Efficacy of Ruxolitinib in Patients With Chronic Neutrophilic Leukemia and Atypical Chronic Myeloid Leukemia

Abstract

Purpose: Colony-stimulating factor-3 receptor (CSF3R)-T618I is a recurrent activating mutation in chronic neutrophilic leukemia (CNL) and to a lesser extent in atypical chronic myeloid leukemia (aCML) resulting in constitutive JAK-STAT signaling. We sought to evaluate safety and efficacy of the JAK1/2 inhibitor ruxolitinib in patients with CNL and aCML, irrespective of CSF3R mutation status.

Methods: We conducted a phase II study of ruxolitinib in 44 patients (21 CNL and 23 aCML). The primary end point was overall hematologic response rate (ORR) by the end of 6 continuous 28-day cycles for the first 25 patients enrolled. We considered a response as either partial (PR) or complete response (CR). We expanded accrual to 44 patients to increase our ability to evaluate secondary end points, including grade ≥ 3 adverse events, spleen volume, symptom assessment, genetic correlates of response, and 2-year survival.

Results: ORR was 32% for the first 25 enrolled patients (8 PR [7 CNL and 1 aCML]). In the larger cohort of 44 patients, 35% had a response (11 PR [9 CNL and 2 aCML] and 4 CR [CNL]), and 50% had oncogenic CSF3R mutations. The mean absolute allele burden reduction of CSF3R-T618I after 6 cycles was greatest in the CR group, compared with the PR and no response groups. The most common cause of death is due to disease progression. Grade ≥ 3 anemia and thrombocytopenia were observed in 34% and 14% of patients, respectively. No serious adverse events attributed to ruxolitinib were observed.

Conclusion: Ruxolitinib was well tolerated and demonstrated an estimated response rate of 32%. Patients with a diagnosis of CNL and/or harboring CSF3R-T618I were most likely to respond.

Trial registration: ClinicalTrials.gov NCT02092324.

Figures

FIG 1.

(A) Schematic of 44 patients enrolled by diagnosis and CSF3R mutation status. Mut, CSF3R-T618I (n = 20), -T615A (n = 1), or -T640N (n = 1); WT, CSF3R–wild type; NR, nonresponder, 15 of 44 (34.1%) withdrew from study before the end of cycle 6; > C6, patients reached the end of cycle 6, 29 of 44 (65.9%). (B) Waterfall plot of absolute change of various hematologic parameters according to individual patients. Numbering of patients is arbitrarily based on the order presented in the heat map by mutation profile in Figure 2. The flat end of the line represents baseline value and the diamond-head end represents the end of cycle 6 value. Presented in this manner, both minimal and significant changes can be assessed on an individual basis. (†) Incidences of complete normalization of WBC, hemoglobin (Hgb) increases by > 2 g/dL, platelet (Plt) increases by > 30 × 109/L, and spleen volume reduction by ≥ 35%. aCML, atypical chronic myeloid leukemia; CNL, chronic neutrophilic leukemia.

FIG 2.

(A) Summary figure of patients according to mutation profile, response assessment, and clinical benefit (CB). The heat map displays the 11 most commonly mutated genes in this study population according to variant allele frequency (VAF; darkest blue 100% and white 0%). The protocol-defined and International Working Group (IWG)–defined response assessments are color-coded by response. CSF3R-T618I mutation status is indicated as well as other pertinent CSF3R variants. CB was assessed by IWG criteria with one modification. We added the CB criterion of decrease in spleen volume by ≥ 35% (used in myelofibrosis pivotal studies). CB was not evaluated in patients who had a complete response (CR) by IWG criteria or in patients who withdrew before the end of cycle 6 (denoted by gray shading). (B) Forest plot for the unadjusted odds ratio (OR) and 95% CI obtained from univariate logistic regression models, with some continuous variables scaled in units of 10 or 100 to produce reasonable range of ORs. Data separated by categorical and continuous variables. Analyses on 43 evaluable patients were performed based on various demographic and disease characteristics and also starting doses of ruxolitinib. Boldface indicates variables that were found to be statistically significant. We also explored the adjusted effects, potential confounders, and effect modifiers using multivariable logistic regression models. Multivariable models were not presented because diagnosis is the single dominating risk factor for response, even after controlling for other important risk factors (ie, CSF3R mutation status and spleen volume). Although CSF3R mutation and spleen volume showed individual confounding effects, the estimated effect of diagnosis from a multivariable logistic regression model with both covariates included was very close to that from the univariate model. aCML, atypical chronic myeloid leukemia; ANC, absolute neutrophil count; C6, cycle 6; CNL, chronic neutrophilic leukemia; Hgb, hemoglobin; MPN-SAF, myeloproliferative neoplasm symptom assessment form; NR, no response; Plt, platelet; PR, partial response; PS, palpable spleen; TSS, total symptom score.

FIG 3.

Different patterns of CSF3R-mutation variant allele frequency (VAF) changes over time with ruxolitinib. The status of the patient and cycle they reached as of data cutoff are indicated. (A) CNL-16, on study C19. (B) CNL-17, on study C16. (C) CNL-05, on study C43. (D) CNL-08, off study C19. C, cycle; CNL, chronic neutrophilic leukemia; CR, complete response; PR, partial response.

FIG 4.

Kaplan-Meier curves estimating probability of survival for all patients and subgroup analyses as indicated. The P values from log-rank (Mantel-Cox) tests are indicated on the graph. We currently have 2-year data for all 44 patients enrolled. For those with longer-term data, we censored the data at 2 years. Six patients were also censored on the date they underwent hematopoietic stem-cell transplantation. aCML, atypical chronic myeloid leukemia; CNL, chronic neutrophilic leukemia; IPSS, International Prognostic Scoring System; Mut, mutation; WT, wild type.