Myeloproliferative neoplasms treated with hydroxyurea, pegylated interferon alpha-2A or ruxolitinib: clinicohematologic responses, quality-of-life changes and safety in the real-world setting

Introduction: Real-world data of responses, quality-of-life (QOL) changes and adverse events in patients with myeloproliferative neoplasms (MPN) on conventional therapy (hydroxyurea ± anagrelide), pegylated interferon alpha-2A (PEG-IFN α -2A) or ruxolitinib are limited. Methods: We prospectively studied MPN patients receiving conventional therapy, PEG-IFN α -2A or ruxolitinib. Next-generation sequencing of 69 myeloid-related genes was performed. Clinicohematologic responses, adverse events, and QOL (determined by the Myeloproliferative Neoplasm Symptom Assessment Form Total Symptom Score, MPN-SAF TSS) were evaluated. Results: Seventy men and ﬁ fty- ﬁ ve women with polycythemia vera (PV) ( N = 23), essential thrombocythemia (ET) ( N = 56) and myelo ﬁ brosis (MF) ( N = 46) were studied for a median of 36 (range: 19 – 42) months. In PV, responses were comparable for di ﬀ erent modalities. CREBBP mutations were associated with inferior responses. In ET, PEG-IFN α -2A resulted in superior clinicohematologic complete responses (CHCR) ( P = 0.045). In MF, superior overall response rates (ORR) were associated with ruxolintib ( P = 0.018) and JAK2 V617F mutation ( P = 0.04). For the whole cohort, ruxolitinib led to rapid and sustained reduction in spleen size within the ﬁ rst 6 months, and signi ﬁ cant improvement of QOL as re ﬂ ected by reduction in MPN-SAF TSS ( P < 0.001). Adverse events of grades 1 – 2 were observed in 44%, 62% and 20% of patients receiving conventional therapy, PEG-IFN α -2A and ruxolitinib respectively; and of grade 3 – 4 in 7% and 9% of patients receiving PEG-IFN α -2A and ruxolitinib. Conclusions: Conventional therapy, PEG-IFN α -2A and ruxolitinib induced responses in all MPN subtypes. PEG-IFN α -2A led to superior CHCR in ET; whereas ruxolitinib resulted in superior ORR in MF, and signi ﬁ cant reduction in spleen size and improvement in QOL.

Although PEG-IFNα and ruxolitinib appear promising in clinical trials of MPN, they have not been prospectively compared with conventional therapy in a non-trial realworld setting. In this study, we prospectively evaluated the efficacy and safety of PEG-IFNα-2A, ruxolitinib and hydroxyurea in a cohort of MPN patients. who received hydroxyurea, PEG-IFNα-2A or ruxolitinib were recruited. All cases were diagnosed according to the World Health Organization (WHO) 2016 criteria. For cases presenting before 2016, all materials were reviewed to ensure that the diagnoses complied with the WHO 2016 criteria. Baseline clinicopathologic and molecular characteristics were determined. Prospective data on treatment responses, quality-of-life (QOL) and adverse events were obtained every 2-4 weeks for the first 6 months and every 3 months thereafter. This study was approved by the institutional review board and registered at the HKU Clinical Trial Registry (Identifier: HKUCTR-2030). Patients gave written informed consent.
Molecular studies and next-generation sequencing (NGS). Patients were annotated for driver mutations of JAK2, CALR and MPL as previously described [24][25][26][27][28][29]. Targeted NGS was performed on archived DNA from diagnostic bone marrow samples. A custom xGen Lockdown Panel targeting 69 myeloid-relevant genes (supplemental file 1) was designed based on GRCh37/hg19 (Integrated DNA Technologies, Coralville, Iowa, USA). All exons of the 69 genes were sequenced, with a total of 2885 probes covering 273.03 kb. The enriched libraries were sequenced pair-ended with the Illumina MiSeq System (Illumina, San Diego, California, USA). FASTQ files containing at least 1 million raw reads with coverage of 500X were generated for bioinformatic analyses as previously described [30].
Treatment. The choice of conventional therapy, PEG-IFNα-2A or ruxolitinib was based on prevailing guidelines [1,3,31], physician choice and patient preferences. We also took into account concomitant medical comorbidities that would increase cardiovascular risks, including smoking, hypertension, hyperlipidemia, type 2 diabetes mellitus, a strong family history of cardiovascular diseases and presence of vascular symptoms for initiation and choice of treatment. Conventional therapy included hydroxyurea for cytoreduction and anagrelide as an adjunct for platelet control. PEG-IFNα-2A was recommended as first-line treatment for MPN patients aged ≤50 years, or as second-line therapy for patients resistant or intolerant to hydroxyurea. It was started at 135 µg subcutaneously, initially every 2 weeks and escalated to weekly. Ruxolitinib was recommended for patients with constitutional symptoms, symptomatic splenomegaly, and intolerance or resistance to hydroxyurea [1]. It was started at 10 mg twice daily and escalated by 10 mg/day every 4 weeks to a maximum of 25 mg twice daily. PEG-IFNα-2A or ruxolitinib was withheld in the event of ≥ grade 3 hematologic or non-hematologic toxicities, and resumed on resolution of toxicities. All patients received anti-platelet therapy with low-dose aspirin (80 mg/day) or clopidogrel (75 mg/day) if sensitive to aspirin. The target hematocrit was <45% for PV.
The target platelet count was 180-450 × 10 9 /L for PV and ET [3,[31][32][33]. In MF, the threshold for blood transfusion in asymptomatic patients without cardiac comorbidities was 7 g/dL. During ruxolitinib therapy, patients positive for hepatitis B virus (HBV) surface antigen (HBsAg) received entecavir 0.5 mg/day as prophylaxis; whereas patients negative for HBsAg but positive for anti-hepatitis B core antigen-antibody (anti-HBc) were regularly monitored for circulating HBV DNA, and started on entecavir once HBV DNA became detectable [34]. All patients gave informed consent to treatment. Patients treated with hydroxyurea or anagrelide prior to this study were not excluded. Off-label use of ruxolitinib was allowed with written informed consent for ET patients with significant symptoms, who refused other treatment options.
Definitions. Risk stratification was conducted as follows: International Prognostic Scoring System (IPSS) [32] and European LeukemiaNet (ELN) recommendations [1] for PV; International Prognostic Score for ET (IPSET) [35] and the IPSET-thrombosis scores [36] for ET; and Dynamic International Prognostic Scoring System (DIPSS) [37] and DIPSS-plus [38] for MF. Treatment responses (clinicohematologic complete response, CHCR; partial response, PR; stable disease, SD; clinical improvement, CI; progressive disease, PD; no response, NR) were defined according to the criteria proposed by the ELN and International Working Group-Myeloproliferative Neoplasms Research and Treatment (IWG-MRT) [39][40][41] (supplemental file 2). Spleen size was defined as the distance from the costal margin to the spleen tip, verified by two independent clinicians. Quality of life (QOL) was evaluated by a Chinese version of the Myeloproliferative Neoplasm Symptom Assessment Form Total Symptom Score (MPN-SAF TSS), which consisted of 10 items for symptom burden on a 0-10 scale [42]. Adverse events (AE) were determined and graded using the NCI Common Terminology Criteria for Adverse Events (CTCAE) version 4.0 [43].
Sample size calculation. To give power of at least 80% (2-sided alpha level of 0.05) to detect ≥20% difference in the outcomes between various treatment groups, it was estimated that a sample size of 108 patients (36 patients per treatment group) would be required.
Statistical analyses. All data were censored on 30 June 2019. Categorical variables were analyzed with the χ 2 test. Continuous variables were analyzed with non-parametric tests. Clinico-hematologic and QOL responses of different treatment modalities (conventional versus ruxolitinib versus PEG-IFNα-2A) were assessed at 3-monthly time points, and compared with one-way analysis of covariance (ANCOVA), incorporating baseline values as a covariate to ensure that differences at different time points were unaffected by baseline inter-group variations. Graphs and charts were constructed using Graphpad Prism version 7.02 and R software (R Project for Statistical Computing, Vienna, Austria). Concentration graph analysis was used to determine the gene relevance network, generating a covariance matrix for Circos plot (Circos software). Statistical analyses were performed using SPSS version 25.0 (Chicago, IL, USA). P-values (2-tailed) of <0.05 were considered significant.

Results
Patients. Seventy-five men and fifty-five women (PV, N = 23; ET, N = 56; MF, N = 46) at a median age of 48.4 (range: 22.7-88.6) years were recruited ( Table 1). None of our pre-MF patients required treatment during the study period and so they were not included. The median duration of follow-up for the cohort was 36. Treatment and outcome in PV. Amongst treatment groups (conventional, PEG-IFNα-2A, ruxolitinib), the gender, age, hemoglobin, hematocrit, platelet count, lactate dehydrogenase (LDH), and splenomegaly were comparable. However, the ruxolitinib group showed a higher leucocyte count (P = 0.009), higher LDH (P = 0.02) and more patient belonging to high-risk categories (IPSS, P = 0.009; ELN, P = 0.005) (supplemental file 6). All patients were assessed for treatment responses (Table 2, Figure 2). At a median treatment duration at 6 (range: 3-18) months, the overall response rate (ORR) was 82% (CHCR: 52%; PR: 30%), which was comparable for various treatment groups. For genes in the NGS panel, only CREBBP mutations were associated with an inferior ORR (P = 0.04) (supplemental file 7). The hemoglobin fell in all groups, with the ruxolitinib group showing the lowest median hemoglobin, which at 15 months was significantly lower than the two other groups (P = 0.03) ( Figure 2). The leucocyte and platelet counts also fell, except in the ruxolitinib group where the platelet count increased progressively during follow-up ( Figure 2). There were no cardiovascular or thrombotic complications. One patient who received hydroxyurea for ten years prior to this study progressed to post-PV MF. Another patient with del(5)(q14q33) progressed to secondary acute myeloid leukemia (AML) with complex karyotypes 8 months after ruxolitinib treatment and 7 years after initial diagnosis.
Responses in splenomegaly. To increase sample size, the whole MPN cohort was evaluated for spleen response. Pre-treatment spleen size was significantly larger in the ruxolitinib-treated group (P < 0.001). Despite this difference, patients treated with ruxolitinib had rapid and sustained spleen responses during the study ( Figure 5(A)).
Responses in QOL. The mean MPN-SAF TSS for patients with PV, ET and MF were 19.8, 24.6 and 23.9 respectively (P = 0.51). The median total symptom scores were comparable for PV (16.5; range: 0-50), ET (25.5; range: 0-72) and MF (19; range: 0-74) ( Figure 5(B)). Amongst symptoms, fatigue was the most serious in all MPN subtypes (supplemental file 13). ET patients had significantly higher scores for bone pain (P = 0.047), whereas MF patients had significantly more unintentional weight   loss (P = 0.044). Ruxolitinib resulted in significantly superior improvement in QOL as compared with conventional therapy and PEG-IFNα-2A (P < 0.001) ( Figure  5(C)). Marked improvement in QOL in ruxolitinibtreated patients was seen within 3 months of therapy, and sustained throughout the study. The superiority of ruxolitinib over conventional therapy and PEG-IFNα-2A was observed across all symptom burden domains (supplemental file 14).

Discussion
In this prospective cohort study, we showed that hydroxyurea, PEG-IFNα-2A and ruxolitinib were efficacious in MPN patients. For PV patients, ORRs were comparable for all three modalities. However, ruxolitinib did not effectively control platelet counts. For ET, practically all patients responded to treatment. PEG-IFNα-2A resulted in superior CHCR rates; whereas ruxolitinib led to significantly worse hemoglobin and inferior platelet control. For MF, CR could not be achieved in any patient. Ruxolitinib achieved the best CI. For the entire MPN cohort, reduction in splenomegaly was only observed in ruxolitinib-treated patients, which was durable throughout the study. Except CREBBP mutations in PV and JAK2V617F in MF, mutations in genes in the NGS panel did not impact on outcome. These results show that other than ORR, other factors also affect the choice of treatment. In PV, hydroxyurea is recommended for high-risk patients [1]. Although a leukemogenic potential of hydroxyurea could not be established in uncontrolled studies, leukemic transformation actually increased with duration of treatment. Hence, hydroxyurea should be used cautiously in young patients. Ruxolitinib is approved for PV patients intolerant to hydroxyurea. However, for patients with concomitant thrombocythemia, ruxolitinib might not achieve optimal control. PEG-IFNα had been reported to achieve ORR of 70-77%, similar to our results. Interestingly, complete molecular response (undetectable JAK2V617F) could be achieved, with a median time to response at 24 months [4,6,11]. Therefore, for young PV patients requiring cytoreductive treatment, PEG-IFNα may be a preferred choice.
In ET, hydroxyurea effectively controls thrombocythemia and had been shown to prevent vascular and thrombotic complications. Its prolonged use in young patients is still restricted by concerns of leukemogenesis. In ET intolerant or resistant to hydroxyurea, ruxolitinib did not achieve superior clinicohematologic responses compared with other forms of second-line therapy [44,45]. It is noteworthy that ET treated with ruxolitinib had more significant and rapid reduction in symptom burden [44]. PEG-IFNα achieved CHCR in about three quarters of patients, similar to our results. Furthermore, molecular responses can be observed in approximately 41% and complete molecular response can be achieved in 5-10% of patients [4,46,47]. Hence, PEG-IFNα remains a preferred choice for young ET patients.
In intermediate-2 and high risk MF, the COMFORT-I and COMFORT-II studies have shown that ruxolitinib use is associated with significant clinical benefits in controlling splenomegaly, ameliorating diseaserelated symptoms, improving QOL and prolonging survivals, compared with placebo or best-availabletherapy. In the COMFORT-I study, ≥35% reduction in spleen volume and ≥50% improvement in MPN-SAF was seen in 42% and 46% of patients respectively at 24 weeks [18]. In the COMFORT-II study, ≥35% reduction in spleen volume was seen in 28% of patients at 48 weeks [19]. The 5-year follow-up data from the COMFORT-I study also demonstrated prolonged median OS compared with placebo [48][49][50]. In the ROBUST trial, ruxolitinib resulted in ≥50% reduction in palpable spleen length and ≥50% reduction in MPN-SAF at 48 weeks respectively [51]. In the JUMP study, 61% of patients with intermediate-1 risk MF achieved ≥50% reduction in palpable spleen length [52]. Similarly, our study showed that ruxolitinib achieved clinical improvement in most patients, associated with rapid and sustained control of symptom burden and splenomegaly. Long-term use of ruxolitinib also resulted in ≥50% reduction in JAK2V617F allele burden [53]. Mutations in ASXL1, EZH2, or IDH1/2, or ≥3 mutations on a multigene panel, were associated with poor treatment responses and outcome following ruxolitinib [54,55]. In MF, PEG-IFNα has only been evaluated in small case series or retrospective studies [56,57]. Similar to our results, ORR (CR + PR + CI) was achieved in 50% of patients, and ≥50% reduction in spleen size in 40% of cases [56]. Responses were encouraging in early MF treated with PEG-IFNα, with control of leukocytosis and thrombocytosis seen in approximately 80% of patients, and spleen size reduction in 47% of cases [57]. Complete responses were however uncommon (<10%), and molecular responses were rarely reported. In overt MF with significant symptom burden and splenomegaly, ruxolitinib achieved the greatest benefit. The role PEG-IFNα in MF remains to be defined, with benefits more probable in early MF [58].
In this study, distinct patterns of adverse events were seen in different treatment modalities. In patients treated with hydroxyurea, the main AEs were neutropenia and mucositis (17% each). This was similar to previous studies. Mucocutaneous ulceration, in particular, was observed in around 13-16% of patients who cannot tolerate hydroxyurea [59][60][61]. Cytopenia, hepatitis and immune-mediated AEs predominate in patients treated with PEG-IFNα. In our cohort, major AEs associated with PEG-IFNα were neutropenia, fatigue and transaminitis, mostly grade 1-2, which were similar to previously reported studies [4,11,46]. One patient developed myasthenia gravis. Thyroiditis or depression was not observed in our cohort. This is in contrast to recent studies in PV and ET reporting psychological complications and thyroid dysfunction in 10-40% and 10-25% respectively [11,46]. In our cohort, anemia and infective complications were the key AEs observed. Anemia was less severe in our cohort compared with published studies, with no patients requiring treatment discontinuation due to anemia. Grade 3-4 anemia was reported in 45% and 42% of patients with MF receiving ruxolitinib in the COMFORT-I and COMFORT-II studies [18,19]. We adopted a starting dose of 10 mg twice daily and a gradual dose-escalation by 10 mg every 4 weeks and a transfusion threshold of 7 g/dL. A dose-escalation approach may reduce the incidence of anemia, which may lead to drug discontinuation or dose reduction [62]. Thalidomide and erythropoietin stimulating agents were not used for the treatment of anemia during the study. In ruxolitinib-treated patients, tuberculosis occurred in 6% of patients in our study, in contrast to 1% reported in the COMFORT-II study [19]. This is likely due to the higher prevalence of tuberculosis in the Asian population. In addition, patients with occult HBV infection had estimated HBV DNA reactivation rates of 8% and 31% at 6 and 12 months [34]. This observation is unique to our population, due to a high seroprevalence of anti-HBc in East Asians [34]. Preemptive use of entecavir effectively prevented clinical hepatitis.
In a non-trial setting, our results showed that conventional therapy, PEG-INFα and ruxolitinib all induced responses in MPN. However, significantly better responses were only associated with PEG-IFNα and ruxolitinib in specific settings. Safety profiles were different for these modalities. Prospective comparative studies between these two modalities are needed in order to critically appraise their relative merits in different MPNs.

Disclosure statement
No potential conflict of interest was reported by the author(s).