l, 42·2% [27/64]). A higher proportion of ruxolitinib-treated patients achieved an ELN response in PLTs (i.e., ≤400 × 109/l) at week 32 compared with BAT, including those with PLTs >400 × 109/l (44·4% [24/54] vs. 21·7% [13/60]) and PLTs >600 × 109/l at baseline (36·4% [12/33] vs. 27·8% [10/36]). The proportion of ruxolitinib-treated patients with normalized PLTs improved over time, with 59·3% (32/54) and 54·5% (18/33) of patients with PLTs >400 and >600 × 109/l, respectively, achieving PLTs ≤400 × 109/l at week 80. Compared with patients in the ruxolitinib arm, those in the BAT arm spent 12-fold more time (in percentage of time on study) with HCT >45% (median [Q1–Q3], 39·1% [12·9–65·9%] vs. 3·0% [0–16·7%]; Figure S4; see Supplemental Methods), 0·9 and 8·1 times more time with WBCs >10 × 109/l (median, 97·7% vs. 50·3%) and >15 × 109/l (3·5% vs. 31·8%), respectively, and 1·7 times more time with PLTs >400 × 109/l (41·0% vs. 15·3%); there was no difference in time with PLTs >600 × 109/l (1·7% vs. 1·7%). The mean proportion of time that patients had control of peripheral blood counts (i.e., HCT ≤45%, WBC ≤10 × 109/l, and PLT ≤400 × 109/l) was also higher in patients receiving ruxolitinib compared with BAT (36·6% vs. 12·9%; Fig 2C). Overall, in addition to controlling HCT and improving splenomegaly and symptoms, ruxolitinib provided durable and comprehensive haematological control in the RESPONSE study. Although the study was not powered to assess thromboembolic events, the lower rate observed in the ruxolitinib versus BAT arms (1·8 vs. 8·2 per 100 patient-years of exposure) is consistent with the effects of ruxolitinib on HCT and WBC counts, which have been shown to be risk factors. Consistent comprehensive haematological control could therefore be associated with improving long-term outcomes; however, further research is warranted. The authors thank Ahmad Naim and Brian Gadbaw for their helpful contributions to data analysis and interpretation. Dany Habr is currently an employee of AbbVie Inc, North Chicago, IL, USA, but was employed by Novartis at time of manuscript development. Medical editorial assistance was provided by Karen Chinchilla, PhD (ArticulateScience LLC, Hamilton, NJ, USA), and funded by Novartis Pharmaceuticals Corporation. FrP, PZ, SD, FaP, SV, DSH, DH, and J-JK contributed to study conception and design. CNH, MG, CM, TM, FrP, PZ, SD, FaP, PG, SV, and J-JK contributed to provision of study materials and patients. CNH, MG, CM, TM, FrP, PZ, SD, FaP, PG, SV, MMJ, DH, and J-JK collected data. WS, JL, and MK performed statistical analysis. CNH, MG, CM, TM, FrP, PZ, SD, FaP, PG, SV, MMJ, DSH, WS, JL, MK, DH, and J-JK analysed and interpreted data. All authors drafted and approved the manuscript for submission. CNH has received research support from Novartis, Cell Therapeutics, Gilead, and Baxalta through the institution; has received personal fees from Novartis, Shire, Gilead, and Baxalta; and has received grant and nonfinancial support from Novartis outside the submitted work. MG has received honoraria and participated in advisory boards for Shire, AOP Orphan, Novartis, Roche, Janssen, and Baxalta. CM has received honoraria and participated in speakers bureaus for Novartis and Incyte. TM has acted as a consultant for Takeda, Novartis, Bristol-Myers Squibb, and Janssen Cilag. FrP has received honoraria and participated in speakers bureaus for Novartis. FaP has received grants from Novartis and has received personal fees from Novartis, Bristol-Myers Squibb, and ARIAD. SV has received research funding for the conduct of the presented clinical study from Incyte and Novartis. MMJ, DSH, and WS are employees of Incyte. JL and MK are employees of Novartis. DH was an employee of Novartis at the time of study execution and analysis, and submission and review of the manuscript. J-JK has received grants from Novartis and AOP Orphan; has received personal fees from AOP Orphan and Shire; and has received nonfinancial support from Novartis. PZ, PG, and SD have nothing to disclose. Figure S1. WBC counts (mean ± SE) over time by baseline quartiles. Figure S2. PLT counts (mean ± SE) over time by baseline quartiles. Figure S3. Mean change from baseline in WBC counts for patient subgroups. Figure S4. Examples for determining time spent with HCT >45%. Table SI. Baseline characteristics of patients in the RESPONSE study. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.