Comparison of Momelotinib, Pacitinib and Ruxolitinib in patients with Myeloproliferative Neoplasm: An Indirect-Comparison Meta-Analysis of Randomized Controlled Trials

The meta-analysis compared the ecacy and safety of momelotinib, ruxolitinib and pacitinib in patients with myeloproliferative neoplasms(MPN), which includes essential thrombocythemia, polycythemia vera and primary myelobrosis. and overall survival rate of patients was not compared, so further clinical studies are needed to prove the results.


Background
The Philadelphia chromosome-negative chronic myeloproliferative neoplasms (MPN) includes essential thrombocythemia (ET), polycythemia vera (PV) and primary myelo brosis (PMF) 1 . MF,a rare myeloproliferative neoplasm, is a clonal stem-cell disorder that leads to ineffective erythropoiesis and dysplastic megakaryocytic hyperplasia 2 . Dysregulation with constitutive activation of the Janus kinasesignal transducer and activator of transcription (JAK-STAT) pathway is central to the pathobiology of MF 3 .The JAK family of enzymes includes JAK1, JAK2, JAK3 and TYK2. These molecules attach to the cytosolic domains of cytokine receptors and are essential for cytokine and growth factor signaling. JAK2 is the only member capable of mediating signaling through the 3 myeloid receptors 4 . JAK2 V617F mediates the activation of downstream signaling through STATs (STAT5, STAT3 and STAT1), extracellular signal-regulated kinase/mitogen-activated protein kinase, and phosphatidylinositol 3kinase/AKT/mammalian target of rapamycin pathways, resulting in uncontrolled myeloproliferation 5 . The JAK1/2 inhibitor ruxolitinib improves splenomegaly and symptoms in patients with intermediate-2 or high-risk myelo brosis, which was approved by the FDA 6 . However, 58-71% of patients treated with ruxolitinib in clinical trials so far have not achieved the primary endpoint of 35% or more reduction in spleen volume from baseline assessed by MRI or CT. Furthermore, more than 50% of patients discontinue ruxolitinib treatment after 3-5 years 7 . When ruxolitinib failure, there are few therapeutic options.Then several small molecular inhibitors were developed subsequently, such as momelotinib and pacitinib. Momelotinib is an ATP-competitive small molecule that potently inhibits JAK1/JAK2 kinases (IC(50) = 11 and 18 nM, respectively), with signi cantly less activity against other kinases, including JAK3 (IC(50) = 155 nM) 8 , which was demonstrated e cacy in patients with primary and secondary myelo brosis and had potential for e cacious treatment of MPN harboring mutated JAK2 and MPL alleles 8,9 . Pacritinib, a potent JAK2 (IC(50) = 23 and 19 nM for JAK2(WT) and JAK2(V617F), respectively) and FLT3 (IC(50) = 22 nM) inhibitor with selectivity against JAK1 and JAK3 (IC(50) = 1280 and 520 nM, respectively) 10 , has demonstrated the ability to favorably impact MF-associated splenomegaly and symptom burden 11 . The phase3 clinical trails of momelotinib and pacritinib had been nished. So far, there has been no clinical studies of comparing momelotinib/ruxolitinib with pacritinib by direct head-to-head comparison. The meta-analysis indirectly compared the e cacy and safety of momelotinib, ruxolitinib and pacitinib in patients with MPN.

Methods
This indirect-comparison meta-analysis was conducted in accordance with the quality of reporting of meta-analysis (PRISMA) statements, and the protocol was registered with PROSPERO, number CRD42020164271.

Inclusive and exclusive criteria
Inclusive criteria: 1.The experimental group were treated with ruxolitinib, momelotinib or pacritinib; The control group can be blank control, placebo, conventional drugs or one of the three experimental group drugs. 2.There is no ethnic restriction in the study. The patients were diagnosed as myeloproliferative disease and the treatment background of the patients was not limited, which can be the initial treatment or recurrent treatment. 3.The results of the experiments need to include spleen response rate, the improvement of symptom and the risk of adverse events. 4.Studies must be randomized controlled trials (RCTs) and o cially published.
Exclusive criteria: 1.The same study was published repeatedly or the results of different stages of the same clinical trial and only the nal or latest results or the results that can be analyzed and utilized shall be taken. 2.The studies had incomplete results and no supplementary data after contacting the author. 3. Basic research and animal experimental research. 4.Studies had a insu cient follow-up time and more than 20% of the patients lost contacting.
Retrieval methods and studies screening By searching PubMed, Embase, Medline, Web of Science, Cochrane Library and Chinese Biomedical Database, then we collected all relevant literature before Dec.20,2019. Language is limited to English. The retrieval terms and methods: 1#:"Primary Myelo brosis" OR "Essential Thrombocythemia" OR "Polycythemia Vera" OR "Myelo brosis" OR "Thrombo-cythemia", 2#:"Ruxolitinib" OR "Momelotinib" OR "Pacritinib" OR"Jaka " OR "SB1518" OR "CYT387", 3#: 1# AND 2#. Two authors independently reviewed the titles and abstracts to screen potentially eligible studies, then two authors reviewed the full text to screen quali ed articles independently. Disagreements between authors were solved by consensus or consultation with a third investigator, if necessary.
Bias risk assessment of the included studies According to "Cochrane collaboration's tool for assessing risk of bias", the assessment contents include the following aspects: 1.Blind methods; 2.Randomized methods; 3.Allocation concealment; 4.Incomplete outcome data; 5.Selective reporting; 6.Other biases. Two authors assessed the quality of the studies independently, disagreements between authors were solved by consensus or consultation with a third investigator.

Data extraction
The contents of data extraction were as follows: the rst author, year of publication, number of patients, the experimental and control group, follow-up time, ≥ 35% reduction in spleen volume from baseline as determined by magnetic resonance imaging or computed tomography 12 , achieving a ≥ 50% reduction in total symptom score (TSS) from baseline assessed using the modi ed Myelo brosis Symptom Assessment Form, the risk of hematological and non-hematological adverse events. Two authors extracted the data independently and inputted them into the database.

Statistical analysis
First step: This meta-analysis used the StataMP14 software to process data from ruxolitinib, momelotinib or pacritinib relevant studies separately. The p-value obtained from chi-squared test and I² value were used to evaluate the heterogeneity of the study. When p-value was bigger than 0.1 and I² was less than 50%, the heterogeneity of the study was considered to be small and the results were analyzed by the xed-effect-model. When p-value was less than 0.1 and I-squared was bigger than 50%, the heterogeneity of the study was considered to be large. When the heterogeneity of the study was large, the meta-analysis would explore the source of heterogeneity or conduct subgroup analysis and all the results were analyzed by the random-effect-model. Relative risk (RR) and 95% con dence interval (CI) were used to analyze binary variable data.
Second step: This meta-analysis used the Indirect Treatment Comparisons(ITC) software to compare RR from the rst step.

Results
The screening results of studies According to the retrieval terms and methods, we obtained 795 studies in the database of PubMed, Embase, Medline, Web of Science, Cochrane Library and Chinese Bio-medical Database. The aboveacquired studies were imported into the NoteExpress software for duplicate checking, then the duplicate studies were removed and 345 studies were obtained. After preliminary screening by reading titles and abstracts for potentially eligible studies, there were remaining 47studies. Then we removed 15 nonrandomized controlled trials and read full text for screening eligible studies again. Eventually, eight phase 3 RCTs with 2125 patients were included in quantitative synthesis. Please see ow-process diagram for details in Fig. 1 and the basic characteristics of the included studies in Table 1. BAT ‡:The most common BAT was ruxolitinib (45%). †≥35% reduction in spleen volume at week 24-32 from baseline as determined by magnetic resonance imaging or computed tomography. ‡achieving a ≥ 50% reduction in TSS from baseline at week 24-32, assessed using the modi ed Myelo brosis Symptom Assessment Form.
Abbreviations:T, trial; C,control; BAT, best available therapy,consisted of any physician-selected treatment, excluding JAK2 inhibitors, and could also include no treatment (watchful waiting) or symptom-directed treatment; MF, myelo brosis; PV, polycythemia vera; TSS,total symptom score.
The results of bias risk assessment of the included studies All studies adopted randomized assignment methods, but only four studies reported the speci c randomized assignment methods and assignment concealment, the remaining four studies did not report. The three studies used double-blind methods and the rest were open-label. In the Verstovsek2012 13 , all patients completed evaluation for 24 weeks and the remaining seven studies had incomplete resulting data. Selective reporting and other biases were all unclear. Please see details in Table 2. All studies were phase 3 large-scale multi-center clinical trails, generally speaking, the risk of bias was relatively small. The Supplementary Appendix was baseline characteristics of the included studies.   Fig. 2. The control group of the ruxolitinib group and the pacritinib group were placebo or best available therapy (BAT) excluding or just including a small part of JKAi small molecule inhibitors, while the control group of the momelotinib group was ruxolitinib,a dual JAK1/2 inhibitor.
The RR greater than 1 favoured the experimental group, otherwise not. As we can see, the ruxolitinib group and the pacritinib group had obviously better effect in spleen response than their corresponding control group, the momelotinib group had similar effect compared with the control group,which was ruxolitinib. Assessments of ≥ 50% reduction in total symptom score(TSS) from baseline assessed using the modi ed Myelo brosis Symptom Assessment Form  Fig. 4. The RR less than 1 indicated that the experimental group had lower incidence of adverse events compared with the control group, otherwise not. The ruxolitinib group had higher incidence of hematologic adverse events (thrombocytopenia) than the control group and had similar incidence of non-hematologic adverse events compared with the control group. In subgroup analysis, heterogeneity may be due to the type of MPN.
The Mesa(2017) and Harrison(2017) compared momelotinib with ruxolitinib, the incidence of all grades of hematologic adverse events(anemia and thrombocytopenia) and the incidence of all grades of non-hematologic adverse events (fatigue, diarrhea and abdominal pain) respectively. The pooled RR(95CI) for the risk of anemia, thrombocytopenia, fatigue, diarrhea and abdominal pain of the momelotinib group vs the control group were 0.60(95%CI0.20-1.85), 0.69(95%CI0.50-0.96), 1.07(95%CI0.71-1.59), 1.32(95%CI0.58-3.01) and 0.95(95%CI0.61-1.48) respectively, as shown in Fig. 5. The higher the RR value, the higher the incidence of adverse events in the experimental group, otherwise not. As we can see, momelotinib had lower incidence of hematologic adverse events than ruxolitinib and had similar incidence of non-hematologic adverse events compared with ruxolitinib.
The  and Mascarenhas(2018) compared pacritinib with the control group, the incidence of all grades of hematologic adverse events(anemia and thrombocytopenia) and the incidence of all grades of non-hematologic adverse events (fatigue, diarrhea and abdominal pain) respectively.  Fig. 6. The RR less than 1 indicated that the experimental group had lower incidence of adverse events compared with the control group, otherwise not. As we can see, the pacritinib group had higher incidence of hematologic adverse events and diarrhea than the control group and had similar incidence of fatigue and abdominal pain compared with the control group. .25) respectively. The RR less than 1 indicated that momelotinib or ruxolitinib had lower incidence of adverse events compared with pacritinib, otherwise not. According to indirect-comparison results, momelotinib had lower incidence of anemia, diarrhea and abdominal pain than pacritinib and had similar incidence of thrombocytopenia and fatigue compared with pacritinib. Ruxolitinib had similar or lower incidence of adverse events than pacritinib excluding thrombocytopenia.

Assessments of the improvement of transfusion-independent
Assessments of the proportion of patients who were transfusion-dependent at baseline changed their classi cation to transfusion-independent on study, using the International Working Group for Myelo brosis Research and Treatment response criteria. In the ruxolitinib vs placebo/BAT studies, only the Verstovsek (2012) reported the speci c results of that, the RR for the improvement of transfusionindependent was 0.878 (95%CI 0.509-1.516). In the pacritinib vs placebo/BAT studies, we adopted xedeffect model to calculate RR and the pooled RR was 5.587(95%CI0.750-41.617), as shown in Fig. 7. We indirectly calculated RR of pacritinib vs ruxolitinib by ITC software, the RR was 6.36(95%CI0.79-50.98).
The RR greater than one favoured pacritinib. According to indirect-comparison results, pacritinib had much better effect in the improvement of transfusion-independent than ruxolitinib.
The indirect-comparison results of e cacy and safety between momelotinib, ruxolitinib and pacritinib three groups by Indirect Treatment Comparisons software was summarized in Table 3. Spleen response § referred to achieving ≥ 35% reduction in spleen volume from baseline at week 24/32 as determined by magnetic resonance imaging or computed tomography in patients with Myeloproliferative Neoplasm.
TSS ¶ referred to achieving a ≥ 50% reduction in TSS (total symptom score) from baseline at week 24-32 assessed using the modi ed Myelo brosis Symptom Assessment Form in patients with MPN. †The symbol referred to all grades of adverse events.

Discussion
The meta-analysis indirectly compared the e cacy and safety of momelotinib, ruxolitinib and pacitinib in patients with myeloproliferative neoplasms. As for spleen response, momelotinib and ruxolitinib had similar effect and both were better than pacritinib, but pacritinib also had obviously better effect than other drugs excluding JAKi inhibitors. In the pacritinib group vs the control group studies, the most common BAT of the Mascarenhas(2018) was ruxolitinib (45%), we regarded the control group of pacritinib as drugs excluding JAKi inhibitors in this meta-analysis, so the results may underestimate pacritinib's e cacy. As for the improvement of symptom in patients, the e cacy of drugs in the treatment of patients with MPN was momelotinib > ruxolitinib > pacritinib > BAT(excluded JAKi inhibitors).
Assessments of the proportion of patients who were transfusion-dependent at baseline changed their classi cation to transfusion independent on study, using the International Working Group for Myelo brosis Research and Treatment response criteria, pacritinib had better effect than ruxolitinib obviously. As for safety, the risk of all grades of anemia was momelotinib < ruxolitinib ≦ pacritinib; The risk of all grades of thrombocytopenia was momelotinib ≈ pacritinib < ruxolitinib; The risk of all grades of fatigue was momelotinib ≈ ruxolitinib ≦ pacritinib; The risk of all grades of diarrhea was ruxolitinib < momelotinib < pacritinib; The risk of all grades of abdominal pain was momelotinib ≈ ruxolitinib ≦ pacritinib.

Conclusions
In conclusion, pacritinib compared with momelotinib or ruxolitinib, had relatively poorer e cacy and higher risk of adverse events excluding the risk of thrombocytopenia and the improvement of transfusionindependent in patients with MPN. Momelotinib had similar or better e cacy and similar or lower risk of adverse events than ruxolitinib, excluding the risk of diarrhea. Generally, among the three JAKi inhibitors that were included in the study, momelotinib has the best potential therapeutic value in patients with MPN, followed by ruxolitinib and the last one was pacritinib, but pacritinib had a unique advantage in the improvement of transfusion-independent. The meta-analysis was the rst study to compare the e cacy and safety of momelotinib, ruxolitinib and pacitinib in patients with MPN by indirect-comparison, so it is likely to provide important information for drug selection in the treatment of MPN in the future. But the meta-analysis also has several limitations, the number of studies that were included were relatively not enough and overall survival rate of patients was not compared, so further clinical studies are needed to prove the results.  Figure 1 Screening process of randomized controlled trials included in the meta-analysis Figure 2 Forest plots of the pooled Relative Risk (95Con dence Interval) for achieving ≥35% reduction in spleen volume from baseline at week 24/32 as determined by magnetic resonance imaging or computed tomography in patients with Myeloproliferative Neoplasm(MPN) of the ruxolitinib group, the momelotinib group and the pacritinib group versus their corresponding control group respectively. The size of the blocks or diamonds represents the weight for the xed-effect-model in the meta-analysis Figure 3 Forest plots of the pooled Relative Risk (95Con dence Interval) for achieving a ≥ 50% reduction in TSS (total symptom score) from baseline at week 24-32 assessed using the modi ed Myelo brosis Symptom Assessment Form in patients with MPN of the ruxolitinib group, the pacritinib group and the momelotinib group versus their corresponding control group respectively. The size of the blocks or diamonds represents the weight for the xed-effect or random-effect model in the meta-analysis Figure 4 Forest plots of the pooled Relative Risk (95Con dence Interval) for all grades of hematologic adverse events(anemia and thrombocytopenia) and non-hematologic adverse events (fatigue, diarrhea and abdominal pain) in patients with MPN of the ruxolitinib group versus the control group respectively. The size of the blocks or diamonds represents the weight for the xed-effect or random-effect model in the meta-analysis. According to sub-types of MPN(polycythemia vera, myelo brosis), the meta-analysis conducted subgroup analysis Page 20/22 Figure 5 Forest plots of the pooled Relative Risk (95Con dence Interval) for all grades of hematologic adverse events(anemia and thrombocytopenia) and non-hematologic adverse events (fatigue, diarrhea and abdominal pain) in patients with MPN of the momelotinib group versus the control group respectively. The size of the blocks or diamonds represents the weight for the xed-effect or random-effect model in the meta-analysis Figure 6 Forest plots of the pooled Relative Risk (95Con dence Interval) for all grades of hematologic adverse events(anemia and thrombocytopenia) and non-hematologic adverse events (fatigue, diarrhea and abdominal pain) in patients with MPN of the pacritinib group versus the control group respectively. The size of the blocks or diamonds represents the weight for the xed-effect model in the meta-analysis