Fedratinib in patients with myelofibrosis previously treated with ruxolitinib: An updated analysis of the JAKARTA2 study using stringent criteria for ruxolitinib failure

Abstract Fedratinib is an oral, selective Janus kinase 2 (JAK2) inhibitor. The phase II JAKARTA2 study assessed fedratinib in patients with intermediate‐ or high‐risk myelofibrosis (MF) who were resistant or intolerant to prior ruxolitinib per investigator assessment. Patients received fedratinib 400 mg/day in 28‐day cycles. The JAKARTA2 outcomes were initially reported using a last‐observation‐carried forward (LOCF) analysis in a “Per Protocol” population. This updated analysis of JAKARTA2 employs intention‐to‐treat analysis principles without LOCF for all treated patients (ITT Population; N = 97), and for a patient subgroup who met more stringent definitions of prior ruxolitinib failure (Stringent Criteria Cohort; n = 79). Median duration of prior ruxolitinib exposure was 10.7 months. The primary endpoint was spleen volume response rate (SVRR; ≥35% spleen volume decrease from baseline to end of cycle 6 [EOC6]). The SVRR was 31% in the ITT Population and 30% in the Stringent Criteria Cohort. Median duration of spleen volume response was not reached. Symptom response rate (≥50% reduction from baseline to EOC6 in total symptom score [TSS] on the modified Myelofibrosis Symptom Assessment Form [MFSAF]) was 27%. Grade 3‐4 anemia and thrombocytopenia rates were 38% and 22%, respectively. Patients with advanced MF substantially pretreated with ruxolitinib attained robust spleen responses and reduced symptom burden with fedratinib.

Many patients treated with ruxolitinib lose response over time, achieve only a suboptimal response, or develop cytopenias during treatment, resulting in therapy discontinuation. 12 The combined ruxolitinib discontinuation rate in the phase III COMFORT-I and COMFORT-II trials was~50% at 3 years and~72% at 5 years. 13,[14][15][16] A retrospective review of data from two large United States (US) claims databases suggests ruxolitinib discontinuation rates in clinical practice during early treatment are at least as high as rates in clinical trials. 17 The prognosis for these patients is generally poor, with median survival ranging from 6 to 28 months, depending on whether a patient is in the chronic phase of MF or has transitioned into the blast phase when ruxolitinib was discontinued. 12,[18][19][20] There is no approved standard of care for patients with MF previously treated with ruxolitinib; thus, there is an important medical need for an effective therapy in this setting. MF. 21 Fedratinib was recently added to the National Comprehensive Care Network guidelines for treatment of MPNs, as an initial treatment option for patients with intermedicate-2 or high-risk MF or as second-line therapy for those who do not respond or lose response to ruxolitinib. 1 Fedratinib has higher inhibitory activity for JAK2 over family members JAK1, JAK3 and TYK2, and is a more selective inhibitor of JAK2 than ruxolitinib. 22 Additionally, an in vitro drug screen identified 211 mutations resistant against ruxolitinib that were fully sensitive to fedratinib, 23 perhaps by a novel mechanism of JAK2 kinase inhibition by fedratinib that prevents emergence of genetic resistance, making it a therapeutic option for patients who are resistant to ruxolitinib therapy. Fedratinib also has a longer effective halflife than ruxolitinib (~41 hours vs 3 hours, respectively), which allows more persistent JAK2 inhibition and makes it suitable for once-daily dosing. 24,25 The international, single-arm phase II JAKARTA2 trial evaluated the efficacy and safety of fedratinib in patients with intermediate-or high-risk primary MF, post-PV MF, or post-ET MF, who were previously treated with ruxolitinib. The fedratinib clinical development program was placed on clinical hold in November 2013 by the US FDA following reports of suspected Wernickeʼs encephalopathy (WE), a rare but serious neurological condition. As a result, the JAKARTA2 trial was terminated; all patients were required to discontinue fedratinib treatment and the study was substantially truncated.
Note, JAKARTA2 enrolled patients who were resistant or intolerant to prior ruxolitinib therapy based on investigator assessment. The primary endpoint was the spleen volume response rate (SVRR); that is, the proportion of patients who achieved a ≥35% reduction from baseline spleen volume at the end of cycle 6 (EOC6). Based on the prospective Statistical Analysis Plan, the primary efficacy analysis of JAKARTA2 was performed in the Per Protocol population, which comprised patients with spleen volume assessments at baseline and at least one post-baseline time point. The original analysis utilized a lastobservation-carried-forward (LOCF) analysis method, which allowed for the last spleen volume assessment to be "carried forward" for patients missing EOC6 assessments. 26 At EOC6, the SVRR (using LOCF method) in the Per Protocol population was 55% (95% confidence interval [CI] 44, 66). 26 The objectives of this updated analysis are to confirm the efficacy of fedratinib in the JAKARTA2 study by employing intention-to-treat (ITT) analysis principles for all treated patients. There is no imputation made for missing data, and to demonstrate efficacy outcomes in a subgroup of JAKARTA2 patients who met new, more stringent criteria for relapsed, refractory, or intolerant to ruxolitinib than were used in the original analysis. Additionally, a sensitivity analysis assessed fedratinib efficacy in patients who met the more stringent criteria for ruxolitinib failure, and for whom the primary endpoint would have been least affected by early termination of the study.

| METHODS
The phase II, international, multicenter, open-label, single-arm JAKARTA2 study (ClinicalTrials.gov NCT01523171) was conducted at 40 sites in 10 countries. The study protocol was approved by relevant independent ethics committees or institutional review boards at each site. All patients provided written, informed consent before study participation. Detailed study design and inclusion/exclusion criteria have been described. 26 Briefly, eligible patients were aged ≥18 years, with primary, post-PV, or post-ET MF; intermediate-1 (with symptoms), intermediate-2, or high risk disease; palpable splenomegaly (≥5 cm below the left costal margin); Eastern Cooperative Oncology Group (ECOG) performance status scores ≤2; platelet counts ≥50 × 10 9 /L, and were considered by their treating investigator to be resistant or intolerant to ruxolitinib (Table S1).
Patients received oral fedratinib 400 mg once-daily in repeated 28-day treatment cycles. Dose escalation was permitted up to 600 mg/day if there was <50% reduction in spleen size by palpation at the end of cycles 2 and 4, and the fedratinib dose could be reduced, interrupted, or discontinued in cases of toxicity.
This updated analysis assesses three patient populations (Table S1): the ITT Population includes all patients who enrolled in JAKARTA2; the Stringent Criteria Cohort comprises a subset of patients who met the new, more stringent criteria for relapsed or refractory to ruxolitinib (based on spleen volume or size assessments), or intolerant to ruxolitinib, than used in the original analysis. 26 The Sensitivity Analysis Cohort includes the subgroup of patients within the Stringent Criteria Cohort who were least affected by early study termination, that is, those who reached fedratinib treatment cycle six, or discontinued fedratinib before cycle six for reasons other than "study terminated by the sponsor". These criteria were presented to and accepted by MF experts from the United States and European Union at an advisory board meeting with the study sponsor. The sponsors also reviewed the proposed criteria with relevant health authorities.

| Efficacy endpoints
The primary endpoint was SVRR, defined as the proportion of patients who achieved a ≥ 35% reduction in spleen volume from baseline to EOC6. Spleen volume assessments were to be performed at baseline, at the end of cycles three and six, and every six cycles thereafter. Blinded review of MRI/CT scans was performed by an independent central imaging laboratory. This updated analysis employed ITT analysis principles; thus, missing spleen volume data were not imputed (no LOCF) for the primary endpoint, and patients missing assessments at EOC6 were considered nonresponders. No formal statistical adjustments were made for possible covariate effects or for multiple comparisons.
A key secondary endpoint was symptom response rate, defined as the proportion of patients with a ≥50% decrease in total symptom score (TSS) from baseline to EOC6. Symptom scores were subjectively evaluated using the modified Myelofibrosis Symptom Assessment Form (MFSAF 27 ) e-diary, which assesses the severity of six key MFassociated symptoms (night sweats, pruritis, abdominal discomfort, early satiety, pain under ribs on left side, and bone or muscle pain), each assigned a score from 0 (absent) to 10 (worst imaginable). TSS is the sum of individual symptom scores. The MFSAF was to be completed beginning 7 days before cycle 1-day 1, and then 7 days before day 1 of each subsequent treatment cycle for six treatment cycles, and at EOC6. The MFSAF Analysis Population included patients with evaluable TSS data available at baseline and at least 1 post-baseline assessment. Confidence intervals for spleen volume and symptom response rates were calculated using the Clopper-Pearson Exact method.
Additional secondary endpoints included the duration of spleen volume response, calculated from the date of first response to the date of disease progression (≥25% spleen volume increase from baseline) or death, whichever came first. Duration of spleen response was estimated by Kaplan-Meier (KM) analysis among patients who responded at any time on-study. In the absence of disease progression or death before the analysis cut-off date (May 7, 2014), duration of response was censored at the date of the last valid assessment before data cutoff. Also assessed were median percent change in spleen volume from baseline to EOC6, proportion of patients with ≥50% reductions in spleen size by palpation at EOC6, and proportion of patients with ≥35% reduction from baseline in spleen volume at EOC3.

| Safety
The safety and tolerability of fedratinib were evaluated based on the

| Patients
In all, 97 patients were enrolled and treated in JAKARTA2 between 30 April 2012, and 7 May 2014, and comprise the ITT Population ( Figure S1). The majority of patients (n = 63; 65%) discontinued treatment due to study termination following the fedratinib clinical hold.
Other common reasons for treatment discontinuation were adverse events (19%) and disease progression (6%).
Based on new, more stringent criteria for ruxolitinib failure, 79/97 patients (81%) were refractory (n = 47; 48%), relapsed (n = 18; 19%), or intolerant (n = 14; 14%) to prior ruxolitinib therapy and comprised the Stringent Criteria Cohort. The remaining 18 patients were excluded from the Stringent Criteria Cohort because they had an adequate response to ruxolitinib (n = 3), were missing ruxolitinib response data (n = 8), or did not receive ≥3 months of ruxolitinib treatment (n = 7). The Sensitivity Analysis Cohort included 66 patients within the Stringent Criteria Cohort who had the opportunity to receive six cycles of fedratinib therapy or discontinued treatment prior to cycle six for reasons other than study termination.
The median age of all patients was 67 years (range 38-83). At entry, patients generally had poor prognostic disease characteristics (Table 1).
Median baseline spleen volume was 2894 mL (~14 times that reported in the healthy population 28 ) and 93 patients (96%) reported experiencing one or more MFSAF symptom at baseline. The majority (79%) of patients had received ≥2 prior MF-directed therapies, and 13% had received ≥4 MF-directed therapies before study entry. One-third of all patients had baseline platelet counts of 50 to <100 × 10 9 /L. Over onehalf (53%) of patients had baseline hemoglobin levels <10 g/dL and 14% were RBC transfusion-dependent. 29 There were no overt differences in baseline characteristics between the ITT Population and the Stringent Criteria or Sensitivity Analysis cohorts (Table 1).
The duration of spleen response was subject to extensive censoring due to early study termination; follow-up ranged from 0 to  Abbreviations: 95% CI, 95% confidence interval; EOC, end of cycle; ITT, intention-to-treat; JAK2, Janus kinase 2; SVRR, spleen volume response rate. a Per enrolling investigator. One patient was classified as "Other: lack of efficacy." b Relapsed/refractory or intolerant per updated stringent criteria (see Table S1).
Median spleen volume response duration was also NR (95%CI
Serious TEAEs were reported for 33 patients (34%), the most common being pneumonia (4%) and pleural effusion (3%). Eleven patients experienced a serious event that was considered treatmentrelated; pneumonia was the only treatment-related serious TEAE reported for more than one patient (n = 2).
Seven patients (7%) experienced a TEAE that led to death during the treatment period or the 30-day follow-up period. The cause of death was determined to be disease progression in four cases, and the other three cases were due to TEAEs (pneumonia, cardiorespiratory arrest, and shock) that were not considered to be related to study treatment.
Proportions of patients with treatment interruptions of ≥7 days or any fedratinib dose-reduction were 26% and 39%, respectively.
Nineteen patients (20%) permanently discontinued fedratinib due to a TEAE (regardless of causality); diarrhea and thrombocytopenia (n = 2 each) were the only TEAEs leading to discontinuation in >1 patient.
Treatment-related TEAEs led to permanent treatment discontinuation  for 10 patients (10%), eight of whom had a grade 3 or grade 4 treatment-related event (Table S4). Only two patients discontinued fedratinib due to treatment-related anemia or thrombocytopenia (n = 1 each). No report of thrombocytopenia was associated with a major bleeding event.
Acknowledging differences in study designs and the absence of headto-head comparisons, results of JAKARTA2 compare favorably with those for other JAK inhibitors in similar patient populations.
In the current study, more than 90% of patients with baseline platelet counts of 50 to <100 × 10 9 /L received ≥80% of their intended fedratinib dose; treatment was generally tolerable, and spleen volume and symptom response rates were comparable to rates for patients who entered the study with platelet counts ≥100 × 10 9 /L. Similarly, baseline platelet count did not significantly influence spleen response rates in the phase III JAKARTA trial of fedratinib in JAK-inhibitor-naive patients with intermediate-2 or highrisk MF. 32 Even though ruxolitinib, the only other approved JAK inhibitor for MF, can be used at lower doses (5 or 10 mg twice-daily) in patients with MF who have platelet counts of 50 to <100 × 10 9 /L, it may be at the expense of drug efficacy. 9,33 Hematologic events are anticipated with JAK inhibitors based on their mechanism of action. 34 As expected, grade 3 or 4 cytopenias were more commonly reported in patients who began the study with platelet counts of 50 to <100 × 10 9 /L. Importantly, cytopenias were rarely cause for permanent fedratinib treatment discontinuation, suggesting that these events could be managed effectively with dose modifications, temporary treatment interruptions, and transfusions.
No report of thrombocytopenia was associated with a major bleeding event.
The most frequent TEAEs in this study were low-grade gastrointestinal events. Clinical data suggest that taking fedratinib with a highfat meal improves gastrointestinal tolerability with minimal effect on bioavailability. 35  Early study termination of JAKARTA2 may have led to underestimation of fedratinib response. Spleen volume reductions of ≥35% from baseline occurred in a higher proportion of patients who had the opportunity to complete three cycles of fedratinib treatment than the rate reported in the ITT analysis at EOC6; thus, patients who were responding to fedratinib therapy may have been discontinued due to the clinical hold before a cycle six measure was taken and would have been considered nonresponders. Early termination also prevented assessment of longer-term efficacy and safety of fedratinib treatment.
Currently, the longest exposure to fedratinib therapy occurred in the extension portion of a phase I dose-finding and expansion study of fedratinib in adult patients with MF (ClinicalTrials.gov NCT00631462, NCT00724334). 40,41 In an interim analysis from that study, 23 of 59 patients (39%) had received long-term fedratinib treatment for a median 30 cycles (range 13-44) at a median current fedratinib dose of 440 mg. 41 No unexpected safety signals emerged during long-term fedratinib therapy. 41 Long-term outcomes with fedratinib in patients previously treated with ruxolitinib are currently under investigation in the aforementioned FREEDOM and FREEDOM2 studies.
Eligibility criteria for the JAKARTA2 study required a relatively limited degree of ruxolitinib exposure as sufficient to determine ruxolitinib failure at enrollment. However, the median prior ruxolitinib treatment duration in the ITT Population was 10.7 months, and outcomes in the Stringent Criteria and Sensitivity Analysis cohorts, which included patients with greater prior ruxolitinib exposure than initially protocol-specified, were consistent with those of the ITT Population.
This rigorous updated analysis of JAKARTA2 data demonstrates that patients with advanced MF who were substantially pretreated with ruxolitinib could attain robust spleen responses and reduced symptom burden with fedratinib. The efficacy of fedratinib was confirmed in the subgroup of JAKARTA2 patients who met stringent criteria for ruxolitinib relapsed, refractory, or intolerant, and in the sensitivity analysis comprising patients who were least affected by the fedratinib clinical hold and early study termination. Fedratinib is an important new treatment option for patients with MF, particularly those who have previously been treated with Ruxolitinib, as well as those patients with low pretreatment platelet counts or hemoglobin levels.

ACKNOWLEDGMENTS
We thank the patients, families, co-investigators, and all study personnel who participated in this trial. The original study was supported by Sanofi