Patient characteristics and outcomes after ruxolitinib discontinuation in patients with myelofibrosis

Abstract

Background: This retrospective analysis evaluates morbidities, outcomes and associated risk factors in patients with myelofibrosis (MF) after ruxolitinib discontinuation, using Truven Health Analytics MarketScan (TR), Optum integrated virtual electronic health records and claims databases (OP), and Surveillance, Epidemiology, and End Results (SEER)-Medicare linked database (SM).

Methods: A total of 290 patients with MF between 2006 and 2015 (using ICD-9 and ICD-O-3 codes), who were treated with and discontinued ruxolitinib were identified. Only patients with ≥90 days of medical history prior to index diagnosis (TR + OP) and Part A, B, and D enrollment at the time of index diagnosis (SM) were included. Morbidities were assessed during the 30-day period each following ruxolitinib initiation, prior to and post ruxolitinib discontinuation. Cumulative incidence of cytopenias and efficacy outcomes were evaluated from baseline.

Results: Median age of patients was 68 years, with equal proportion of either gender. Median time to ruxolitinib discontinuation was 284 days and median follow-up after discontinuation was 70.9 days. The majority of patients were diagnosed with anemia and >30% of the patients received RBC transfusions during 30-day period prior to and the 30-day period post ruxolitinib discontinuation. After ruxolitinib discontinuation, half of the patients developed cytopenias. The median treatment progression-free survival, and overall survival after ruxolitinib discontinuation were 6.0 (4.4, 8.3) months and 11.1 (8.4, 14.5) months, respectively. Age at ruxolitinib discontinuation (HR [95% CI] = 2.071 [1.320, 3.248]), Charlson Comorbidity Index score (HR [95% CI] = 1.172 [1.093, 1.257]) and gender (HR [95% CI] = 1.620 [1.108, 2.369]) increased the risk of treatment progression (start of the subsequent treatment regimen) or death.

Conclusion: Results from this large, retrospective, US population-based outcome analysis of MF patients show an increase in morbidity burden and identifies the risk factors of survival outcomes among real-world patients who have discontinued ruxolitinib.

Keywords:

• Ruxolitinib
• retrospective analysis
• treatment discontinuation
• myelofibrosis
• survival outcomes

JEL CLASSIFICATION CODES:

• I10
• I18

Introduction

Myelofibrosis (MF), essential thrombocythemia (ET), and polycythemia vera (PV) form a group of clonal hematologic malignancies known as myeloproliferative neoplasms (MPNs), characterized by dysregulated proliferation of the myeloid lineages¹. MF can either be idiopathic, known as primary myelofibrosis (PMF) or occur as the result of ET (Post-ET MF) or PV (Post-PV MF), referred to as secondary myelofibrosis (sMF)².

Advanced age is a known adverse risk factor for MF and majority of patients are diagnosed after 60 years of age^3,4. According to a 2014 estimate based on a 3-year longitudinal analysis, the age-adjusted prevalence of MF in the US ranged from 3.6 to 5.7 per 100,000 persons. Age-adjusted prevalence of PMF was higher (range: 1.3–2.2) than sMF (range: Post-ET MF, 0.46–1.1; Post-PV MF, 0.29–0.67) per 100,000 persons⁵. However, overall survival in the PMF population (45 months) and post-PV (48 months) sMF populations tends to be shorter than that observed in the post-ET population (73 months)⁶.

MPNs are characterized by point mutations in Janus associated kinase 2 (JAK2V617F), CALR and MPL genes which are reported in the majority of patients^7–11. Published guidelines suggest similar treatment approaches for patients with PMF and sMF¹². Ruxolitinib, a JAK 1/2 inhibitor, is the only approved therapy for intermediate-to-high risk MF (PMF + sMF) defined by the International Prognostic Scoring System (IPSS)^13,14.

Despite significant improvements in disease related symptoms and quality of life, over half of the patients discontinue ruxolitinib treatment within 2–3 years due to adverse events or loss of treatment response¹⁵. Recent studies in MF patients following ruxolitinib discontinuation demonstrated poor survival outcomes, as well as low response rates to salvage therapies^16,17.

Research has been conducted to determine the treatment pattern among PMF and sMF patients^16–18, however, patient outcomes associated with treatment discontinuations or failures in these populations have not been well studied, especially using real-world data sources which provide important evidence to support clinical and regulatory decision-making. To address these gaps in the literature, the present study aims to characterize and determine patient outcomes in the MF (PMF + sMF) population who discontinued ruxolitinib treatment. This is a large US population-based study that identifies an important unmet need in MF patients who discontinue ruxolitinib and provides a comprehensive assessment of morbidities, outcomes and associated risk factors in this population.

Methods

Data sources

Data were obtained from three claims databases, the Truven Health Analytics MarketScan (Commercial Claims and Encounters and Truven Medicare) (TR), the Optum (Optum, Inc, Eden Prairie, MN, USA) integrated virtual electronic health records and claims databases (OP), and the Surveillance, Epidemiology, and End Results (SEER)-Medicare linked database (SM). OP and TR databases were used to identify patients with MF diagnosed between 2006 and 2015; the SM database was used to identify patients with MF diagnosed between 2007 and 2014.

The OP database combines adjudicated claims data with Humedica’s Electronic Health Record data, which covers over 10 million lives and includes historical, linked administrative claims data, pharmacy and physician claims, facility claims, with clinical information, inclusive of prescribed and administered medications, lab results, vital signs, body measurements, diagnoses, procedures, and information derived from clinical notes. The TR database comprises inpatient medical, outpatient medical, and outpatient prescription drug claims, diagnosis and encounter data on 70 million lives. The SM database reflects the linkage of two large population-based sources of data that provide detailed information about elderly persons with cancer, which can be used for an array of epidemiological and health services research. Clinical and diagnosis information from the SEER cancer registry (covering approximately one-quarter of the total US population¹⁹) is combined with medical and pharmacy claims for Medicare enrollees. Medicare Part A covers hospital stays, hospice care, and some skilled nursing care; Part B covers services and supplies that are medically necessary to treat a patient’s health condition, such as outpatient care, preventive services, ambulance services, and durable medical equipment; and Part D covers prescription drugs. For each incident cancer diagnosis reported in SEER, patient-level information is captured on demographics, specific International Classification of Diseases for Oncology, Third Revision (ICD-O-3) codes, and survival.

Patient cohort

Patients ≥65 years (categorized as elderly) were included from the SM database, as it is the minimum eligible age to be enrolled in the Medicare database. Patients ≥18 years and <65 years (categorized as non-elderly) were included from the TR and OP databases. Including different age groups across the data sources ensured both adequate representation of all age groups and elimination of possible duplication patients. Also, those patients who had ≥90 days of medical history prior to index diagnosis (TR and OP) and who were enrolled in Part A, B, and D at the time of index diagnosis (SM) were included in the analysis. Patients who had PMF diagnosis in Medicare database but did not have a PMF diagnosis in the SEER registry were required to have a minimum of 6 months of prior enrollment to ensure index diagnosis.

Patients with MF were identified according to ICD-9 (TR, OP, SM) and ICD-O-3 (SM) codes. Patients with MF diagnosis (ICD-9: 238.76, 289.83; ICD-O-3: 9961/3) without prior history of PV (ICD- 9: 238.4, ICD-O-3: 9950/3) or ET (ICD-9: 238.71, ICD-O-3: 9962/3) were classified as PMF, while those with prior history of PV or ET were classified as sMF patients (Figure 1).

Figure 1. Cohort selection. Abbreviations. PMF, Primary Myelofibrosis; sMF, Secondary Myelofibrosis; PV, Polycythemia Vera; ET, Essential Thrombocythemia; SM, SEER Medicare; TR, Truven; OP, Optum.

Among PMF and sMF patients, lines of treatments (LOTs) were identified from index diagnosis date on the basis of business rules centered on continuity and concurrence of usage of relevant drugs (chemotherapy, erythropoiesis stimulating agents, hydroxyurea, ruxolitinib).

Patient characteristics and outcomes of interest were evaluated in the PMF + sMF population who experienced ruxolitinib discontinuation.

The study protocol was approved by Institutional Review Board (Fox Commercial IRB, Ltd). Because no information was collected on human subjects (all data were retrospective, de-identified, and non-interventional), the study was exempted from patient consent requirements. The data and protocol of this study is not accessible publicly, hence please contact [email protected] for the original data or protocol.

Outcomes of interest

Baseline patient characteristics were summarized at ruxolitinib discontinuation date. Patients were followed up from ruxolitinib discontinuation date to death date, in case of death, or end of follow-up (defined as 31 December 2014 and last claim date in the SM and OP + TR databases, respectively), in others. Descriptive measures of morbidities and red blood cell (RBC) transfusion (codes provided in supplementary file) were assessed during the 30, 60, 90, and 180-day period following ruxolitinib initiation, prior to ruxolitinib discontinuation, and post ruxolitinib discontinuation date. Cumulative incidence of cytopenias, including anemia (ICD-9 code 285.22), neutropenia (ICD-9 code 288.0) and thrombocytopenia (ICD-9 code 287.3, 287.4, 287.5) were evaluated from baseline. Efficacy outcomes including, time to next treatment (TTNT; start of next LOT was defined as an event), treatment progression-free survival (TPFS; start of next LOT or death, whichever occurred first was defined as an event), and overall survival (OS; death was defined as an event) were also evaluated from baseline.

Statistical analysis

Frequencies were used to summarize categorical and, mean (standard deviation, SD) and median (interquartile range, IQR) were used to summarize continuous variables. Cumulative incidence of cytopenias (with death as a competing risk event) was evaluated using Fine and Gray’s method²⁰. Time-to-event outcomes (TTNT, TPFS and OS) were assessed using Cox proportional hazard models. Median OS was evaluated using Kaplan Meier analysis. All analyses were conducted using SAS^® (version 9.4) statistical package (SAS Institute Inc., Cary, North Carolina, USA).

Results

Baseline characteristics

The analysis included 290 MF (PMF = 125, sMF = 165) patients with ruxolitinib discontinuation, from all the three databases (OP + TR [n = 127] + SM [n = 163]). The median (IQR) age of patients at ruxolitinib discontinuation was 68.0 (62.0−77.0) years. Patients from <45–64 years age group comprised the greatest proportion (39%) of the cohort. Half of the patients were men. Median (IQR) time to ruxolitinib discontinuation from ruxolitinib initiation date was 284 (113.0−562.0) days. Median (IQR) follow-up time after ruxolitinib discontinuation was 70.9 (6.9−251.8) days. Median (IQR) number of prior LOTs was 1.0 (0.0−1.0). Majority of patients (83.1%) did not receive any supportive care treatment (that included erythropoietin ± steroids [danazol, fluoxymesterone, prednisolone, or dexamethasone]) after ruxolitinib was discontinued, while a few received recombinant erythropoietin (2.4%), hydroxyurea (4.5%), or chemotherapy (0.3%) (Table 1).

Table 1. Baseline characteristics.

	Cohort with ruxolitinib discontinuation, (N = 290)
Age, years
Mean (SD)	69.0 (10.2)
Median (IQR)	68.0 (62.0−77.0)
Minimum, Maximum	39, 93
Age at ruxolitinib discontinuation, years, n (%)
<45−64	113 (39)
65−74	83 (29)
75−84	77 (27)
≥85	17 (6)
Gender, n (%)
Male	145 (50)
Number of prior LOTs
Mean (SD)	0.9 (1.1)
Median (IQR)	1.0 (0–1.0)
Minimum, Maximum	0, 6
Number of days from ruxolitinib initiation date to discontinuation date
Mean (SD)	372 (299.0)
Median (IQR)	284 (113.0−562.0)
Minimum, Maximum	0, 1225
Number of days from ruxolitinib discontinuation date to end of follow-up
Mean (SD)	167.8 (213.9)
Median (IQR)	70.9 (6.9−251.8)
Minimum, Maximum	0, 979
Post ruxolitinib regimens containing, n (%)
Erythropoietin, Hydroxyurea	20 (6.89)
Others*	29 (10.00)
None of the above	241 (83.10)

Note: Patients exposed to ruxolitinib after index MF diagnosis were included in this table. Ruxolitinib discontinuation date is defined as the last date on which a patient was exposed to ruxolitinib.

Percentages are based on N within the column header.

Abbreviations. SD, Standard deviation; LOTs, Line of treatments; MF, Myelofibrosis.

Regimens defined as *Others include radiation, best supportive care, steroids only, hormones, immunomodulatory drugs (e.g. lenalidomide and thalidomide).

Morbidities and transfusions

During the 30-day period following ruxolitinib initiation, 36% had anemia, 12% had thrombocytopenia, and 2% had neutropenia. The proportion of patients diagnosed with these morbidities increased during the 30-day period prior to ruxolitinib discontinuation (anemia [45%], and thrombocytopenia [18%]). The comorbidities increased further in the 30-day period after ruxolitinib discontinuation (AML [7%], anemia [53%], thrombocytopenia [23%], and neutropenia [10%]). Only 19% patients were found to receive RBC transfusion within first 30 days of ruxolitinib initiation, compared to, 33% each during the 30 days prior to and after ruxolitinib discontinuation. The proportion of patients diagnosed with any morbidity or those who received RBC transfusion increased with time at every stage of assessment (following/prior to/or after ruxolitinib initiation) (Table 2).

Table 2. Treatment emergent morbidities and transfusions.

Event	Timepoint of assessment
	Following ruxolitinib initiation (baseline) (N = 290)				Prior to ruxolitinib discontinuation* (N = 290)				After ruxolitinib discontinuation^† (N = 163)
	30 Days	60 Days	90 Days	180 Days	30 Days	60 Days	90 Days	180 Days	30 Days	60 Days	90 Days	180 Days
Diagnosis of AML, n (%)**				15 (5)	12 (4)	13 (4)	14 (5)	17 (6)	12 (7)	14 (9)	16 (10)	19 (12)
Diagnosis of anemia, n (%)	105 (36)	153 (53)	174 (60)	191 (66)	131 (45)	163 (56)	183 (63)	203 (70)	86 (53)	106 (65)	112 (69)	125 (77)
Diagnosis of thrombocytopenia, n (%)	31 (11)	53 (18)	66 (23)	84 (29)	51 (18)	70 (24)	83 (29)	100 (34)	38 (23)	54 (33)	58 (36)	69 (42)
Diagnosis of neutropenia, n (%)**			12 (4)	19 (7)		12 (4)	15 (5)	18 (6)	16 (10)	18 (11)	20 (12)	27 (17)
Diagnosis of RBC transfusion, n (%)	54 (19)	103 (36)	116 (40)	138 (48)	97 (33)	108 (37)	118 (41)	135 (47)	53 (33)	68 (42)	75 (46)	86 (53)

Note: Patients exposed to ruxolitinib after index MF diagnosis were included in this table. Ruxolitinib discontinuation date is defined as the last date on which a patient was exposed to ruxolitinib. Percentages are based on N within the column header.

**Values of 1–10 patients in any cell in this table have been suppressed to conform with CMS Cell Size Suppression Policy.

Abbreviations. AML, Acute Myeloid Leukemia; MF, Myelofibrosis; RBC, Red Blood Cells.

*This time point includes the date of ruxolitinib discontinuation.

^†This time point does not include date of ruxolitinib discontinuation. N is based on patients with 60 days of follow up after ruxolitinib discontinuation date.

Cumulative incidence of cytopenias

Half of the patients who discontinued ruxolitinib developed cytopenias at some time during their remaining follow-up. Among all patients, 11% had death as the competing event. Cumulative incidence for all the cytopenias (in the presence of death as a competing event) reached 75% in the 18th month after ruxolitinib discontinuation (Figure 2), after which it remained steady.

Figure 2. Cumulative incidence of cytopenias in the presence of death as a competing event. The Y-axis is the cumulative probability of development of new cytopenia and the x-axis is time of first diagnosis of cytopenia. Cytopenias include anemia, neutropenia and thrombocytopenia.

Efficacy outcomes

After ruxolitinib discontinuation, the median (95% CI) TTNT was NE (11.2, NE), TPFS was 6.0 (4.4−8.3) months, and OS was 11.1 (8.4−14.5) months (Figure 3). Female gender was associated with increased hazard of progression to next treatment (HR [95% CI] = 1.853 [1.007, 3.408]). Age >65 years at time of ruxolitinib discontinuation (HR [95% CI] = 2.071 [1.320, 3.248], unit increase in Charlson Comorbidity Index (CCI) score (HR [95% CI] = 1.172 [1.093, 1.257]) and female gender (HR [95% CI] = 1.620 [1.108, 2.369]) were associated with increased hazard of treatment progression/death. In addition, age >65 (HR [95% CI] = 3.845 [2.118, 6.981]), and unit increase in CCI score (HR [95% CI] = 1.248 [1.154, 1.349]) were associated with increased hazard of death (Table 3).

Figure 3. OS, TPFS and TTNT of patients. Time to event analysis in the sample of myelofibrosis patients that discontinued ruxolitinib (n = 290). X-axis represents the event free probability and y-axis represents the follow-up time after ruxolitinib discontinuation. Event = death (Overall survival [OS] analysis), earlier of treatment progression/death (time to progression free survival [TPFS] analysis) and treatment progression (time to next treatment [TTNT] analysis).

Table 3. Time to event efficacy outcomes.

	Time to next treatment^∗	Treatment progression-free survival^†	Overall survival^‡
Summary of censored and event cases
Total number of patients assessed	290	290	290
Number of patients with an event	49	123	102
Proportion of patients with an event, (%)	16.90	42.40	35
Kaplan-Meier statistics, months
Median (95% CI)	NE (11.2, NE)	6.0 (4.4, 8.3)	11.1 (8.4–14.5)^∗∗
Mean (SE)	10.828 (0.563)	9.495 (0.922)	13.688 (0.970)
Quarterly rates, %
3 month rate	80.80	68.70	84
6 month rate	73.60	50.50	69
9 month rate	65.60	37.70	55
12 month rate	61.30	27.10	47
Predictors of time to event analysis, HR (CI)
Age at ruxolitinib discontinuation date (Reference, ≤65 years)	1.311 (0.696, 2.469)	2.071 (1.320, 3.248)	3.845 (2.118, 6.981)
Prior RBC transfusion (Reference, No RBC transfusion)	1.046 (0.545, 2.007)	0.910 (0.610, 1.356)	0.833 (0.527, 1.316)
Charlson comorbidity index score	1.111 (0.994, 1.242)	1.172 (1.093, 1.257)	1.248 (1.154, 1.349)
Gender (Reference, Female)	1.853 (1.007, 3.408)	1.620 (1.108, 2.369)	1.324 (0.864, 2.029)

Note: Erythropoietin, hydroxyurea, chemotherapy, radiation, best supportive care, steroids, hormones, immunomodulatory drugs (e.g. lenalidomide and thalidomide) were considered as progression therapies.

Comorbidities included for calculating Charlson Comorbidity Index score are myocardial infarction, congestive heart failure, peripheral vascular disease, cerebrovascular disease, dementia, chronic pulmonary disease, rheumatic disease, peptic ulcer disease, mild liver disease, diabetes without chronic complications, diabetes with chronic complications, hemiplegia or paraplegia, renal disease, any malignancy without metastasis, leukemia, lymphoma, moderate or severe liver disease, metastatic solid tumor, and acquired immunodeficiency syndrome (excluded a symptomatic infection).

Abbreviations. CI, Confidence interval; HR, hazard ratio; NR, Not Estimable; SE, standard error; RBC, Red blood cells.

*Patients observed with any of the treatments mentioned above were considered to have an event. The event date is the earliest date on which a patient received a progression therapy. Patients without an observed event were censored on enrollment end date.

^†Patients with an observed death or progression to any of the treatments mentioned above were considered to have an event. The event date is the earliest date on which a patient received a progression therapy. If a patient did not receive a progression therapy but died, then the event date is date of death. Patients without an observed event were censored on enrollment end date.

^‡Patients with an observed death were considered to have an event. Patients without an observed death were censored on 31DEC2014 in SEER Medicare data and last claim date in Optum and Truven studies.

**After excluding patients who developed AML on or 30 days before ruxolitinib discontinuation date, median overall survival was 10.68 (8.21–14.49) months.

Discussion

The present retrospective analysis evaluated various outcomes before and after ruxolitinib discontinuation. The results show that after discontinuation of ruxolitinib, a greater proportion of patients experienced morbidities resulting in worsened outcomes. MF patients had poor efficacy outcomes post ruxolitinib discontinuation with median TPFS of <6 months and median OS of <12 months likely due to limited treatment modalities available after ruxolitinib discontinuation. These findings are aligned with the results obtained in two real-world studies demonstrating the unmet need to improve the safety and efficacy outcomes associated with the treatments used after ruxolitinib discontinuation^4,21. In this study, median time to ruxolitinib discontinuation was <1 year, which is similar to the results obtained from another study based on US claims database²². However, longer median time to ruxolitinib discontinuation (3 years) in COMFORT trials^14,23 could be due to the inclusion of healthier patients in controlled studies as compared to real-world population.

To our knowledge, this is the first population-based study assessing morbidities in a treatment setting following ruxolitinib discontinuation. In the present study, during ruxolitinib treatment, anemia (36%−45%) was the most frequently reported hematological complication morbidity followed by thrombocytopenia (12%−18%) and neutropenia (5%−9%) as assessed by ICD-9 codes. Study findings are aligned with the COMFORT-1 trial, where 45.2% of patients developed anemia, and 12.9% and 7.1% of patients developed thrombocytopenia and neutropenia, respectively¹⁴. Also, a real-world study that retrospectively reviewed medical records data showed that, only 22.2% patients experienced anemia and 7.4% experienced thrombocytopenia while being treated with ruxolitinib²⁴. However, findings from the current study point to a higher burden of cytopenias after ruxolitinib discontinuation compared to the 30-day period prior to and the 30-day period post ruxolitinib discontinuation. Although differing from the current study, Cervantes et al.²³ reported occurrence of cytopenias within 28 days after ruxolitinib discontinuation to be similar to that reported during treatment. The greater cytopenic burden we observed after discontinuation could be attributed to low platelet counts at the time of discontinuation, as such patients often have a poor prognosis¹⁷. This outcome could be a result of disease progression as also observed by Newberry et al.¹⁷. Therefore, patients who are truly failing ruxolitinib due to progressive disease may not experience a rebound of platelet count on discontinuing JAK inhibitors as this outcome is a function of the disease rather than treatment-related cytopenias.

Although ruxolitinib can worsen anemia in some patients, previous safety studies do not suggest presence of anemia as a contraindication for ruxolitinib use; where it does occur, it can be mitigated by ruxolitinib dose reductions and RBC transfusions^25–27. In a real-world study, 12% of patients received RBC transfusions during ruxolitinib treatment, while the present study showed that 30% of the patients required transfusion post ruxolitinib discontinuation in order to manage anemia. Similar results were observed in a retrospective chart review study, in which the transfusion dependency in MF patients increased from 18% at the start of ruxolitinib to 36% after discontinuation¹⁷. These findings are consistent with progressive MF disease and highlight a need for novel treatment approaches to both effectively manage cytopenias post ruxolitinib discontinuation and to extend survival.

A comprehensive review of past literature did not reveal any previous study that sought to evaluate time to next treatment and treatment progression-free survival in MF patients after discontinuation of ruxolitinib treatment. Until now, OS is the only efficacy outcome assessed in real-world populations who discontinued ruxolitinib treatment. The median OS observed in the present study (11.1 months) is consistent with the OS in previously reported real-world single center studies by Kuykendall et al. (13 months)¹⁶ and Newberry et al. (14 months)¹⁷. However, another real-world study conducted using claims data reported median OS of 30 months in MF patients on ruxolitinib therapy, while in patients who failed or discontinued ruxolitinib was 7 months^4,17.

Unlike past studies that were separately conducted in elderly and non-elderly cohorts^4,21, the current study is generalizable to patients of both age groups as a population-based study. Also, the present study provides the largest data sample that has been available to determine the OS of patients who discontinued ruxolitinib. A majority of the data (∼56%) was drawn from a nationally representative, population-based cancer registry linked to Medicare, which provides health insurance to over 95% of the elderly in the US²⁸. Hence, diagnosis of MF, use of ruxolitinib treatment, and record of mortality can be considered reliable and accurate²⁹. Moreover, variables derived from SM and TR + OP databases were complete hence, the results are not affected by missing values.

The study findings should be considered in light of several limitations. Due to the observational nature of the research, no cause and effect relationship could be determined in the present study. Among patients who received at least one LOT, only 20% received ruxolitinib and were included in this study. The low use of ruxolitinib could be attributed to its approval only during the latter half of the study period (November 2011) as well as the population in which it was approved for use (intermediate or high-risk MF only). Moreover, it reflects the low usage of ruxolitinib in clinical practice. Information regarding important determinants of ruxolitinib discontinuation and OS (e.g. IPSS, cytogenetic and mutational profile) were unknown. The state of the disease (chronic/blast phase) as well as the reasons for ruxolitinib discontinuation were unavailable in the databases. The treatment regimens as identified in our study may only reflect standard of care up to 2014 in the SM population and up to 2015 in the TR + OP population. Mortality data in the administrative claims databases may be incomplete, as these data rely on completeness of linkage to national death registries as well as compliance of reporting of death information. Also, a majority of patients in the study did not receive subsequent therapy post ruxolitinib discontinuation; this may be due to inadequate follow-up, lack of effective treatment options or patient choice to discontinue active treatment.

Conclusions

In summary, we found that morbidities worsened after ruxolitinib discontinuation. With limited treatment options being available in the real-world post ruxolitinib use, our study findings underscore the need for more treatment options to manage MF after ruxolitinib is discontinued. We also found that efficacy outcomes were associated with age at ruxolitinib discontinuation (TPFS, OS), CCI score (TPFS, OS), and gender (TTNT, TPFS). Further research is warranted to not only assess survival over a longer period of time, but to also critically examine potential prognostic variables (i.e. patient and clinical factors) which affect survival. Such information may further assist clinicians to evaluate the optimal care for MF and also determine effective treatment strategies to manage all the MF subtypes in a population which experiences ruxolitinib discontinuation.

Transparency

Declaration of funding

This study was supported by research funding from Janssen Global Services LLC.

Declaration of financial/other relationships

MM, CL and JH are employees of Janssen Global Services, LLC.

JME peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Author contributions

All authors were involved in the study design, interpretation of data and drafting the manuscript. JH, RP & MM were involved in data collection and statistical analysis.

Ethics approval and consent to participate

The study protocol was approved by Institutional Review Board (Fox Commercial IRB, Ltd). Because no information was collected on human subjects (all data were retrospective, de-identified, and non-interventional), the study was exempted from patient consent requirements.

Acknowledgements

The authors acknowledge Rishabh Pandey, PhD (SIRO Clinpharm Pvt. Ltd.) for providing writing assistance and Harry Ma, PhD (Janssen Global Services, LLC) for additional editorial support.

Availability of data and material

The data and protocol of this study are not publicly available but are available upon contact with [email protected] on reasonable request.