Real-World Outcomes of Ruxolitinib Treatment for Polycythemia Vera

Abstract

This is a retrospective study of patients with polycythemia vera treated with ruxolitinib. Data was collected from 11 participating centers across the United States. Ruxolitinib demonstrated efficacy in reducing phlebotomy requirements and palpable splenomegaly. This is the first real-world review of patients with polycythemia vera treated with ruxolitinib, and the results are similar to previously published clinical trials.

Introduction:

Ruxolitinib is approved for the treatment of polycythemia vera (PV) with hydroxyurea resistance or intolerance. Approval was based on the phase III RESPONSE trial, which demonstrated efficacy in a highly selected patient population.

Materials and Methods:

To characterize the tolerability and outcomes of ruxolitinib outside of a clinical trial, we performed a multi-center retrospective analysis of patients with PV treated with ruxolitinib at 11 participating sites across the United States. Outcomes of interest included change in phlebotomy requirements after starting ruxolitinib and spleen response, as these were included in the primary composite outcome in the RESPONSE trial.

Results:

One hundred twenty-six patients met eligibility criteria, and the median duration of follow-up was 22.4 months (range, 0–63.0 months). At 32 weeks after starting ruxolitinib, the percentage of patients who received at least 1 phlebotomy was significantly decreased compared with before ruxolitinib (37% vs. 56%; relative risk [RR], 0.66; 95% confidence interval [CI], 0.52–0.84; P < .001). Phlebotomy requirements were similarly decreased in patients who had received at least 3 phlebotomies prior to ruxolitinib initiation (28% vs. 17%; RR, 1.65; 95% CI, 1.13–2.40; P < .01). Resolution of palpable splenomegaly was also documented (48% vs. 20%; RR, 2.45; 95% CI, 1.70–3.53; P < .0001). A total of 9.5% of patients discontinued ruxolitinib owing to treatment-emergent adverse events, and 81.7% of patients were receiving ruxolitinib at last known follow-up.

Conclusion:

These real-world results are similar to those reported from the RESPONSE trial, although additional follow-up is necessary to assess long-term outcomes and potential for late-onset toxicity.

Introduction

Polycythemia vera (PV) is a chronic myeloproliferative neoplasm (MPN) that is characterized by a clonal proliferation of red blood cells, resulting in an increased erythrocyte mass and frequently concomitant increases in the white blood cell and platelet counts. PV is associated with a significant symptom burden, decreased quality of life,¹ increased risk of venous and arterial thromboses,² hemorrhagic complications,³ and progression to myelofibrosis (MF) and/or leukemic transformation.⁴ Multiple epidemiologic studies have demonstrated that patients with PV experience a shortened survival and excess mortality when compared with age- and gender-matched controls.^5–7 Nearly all patients with PV harbor an activating somatic mutation in Janus Associated Kinase (JAK) 2, a member of the JAK-STAT family of signal transduction proteins that is directly linked to erythropoietin receptor signaling. Greater than 95% of patients with PV harbor a JAK2 exon 14 mutation (JAK2V617F), whereas the remainder of patients harbor JAK2 mutations at other loci, most commonly at exon 12.^8,9

Management of PV consists of daily low-dose aspirin and intermittent phlebotomy to maintain hematocrit levels of less than 45%.^10,11 Cytoreductive therapy is indicated for “high-risk” patients (age > 60 years or history of a thrombohemorrhagic event) and is often considered for patients who cannot tolerate therapeutic phlebotomy, have burdensome splenomegaly, or have vascular comorbidities.¹² Hydroxyurea (HU) is the consensus recommended first-line agent for cytoreduction in appropriate patients with PV.¹³ Resistance or intolerance to HU occurs in approximately 25% of patients with PV, and as many as one-third to one-half of patients who initially respond to HU may lose response over time.¹⁴ Additionally, HU resistance has been associated with a shortened survival in patients with PV. Alternative cytoreductive agents, which include pegylated interferon-α and busulfan, are less well-studied and have toxicity profiles that differ from HU and require individualized decision-making.^15–17

Ruxolitinib is a small-molecule inhibitor of JAK1/2 that targets the dysregulated JAK-STAT pathway in PV.¹⁸ Ruxolitinib is United States (US) Food and Drug Administration (FDA)-approved for the treatment of intermediate or high-risk MF, based on the results of the COMFORT-I and COMFORT-II trials, where it was shown to produce durable reductions in splenomegaly and improvement in disease-related symptoms.^19,20 More recently, ruxolitinib has gained FDA approval for the treatment of PV in the setting of HU resistance or intolerance, largely based on the positive results of the RESPONSE trial and follow-up analyses.^21,22 Additional evidence of ruxolitinib benefit for the treatment of PV came from the RESPONSE-2 trial, which demonstrated efficacy in a similar patient population lacking palpable splenomegaly.²³ The RESPONSE trials were international, randomized, multicenter phase III studies whose results may not be entirely translatable to common clinical practice. In an attempt to bridge the gap between a clinical trial and real-word practice outcomes, we conducted a US-based, multicenter, retrospective review of patients with PV who were treated with ruxolitinib.

Materials and Methods

This study was initially approved by the Program for Protection of Human Subjects at the Icahn School of Medicine at Mount Sinai and subsequently at internal review boards at each participating center. The study included 126 patients with World Health Organization-defined PV who were treated with ruxolitinib at one of 11 participating sites in the US between January 1, 2011 and January 1, 2018. Inclusion criteria included: 18 years of age or older and a diagnosis of PV without progression to MF or other hematologic malignancy at the time of initiation of ruxolitinib therapy. HU resistance or intolerance was not an inclusion criterion for this study, nor was concomitant administration of HU and ruxolitinib an exclusion criterion. Classification of patients as intolerant to HU was based on investigator description from the electronic medical record. Data from each participating site was centralized and stored using Research Electronic Data Capture (REDCap), an online tool for data aggregation and storage.²⁴

Baseline demographic information including age and gender, previous thrombohemorrhagic (thrombotic or hemorrhagic) events, JAK2 variant allele fraction, phlebotomy requirements within the prior 24 weeks of ruxolitinib initiation, palpable splenomegaly, and cytoreductive medication history both prior to and at the initiation of ruxolitinib were collected. Primary outcome measures included phlebotomy requirements (defined as the number of therapeutic phlebotomy events) during the 32 weeks after initiation of ruxolitinib, change in palpable splenomegaly after 32 weeks of ruxolitinib therapy, thrombohemorrhagic complications, and myelofibrotic and leukemic transformation while receiving ruxolitinib. The intervals of 24 and 32 weeks were chosen to assess for phlebotomy requirements prior to and after initiation of ruxolitinib, respectively, as the RESPONSE trial utilized the same intervals to assess for freedom from phlebotomy. RESPONSE assessed phlebotomy dependence between weeks 8 and 32, to allow for ruxolitinib dose titration during the first 8 weeks after starting the medication. We assessed phlebotomy requirements during the complete 32 weeks after starting ruxolitinib, given the difficulty of documenting accurate ruxolitinib dose titration with the retrospective nature of this study. Additionally, phlebotomies were recorded during the first 8 weeks of ruxolitinib therapy in RESPONSE, and patients needing > 1 phlebotomy during that time period were classified as phlebotomy-dependent regardless of their subsequent phlebotomy requirements during weeks 8 to 32. Symptom burden assessment for fatigue, pruritus, and night sweats was based on retrospective review of outpatient documentation. Patients whose medical records showed no indication of symptoms were classified as having ‘no symptoms,’ whereas patients whose medical records showed an indication of any degree of symptoms were classified as experiencing ‘some degree of symptoms.’ Outcomes for phlebotomy requirements, splenomegaly, and symptom burden are based upon data available from the retrospective study, and the variable denominators for these outcome measures are based upon availability of patient data.

Continuous variables were summarized by the median and range, and categorical variables were summarized by number and percentage. Generalized estimating equations with a log-link function and binomial distribution were used to estimate the relative risks (RRs) and corresponding 95% confidence intervals (CIs) comparing incidence of phlebotomy requirements, splenomegaly, and symptom burden before and after 32 weeks of ruxolitinib. A mixed model analysis of variance was used to estimate the changes in natural log transformed blood counts (hematocrit, white blood cells, and platelets) from baseline to 32 weeks post-ruxolitinib initiation. Compound symmetric covariance structures were assumed in repeated measures modeling to account for intrasubject correlation. All hypothesis testing was 2-sided and conducted at the 5% level of significance. Statistical analyses were performed with the SAS, version 9.4 (SAS Institute Inc, Cary, NC) software package.

Results

Baseline Characteristics

In total, 126 patients were identified that met eligibility criteria. The cohort was 47% female with a mean age of 63 years (range, 24–92 years) at the initiation of ruxolitinib. A total of 107 (85%) patients had been treated with HU prior to starting ruxolitinib, and 17 of those patients were still receiving HU at the time of ruxolitinib initiation. Other than HU, the most common cytoreductive agents received prior to ruxolitinib therapy were pegylated interferon-α (27 patients; 21%) and anagrelide (16 patients; 13%). Six (5%) patients had previously been treated with multiple cytoreductive agents. Additional baseline data, including JAK2 mutation status, mean variant allele frequency, and HU use, can be found in Table 1.

Table 1.

Baseline Disease Characteristics (N = 126)

Baseline Characteristics	N (%)
Female	59 (47)
Prior venous thrombosis
Yes	23 (18)
No	102 (82)
Missing	1 (<1)
Prior arterial thrombosis
Yes	28 (22)
No	97 (78)
Missing	1 (<1)
Prior hemorrhage
Yes	12 (10)
No	113 (90)
Missing	1 (<1)
JAK2 mutation status
Yes	118 (96)
No	5 (4)
Missing	3 (2)
JAK2 mutation (n = 118)
JAK2V617F	116 (98)
JAK exon 12	2 (2)
Median JAK2V617F variant allele fraction, % [min, max]	50 [4, 98]
Previous hydroxyurea therapy
Yes	107 (85)
No	19 (15)
Discontinued hydroxyurea therapy
Yes	90 (84)
Discontinued owing to resistance	25 (28)
Discontinued owing to intolerance	54 (60)
Discontinued owing to other reason	11 (12)
No	17 (16)
Additional prior polycythemia vera therapies
Yes	47 (37)
Pegylated interferon-α	27 (57)
Anagrelide	16 (34)
Other	7 (15)
No	79 (63)

The starting dose of ruxolitinib in 90 (71%) patients was 10 mg twice daily (BID), 5 mg BID in 18 (14%) patients, 15 mg BID in 9 (7%) patients, and 20 mg BID in 5 (4%) patients. The most common dose of ruxolitinib at 32-week follow-up in the 94 evaluable patients was 10 mg BID in 32 (34%) patients. Other doses of ruxolitinib utilized at 32-week follow-up were 15 mg BID in 21 (22%), 5 mg BID in 14 (15%), and 20 mg BID in 11 (12%) patients.

The median duration from initiation of ruxolitinib treatment to last known follow-up was 22.4 months (range, 0–63.0 months). Five patients did not have follow-up recorded after starting ruxolitinib. Overall, 103 (81.7%) patients were receiving ruxolitinib at their last known follow-up. Eighteen patients discontinued ruxolitinib by the time of their last known follow-up. These 18 patients had received ruxolitinib for a median of 188 days (range, 6–1245 days). Reasons for discontinuation included treatment-emergent adverse events (TEAEs) in 12 (9.5%) patients, unknown/not documented reasons in 5 (4.0%) patients, and financial toxicity in 1 (0.8%) patient. TEAEs leading to ruxolitinib discontinuation were varied, and a comprehensive list can be found in Table 2.

Table 2.

Treatment-emergent Adverse Events Leading to Ruxolitinib Discontinuation

Patient No.a	Adverse Event
3	Headache
7	Weight gain
11	Dizziness, gastrointestinal disturbance
23	Headache
53	Anemia
103	Weakness, memory problems, depression
109	Hypersensitivity pneumonitis
122	Herpes zoster infection
201	Not detailed
302	Muscle cramps, fatigue, dizziness, tinnitus
304	Arthralgias
403	Arthralgias, breast tenderness, headaches

^aPatient number correlates to pre-designated patient identifiers based on contributing institution and do not reflect the sample size of this study

Efficacy Measures

Therapeutic Phlebotomy.

With regards to phlebotomy requirements, 53 (56%) of 94 evaluable patients receiving ruxolitinib had no phlebotomy requirements at the 32-week follow-up compared with 44 (37%) of 118 evaluable patients who had no phlebotomy requirements in the 24 weeks prior to starting ruxolitinib (RR, 0.66; 95% CI, 0.52–0.84; P < .001). A greater number of patients required ≥ 1 phlebotomy within the 24-week period prior to the initiation of ruxolitinib as compared with the 32-week follow-up (63% vs. 43%; RR, 1.45; 95% CI, 1.15–1.82; P < .01). Phlebotomy rate was similarly higher in the pre- versus post-groups when evaluating at least 2 phlebotomies (45% vs. 27%; RR, 1.70; 95% CI, 1.23–2.33; P < .002) and at least 3 phlebotomies (28% vs. 17%; RR, 1.65; 95% CI, 1.13–2.40; P < .01). Complete results of therapeutic phlebotomy requirements can be found in Table 3.

Table 3.

Baseline Disease Data and 32-week Outcomes

Disease Characteristics	Baselinea	32 Weeksb	Risk Ratio (95% Confidence Interval)	P Value
Phlebotomy requirements, n (%)	N = 126	N = 94
0	44 (37)	53 (56)	0.66 (0.52–0.84)	.0008
≥1	74 (63)	41 (44)	1.45 (1.15–1.82)	.0015
≥2	53 (45)	25 (27)	1.70 (1.23–2.33)	.0011
≥3	33 (28)	17 (18)	1.65 (1.13–2.40)	.0094
Missing	8 (6)	32 (28)
Splenomegaly, n (%)		N = 90
Palpable splenomegaly
Yes	55 (48)	18 (20)	2.45 (1.70–3.53)	<.0001
No	59 (52)	72 (80)
Missing	12 (9)	36 (29)
Spleen length, cm (min, max)	4.9 (0, 40)	2.5 (0, 30)	2.00 (1.36–2.96)	.0005
Symptom burden, n (%)		N = 97
Night sweats	35 (28)	9 (9)	2.97 (1.64–5.40)	.0004
Pruritis	49 (40)	18 (19)	2.17 (1.54–3.05)	<.0001
Fatigue	78 (63)	41 (42)	1.40 (1.14–1.71)	.0013
Missing	2 (2)	29 (23)
Geometric mean blood counts (min, max)
Hematocrit, %	41.80 (20.5, 60.2)	38.49 (23.1, 56.6)	1.09 (1.06–1.12)	<.0001
White blood cell count, × 103/uL	10.67 (2.9, 64.7)	8.69 (1.8, 41.3)	1.23 (1.10–1.38)	.0005
Platelet count, × 103/uL	328.19 (21, 2196)	312.53 (14, 2196)	1.05 (0.9–1.22)	.5157

^aBaseline assessed within 24 weeks of starting ruxolitinib

^b32-week treatment data available for 94 patients for phlebotomy requirements, 90 patients for spleen assessment, and 97 patients for symptom burden

Splenomegaly.

The prevalence of palpable splenomegaly was decreased at 32-week follow-up (48% vs. 20%; RR, 2.45; 95% CI, 1.70–3.53; P < .0001) (Figure 1). The mean palpable spleen length measured below the left costal margin was higher in the pre-ruxolitinib period than in the 32-week follow-up periods (4.9 cm vs. 2.5 cm; RR, 2.00; 95% CI, 1.35–2.96; P = .0005).

Figure 1.

Splenomegaly Responses of Patients With Polycythemia Vera Treated With Ruxolitinib. Percentage Change in Palpable Splenomegaly From Baseline to the 32-Week Assessment after Starting Ruxolitinib Therapy. Splenomegaly was Assessed by Palpation During Clinic Visits and Recorded in cm Below the Left Costal Margin. Responses are Colored Coded by whether the Baseline Spleen was Non-palpable, < 10 cm, or ≥ 10 cm. None of the Patients With Non-palpable Spleens at Baseline Developed Palpable Spleens while Receiving Ruxolitinib Treatment

Symptoms.

Patients experienced a reduction in symptom burden from baseline to 32 weeks of ruxolitinib treatment with regards to the pre-specified complaints of night sweats, pruritis, and fatigue. Night sweats were documented in 28% of patients at baseline versus 7% at their 32-week follow-up (RR, 3.89; 95% CI, 2.08–7.26; P < .0001). Pruritis was documented in 39% of patients at baseline versus 14% at their 32-week follow-up (RR, 2.72; 95% CI, 1.86–3.98; P < .0001). Fatigue was documented in 62% of patients at baseline versus 33% at their 32-week follow-up (RR, 1.90; 95% CI, 1.49–2.44; P < .0001).

Safety

Twelve patients discontinued ruxolitinib treatment owing to adverse events (Table 2). Seventeen (13.5%) patients experienced headaches after starting ruxolitinib, 12 (9.5%) patients experienced dizziness, and 7 (5.6%) patients reported new onset diarrhea. Nine (7.1%) patients had documented infections while receiving ruxolitinib. Herpes zoster (n = 2), pneumonia (n = 2), skin or soft tissue infections (n = 2), streptococcal pharyngitis (n = 1), urinary tract infection (n = 1), and infectious colitis (n = 1) occurred while receiving ruxolitinib. Three infections (pneumonia, cellulitis, and disseminated zoster) required hospitalization, and 1 death owing to severe pneumonia occurred while receiving ruxolitinib.

One patient experienced a venous thrombotic event (upper extremity deep venous thrombosis) while receiving ruxolitinib. Two patients experienced arterial thrombotic events (1 transient ischemic attack and 1 cerebral vascular accident) while receiving ruxolitinib. Only 1 patient discontinued ruxolitinib owing to the emergence of significant cytopenias. Mean white blood cell counts (10.67 × 10³/uL vs. 8.69 × 10³/uL, P < .001) were significantly higher at baseline, and there was no statistically significant difference in mean platelet counts between baseline and 32-week follow-up (328.19 × 10³/uL vs. 312.53 × 10³/uL; P = .52).

Eleven (8.7%) patients experienced disease progression while receiving ruxolitinib (see Supplemental Table 1 in the online version). Ten (7.9%) patients had documented progression to MF, and 1 (0.79%) patient had documented evolution to acute myeloid leukemia (AML). Progression from PV to MF in 10 patients occurred at a median of 7.2 months (range, 3.16–27.5 months) after starting ruxolitinib. The 1 progression from PV to AML occurred 19 days after the initiation of ruxolitinib. Four secondary malignancies developed while receiving ruxolitinib; squamous cell cancer of the anus (n = 1), diffuse large B-cell lymphoma (n = 1), endometrial adenocarcinoma (n = 1), and Kaposi sarcoma (n = 1). Diagnosis of a secondary malignancy occurred at a median of 292 days (range, 122–905 days) after initiation of ruxolitinib.

Three (2%) patients died while receiving ruxolitinib, from pneumonia (n = 1) at 1221 days, acute respiratory failure of undocumented etiology (n = 1) at 225 days, and from unspecified causes (n = 1) at 420 days after ruxolitinib initiation.

Discussion

Results from this retrospective investigation of real-world outcomes of patients with PV treated with ruxolitinib were largely congruous with published registration trial results.^21,23 Prospective clinical trial data from ruxolitinib studies for the treatment of PV have relatively stringent inclusion and exclusion criteria.^{21–23,25–30} The majority of published data that details the outcomes of patients with PV treated with ruxolitinib has been derived from subgroup analyses of the original patient cohort in the RESPONSE trial.^27–29 A publication detailing the outcomes of 2 patients with PV treated with ruxolitinib is the only description of real-world outcomes in the current literature.³⁰

The primary endpoint for the RESPONSE trial utilized a composite outcome of phlebotomy freedom and spleen volume reduction. Using similar intervals for pre-ruxolitinib and post-ruxolitinib phlebotomy requirements (24 and 32 weeks, respectively), we found a significantly lower rate of therapeutic phlebotomy after the initiation of ruxolitinib. This finding also extended to patients with relatively poor hematocrit control prior to the initiation of ruxolitinib, as the proportion of patients who required more than 3 phlebotomies in the 24 weeks prior to starting ruxolitinib significantly decreased at the 32-week follow-up. Additionally, significant improvement in splenomegaly was observed in our study, as over 50% of patients with palpable splenomegaly at time of ruxolitinib initiation attained a non-palpable spleen on examination.

Ruxolitinib was generally well-tolerated. Approximately 10% of patients discontinued ruxolitinib owing to TEAEs; similar to but slightly higher than reported in the RESPONSE trial (3.6%). The incidence of infectious complications while receiving ruxolitinib was much lower than that in the RESPONSE trial, likely a result of under-documentation and the retrospective nature of the study. A higher rate of disease progression was seen in our study compared with the RESPONSE trial (8.7% vs. 3.6%), although this may be in part be explained by the longer median follow-up (99 vs. 81 weeks, respectively). The 1 patient who progressed to AML while receiving ruxolitinib did so only 19 days after starting the medication, and likely had more advanced disease than documented at time of ruxolitinib initiation. Rates of thromboembolic events, both venous and arterial, were similarly low in both studies.

There are multiple strengths of this retrospective study, one of which was the relatively large number of patients with PV who were included. Our study had a relatively long follow-up period and included patients from 11 institutions located throughout different geographic regions of the US.

The conclusions from this study are limited, given the retrospective process of data collection, although this was unavoidable with the intention of documenting “real-world” outcomes. This aspect of the study likely biased certain outcomes such as symptom burden and splenomegaly assessments more than other objective data points, such as phlebotomy requirements and disease evolution. Additionally, 15% of patients in this study had never been exposed to HU, a requirement for entry in the RESPONSE trials. The varying doses of ruxolitinib noted at time of drug initiation and throughout the follow-up period likely reflected real-world differences in physician perception of therapeutic goals, patient preference, and clinical factors that may influence treatment decision such as baseline and treatment-emergent thrombocytopenia. Also, the frequency of progressive disease at a relatively short median follow-up of 7.2 months noted here may likely reflect a subset of patients with advanced PV at risk of impending progression to MF and in need of JAK inhibition.

The approval of ruxolitinib for HU-resistant or intolerant PV was a significant advance for patients with this disease, as symptom burden can be particularly debilitating and lead to substantial decreases in quality of life. The conclusions drawn from our retrospective review of patients treated with ruxolitinib for this indication were largely congruous with the RESPONSE trial, which led to the aforementioned approval of ruxolitinib.

Clinical Practice Points.

• Ruxolitinib has been demonstrated to have clinical efficacy in patients with PV with regards to hematocrit control, splenomegaly, and symptom management. The data supporting this and the FDA approval of ruxolitinib in this disease is primarily based upon randomized, controlled trials such as RESPONSE, RESPONSE-2, and RELIEF, which consisted of highly-selected patient populations. To date, reports of ruxolitinib efficacy in patients with PV outside of clinical trials have largely consisted of case reports, limited case series, and meta-analyses of the aforementioned clinical trials.

• This study represents the first large review of patients with PV who were treated with ruxolitinib outside of a clinical trial. Our findings are consistent with earlier studies that demonstrated a reduction in phlebotomy requirements and palpable splenomegaly. Furthermore, ruxolitinib appeared to be well-tolerated in this patient population, with only 12 patients discontinuing the drug owing to adverse events.

• This study serves to demonstrate the efficacy of ruxolitinib for patients with PV outside of a clinical trial and should be of interest to clinicians who treat this patient population.