Prevalence and Prediction of Ruxolitinib Treatment Emergent Thrombocytopenia in Myelofibrosis

Introduction

Ruxolitinib (rux) treatment for myelofibrosis (MF) is frequently complicated by the development of thrombocytopenia, leading to dose reduction and treatment discontinuation, an event which is associated with poor outcomes. With the recent availability of JAK inhibitors that are safe and effective in thrombocytopenic MF patients, identifying predictors for rux-induced thrombocytopenia may allow for personalized JAK inhibitor selection in the upfront setting.

Methods

We conducted a retrospective cohort study of JAK inhibitor naïve MF patients treated with rux at two large academic institutions (Mount Sinai and Moffitt Cancer Center). Included patients had a baseline platelet count of ≥100 x 10 ⁹/L. The primary outcome was the development of a platelet count of <100 x 10 ⁹/L by 3 months (measured at both 1 and 3 months), a time period selected to isolate rux-induced thrombocytopenia as opposed to disease progression. Predictors of moderate or severe thrombocytopenia development were identified using univariate and multivariate logistic regression. A p-value of 0.1 was chosen as a cutoff for inclusion into the multivariable model. Variables that had a skewed distribution, including platelet count, were log transformed for normalization.

Results

A total of 246 MF patients were identified with baseline characteristics shown in the Table. The median platelet count at baseline was 289 x 10 ⁹/L, with 42 patients (17%), 35 patients (14%), and 27 patients (11%) having a platelet count of 100-150 x 10 ⁹/L, 151-200 x 10 ⁹/L, and 201-250 x 10 ⁹/L, respectively. 52 patients (21%) were RBC transfusion dependent. Driver mutational prevalence was as expected and one-third of patients were high molecular risk. 17 patients (6.9%) harbored a U2AF1 mutation, previously noted to be associated with thrombocytopenia in MF. The median time from MF diagnosis to rux initiation was 6 months (IQR: 1 to 30 months).

A total of 37 patients (15%) developed a platelet count of <100 x 10 ⁹/L by 3 months. On univariate analysis of baseline characteristics including hematologic factors, disease duration, mutational profile, bone marrow cellularity and fibrosis, and rux dose, log transformed baseline platelet count (odds ratio [OR]: 0.08, 95% CI: 0.03-0.18), symptoms at baseline (OR: 3.18, 95% CI 1.40-8.18) and DIPSS category (OR: 1.26, 95% CI: 0.99-1.62), were found to be predictive of platelet <100 x 10 ⁹/L by 3 months. However, on multivariate analysis, baseline constitutional symptoms and DIPSS category lost statistical significance. Of note, baseline rux dose was not associated with development of moderate thrombocytopenia on univariate analysis (OR: 1.02, 95% CI 0.98-1.06).

To analyze the impact of degree of baseline platelet count on development of rux-induced thrombocytopenia, we analyzed the predictive capacity of different baseline platelet cutoffs in a multivariable model. As expected, deeper baseline thrombocytopenia was associated with a higher OR of developing a platelet count of <100 x 10 ⁹/L by 3 months (Figure). Specifically, a platelet count of 100-150 x 10 ⁹/L had the highest risk (OR: 25.1, 95% CI: 9.3-7.77), as compared with patients with >250 x 10 ⁹/L platelets at baseline. Baseline platelet count of 151-200 x 10 ⁹/L still retained significance (OR: 4.74, 95% CI: 1.37-16.5), while baseline platelets 200-250 x 10 ⁹/L were not associated with a significantly elevated risk of moderate thrombocytopenia.

Conclusions

In this relatively large cohort of rux-treated MF patients, baseline platelet count near or below the lower limit of normal (100-200 x 10 ⁹/L) was significantly and independently associated with the development of rux-induced moderate or severe thrombocytopenia, with platelet counts 100-150 x 10 ⁹/L being the most potent predictor. Rux dose did not predict the development of thrombocytopenia, likely due to initial dose choice being a function of baseline platelet count. Interestingly, in this analysis, baseline mutational profile did not predict for rux-induced thrombocytopenia. These data may guide the preferential use of alternative JAK inhibitors in patients who are at increased risk of developing treatment-related thrombocytopenia that would necessitate dose reduction or hold.