Preventing thrombosis with ruxolitinib in polycythemia vera

Myeloproliferative neoplasms (MPNs) are clonal disorders of hematopoietic stem cells that cause increased proliferation of cells from the myeloid lineage. Polycythemia vera (PV) is an MPN that is characterized by the overproduction of erythrocytes and may also be associated with an increase in leukocytes and thrombocytes. In most patients, PV arises due to a dysregulated JAK-STAT signaling pathway, resulting from an activating JAK2 mutation. In PV, the specific JAK2V617F mutation occurs at a frequency of 95%.¹

The clinical features of PV include¹:

• headache, tiredness, dizziness, and sweating
• pruritus, due to high levels of histamine released by mast cells in the skin (~40% of patients)
• thrombotic events, e.g. strokes and pulmonary embolism (~33% of patients)
• bleeding (~25% of patients)

The main aim of therapeutic management of PV is to alleviate the symptom burden, prevent or reduce the recurrence of thromboembolic events, and prevent disease transformation.² The first-line treatment of patients with high-risk PV is hydroxyurea (HU), and studies suggest HU may have an anti-thrombotic effect.^3,4 The current recommendation from the European LeukemiaNet (ELN) to prevent thrombosis in patients who become resistant or intolerant to HU is to administer the JAK1/JAK2 inhibitor ruxolitinib or recombinant interferon α therapy (rINFα). Since there is no comparison between these agents, the decision is usually made based on patient age and drug availability.⁴

Data supporting the use of ruxolitinib in this setting predominantly comes from the RESPONSE trials (RESPONSE and RESPONSE-2). Based on the results of these studies, the United States Food & Drug Administration (FDA) and European Medicines Agency (EMA) approved ruxolitinib for the treatment of patients who have an inadequate response to, or who cannot tolerate, HU.^5,6 In the ELN recommendations, due to performance bias and a lack of comparator arm, the overall quality of evidence for the critical outcomes of the RESPONSE trials was downgraded, however, the hematocrit response and primary response were deemed to have a high level of evidence.⁴  So far, there is limited data evaluating the efficacy of ruxolitinib with regards to the reduction of thrombosis. For this reason, Arianna Masciulli, Papa Giovanni XXIII Hospital, Bergamo, IT, and colleagues conducted a systemic review and meta-analysis of randomized controlled trials (RCTs) comparing ruxolitinib to best available therapy (BAT), specifically for the purpose of reducing the incidence of thrombosis.³

Methods³

• RCTs that evaluated the incidence of thrombosis in patients with PV receiving ruxolitinib vs. BAT were identified from medical databases and abstracts presented at congresses from 2014
• Data were extracted: number of patients treated per arm, median and/or total follow-up time, number of patients with thromboembolic events, median/mean age of patients, and number of patients with history of thrombosis
• The final selection for meta-analysis (quantitative data) included 15 records
• Data came from four main RCTs, as shown in Table 1, with data specifically used for the meta-analysis coming from the publications indicated in bold

Table 1. RCTs used for systematic review³

The risk of bias was assessed and deemed low. However, to avoid bias from the possibility of cross-over, data were only included up to the cut-off period for cross-over (where applicable).

Results³

Qualitative synthesis of results

All four studies (RESPONSE, RESPONSE-2, RELIEF, and MAJIC) used surrogate efficacy endpoints. Specifically, these included complete hematologic response (CHR) and molecular response (MR), with the incidence of thromboembolic events reported in the safety analyses. All four studies found ruxolitinib had a significant effect on CHR. The MAJIC substudy (of patients with PV only) found ruxolitinib was more efficient than BAT in obtaining MR, which is important because MR may correlate with a lower risk of thrombosis.

Follow-up analyses of the RESPONSE trial considered thrombosis incidence after long-term treatment with ruxolitinib. These analyses compared:

• Ruxolitinib vs. propensity-score matched patients: advantage of ruxolitinib in overall survival and thrombosis
• BAT (32-week follow-up) vs. ruxolitinib (from the beginning of treatment with 5-year follow-up): thrombosis incidence was lower in patients receiving ruxolitinib (1.2% vs. 8.2%) but significance was not reported

Patients treated with BAT on the RESPONSE trials who crossed over to ruxolitinib had a thrombosis rate approximately in the middle of the two main patient populations analyzed.

Meta-analysis of thrombosis incidence: ruxolitinib vs. BAT³

For all studies, and for both arms, the investigators calculated the thrombosis incidence and relative risk (RR) of thrombosis. Of the 663 patients included in the analysis, patient characteristics were overall balanced (median age, years [BAT vs ruxolitinib]: 63.9 vs. 63.3). However, follow-up duration varied between studies and from arm to arm, ranging from 0.3 to 2.6 years. Overall, the data do not show a clear advantage of ruxolitinib over BAT in preventing thrombosis in the follow-up period examined.

Table 2. Annual incidence of thrombosis per study arm³

Based on a meta-regression of ruxolitinib treatment effect (n = 663), thrombosis incidence rates are shown below:

• Overall: 4.30% (95% CI, 3.00–5.60)
• BAT: 5.51% (95% CI, 3.72–7.30)
• Ruxolitinib: 3.09% (95% CI, 1.22–4.96)
• The RR of thrombosis for ruxolitinib vs. BAT was 0.56, however this was not statistically significant (p = 0.098)

Conclusion³

• The overall thrombosis rate was numerically lower with ruxolitinib, but the hazard ratio was not statistically significant
• Limitations of this study include:
  • Limited number of events
  • Annual thrombosis rate estimates are limited to a short follow-up duration in some treatment arms
• Evidence that ruxolitinib prevents thrombotic events is lacking and a prospective clinical trial in a selected patient population is warranted. However, the authors question whether such a trial would be feasible as it would require thousands of patients to be enrolled
• Alternatively, preliminary evidence seems to suggest that maintaining a target hematocrit does correlate with lower incidence of cardiovascular events. The authors therefore suggest using combined or surrogate endpoints (such as hematocrit) may better assess impact of treatment on thrombosis rates